Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
History for cwal.c
2022-01-20
| ||
16:57 | Renamed to src/cwal.c. Major build reorg, step 1 of N. Moving to a top-level non-recursive make, based on the recent libfossil build tree re-org. Currently onlythe main libcwal builds. check-in: [a6705cc58a] user: stephan branch: build-reorg, size: 0 | |
2021-08-01
| ||
10:23 | Removed CWAL_FEATURE_SCOPE_STORAGE_HASH and changed the default of CWAL_OBASE_ISA_HASH to 1. file: [0f6a4fbc35] check-in: [deddd35e15] user: stephan branch: trunk, size: 459263 | |
2021-07-24
| ||
10:45 | Implemented cwal_hash_take_props() for the CWAL_OBASE_ISA_HASH case. That was the last known piece which required/requires fiddling for that build option. Time to try it out on my various CGI backends. file: [b7dccc3075] check-in: [a98b4b273d] user: stephan branch: trunk, size: 460792 | |
09:36 | Some cwal_size_t to cwal_midsize_t conversions. Minor doc and constness tweaks. file: [9af2c57667] check-in: [601b514dd3] user: stephan branch: trunk, size: 458869 | |
2021-07-23
| ||
05:46 | Minor doc updates. file: [6175cc5494] check-in: [0a79496327] user: stephan branch: trunk, size: 458983 | |
2021-07-22
| ||
12:41 | Corrected a silly cwal_obase_kvp_iter_next() bug which caused it to skip over colliding tail-end entries in the hashtable (this resolves my website build failures). Corrected some of the scope propery handling for the case when scopes are configured to use hashes (that feature can reasonably be removed now that CWAL_OBASE_ISA_HASH works). Valgrind seems happy but needs another 20-30 minutes to finish crunching the various memory configurations of the s2 unit tests. file: [e82d683fd7] check-in: [bbe082639a] user: stephan branch: trunk, size: 458922 | |
2021-07-20
| ||
02:21 | Found and documented (for the time being) an unexpected failure case when using the new cwal_obase hashtable storage. Renamed cwal_scope_import_properties() to cwal_scope_import_props() for consistency. file: [488c06017d] check-in: [b4f045c66a] user: stephan branch: trunk, size: 457781 | |
2021-07-11
| ||
23:29 | A full valgrind of the s2 test scripts is happy with CWAL_OBASE_ISA_HASH mode, but cwal_hash_take_properties() is disabled in that mode - it needs to be reimplemented for that case and is a bit of a beast. Aside from the unavoidable behavioural change of using type-strict property keys in hash mode, all seems to be working exactly as before. Memory cost comparisons are pending. file: [9629ee39c3] check-in: [5e954ac3be] user: stephan branch: obase-isa-hash2, size: 457786 | |
22:06 | CWAL_OBASE_ISA_HASH is mostly implemented and seems to basically work, but cwal_hash_take_props() still has to be reimplemented. file: [557bea1c01] check-in: [273b214789] user: stephan branch: obase-isa-hash2, size: 457765 | |
18:27 | More baby steps towards hash-stored object properties. file: [d55df2f5b2] check-in: [e0d28b94d6] user: stephan branch: trunk, size: 456020 | |
17:45 | More cwal_size_t ==> cwal_midsize_t changes. file: [01e620b62d] check-in: [92f0fb90be] user: stephan branch: trunk, size: 454830 | |
17:36 | Inconsequential minor cleanups. file: [2676cbf15f] check-in: [3310bf0b1c] user: stephan branch: trunk, size: 454595 | |
17:20 | More char-to-bool API conversions. Started changing the string APIs to expose cwal_midsize_t string lengths, though a couple such changes (namely cwal_string_cstr2()) are pending because they'll break a bit of s2's core. file: [65fd972fef] check-in: [cdefef67f8] user: stephan branch: trunk, size: 454683 | |
10:00 | Renamed cwal_lsize_t cwal_midsize_t. Should probably just use uint32_t, but like the "documentation factor" of using this typename. file: [d9ba166784] check-in: [9ec0b252f6] user: stephan branch: trunk, size: 454149 | |
08:42 | core infrastructure: re-added the obase kvp iterator, a flag on cwal_list to flag whether it's being visited, and strengthened some internal assertions related to list/child visitation. file: [ec5767e8eb] check-in: [45f6d620a6] user: stephan branch: trunk, size: 454011 | |
2021-07-10
| ||
22:22 | Refactored the core cwal_hash operations into the internal-only cwal_htable type, for eventual re-use in implementing hash-hosted object-level properties (as opposed to the current sorted list impl). A full valgrind of the s2 test scripts is happy. file: [042c4a5c6a] check-in: [5e0ce36380] user: stephan branch: trunk, size: 451374 | |
20:14 | Got CWAL_ENABLE_TRACE compiling again and added --trace-cwal flag to s2sh/s2sh2. file: [3c7f3a3079] check-in: [b73445b7ca] user: stephan branch: trunk, size: 447746 | |
20:06 | Pulled in part of [16c11f2139]. Preparing to take a fresh whack at the [obase-isa-hash] branch, as that branch is apparently beyond repair. file: [05c0244c18] check-in: [8c86fbdedf] user: stephan branch: trunk, size: 447742 | |
19:56 | Now it doesn't work in either mode. Stashing this for later analysis and rolling back to some previous version (possibly the start). file: [7d4ba1b50f] check-in: [9522bcf73e] user: stephan branch: obase-isa-hash-wha, size: 457014 | |
15:41 | Minor cleanups. Fixed an ancient bug in cwal_hash_insert_with_flags() when replacing a property using a string literal key (it was storing that key as the new value). Still having the same refcount/lifetime issues with hash-hosted object properties :/. file: [a38777a688] check-in: [16c11f2139] user: stephan branch: obase-isa-hash, size: 455781 | |
14:28 | Some code de-duplication. Still getting values without a scope at cleanup-time when using hash property storage. file: [2f9950168d] check-in: [250dad0420] user: stephan branch: obase-isa-hash, size: 455542 | |
12:58 | test.c tests of hash property storage basically works but any non-empty script in s2sh triggers assertions at scope cleanup related to unexpected refcounts and missing scope ownership. file: [cf125db2e8] check-in: [d7a5284c55] user: stephan branch: obase-isa-hash, size: 455937 | |
11:46 | Both property storage modes now compile but the new one is (predictably) currenly broken. file: [b2e9938512] check-in: [3c808939d6] user: stephan branch: obase-isa-hash, size: 453710 | |
11:21 | Resolved a new memory leak, fixed a cwal_list_reserve() result misuse, and changed some cwal_size_t decls to cwal_lsize_t. file: [c929a63c0f] check-in: [4d6229e505] user: stephan branch: obase-isa-hash, size: 452790 | |
10:35 | More work on the new hashtable infrastructure. The old/non-hash obase case works with this new code, but still hammering on the new case. file: [18f5ab6bf5] check-in: [4fdff30fc3] user: stephan branch: obase-isa-hash, size: 452394 | |
04:26 | More work on enabling hash storage for cwal_obase properties. file: [ee940cd1bb] check-in: [a6b871bfd0] user: stephan branch: obase-isa-hash, size: 449858 | |
00:04 | Initial work on adding a compile-time option to make cwal_obase use a hashtable for property storage. That mode does not yet compile. file: [047fc7f2c3] check-in: [044ee16592] user: stephan branch: obase-isa-hash, size: 447142 | |
2021-07-09
| ||
22:06 | Changed a bunch of predicate function signatures from returning char to bool (now that we're C99, we might as well own it). file: [ed18f29ab3] check-in: [9b35720349] user: stephan branch: trunk, size: 447665 | |
21:59 | Added cwal_prop_key_can() to check the data type of object property keys and hashtable keys so that we can reject any types for which their mutable state is accounted for in their comparison or hashing operation. Thus Buffers and Tuples are no longer legal as property keys and will trigger a CWAL_RC_TYPE error. Allowing those types as keys in the first place was a design error. file: [3a7ddd66e8] check-in: [539060fab2] user: stephan branch: trunk, size: 447666 | |
20:46 | Refactored cwal_printf() interface to use return-0-on-success semantics because the classical printf semantics are next to useless for this particular API and this gives us the ability to propagate errors out properly. file: [ce3d556175] check-in: [043c912982] user: stephan branch: trunk, size: 447073 | |
2021-06-26
| ||
08:16 | Added cwal_rc_cstr_fallback() to allow us to get rid of s2_rc_cstr() so that being-factored-out t10n/c9n code can make use of cwal_rc_cstr() to report non-core-cwal result code strings. s2 now registers a fallback for its S2_RC_xxx codes. file: [868eb86538] check-in: [e679463d26] user: stephan branch: trunk, size: 447102 | |
02:40 | A few cleanups to the s2_error/cwal_error port. file: [c574b8611b] check-in: [caa168b835] user: stephan branch: s2-remove-err-obj, size: 446266 | |
2021-06-24
| ||
11:00 | Had emacs reformat the core lib to 2-space indentation. file: [963a0988cb] check-in: [29cf299dec] user: stephan branch: trunk, size: 446093 | |
09:29 | More work on the standalone t10n/c9n subsystem. file: [4652f75073] check-in: [fb2e2b8f05] user: stephan branch: trunk, size: 485252 | |
07:47 | More work on the standalone t10n/c9n subsystem. file: [1a49609e1d] check-in: [24029cb1d9] user: stephan branch: trunk, size: 485062 | |
05:25 | Ported much of the s2_error infrastructure to cwal_error in the core lib for use by the pending t10n/c9n modules (and similar code). file: [418820672b] check-in: [3d41e2d40f] user: stephan branch: trunk, size: 484980 | |
2020-05-22
| ||
06:31 | Random drive-by doc tweaks. file: [e63dd82909] check-in: [cf93734fae] user: stephan branch: trunk, size: 479789 | |
2020-02-20
| ||
01:35 | Insignificant internal tweak. file: [8c958a8694] check-in: [37a77d4f5f] user: stephan branch: trunk, size: 479764 | |
2020-02-18
| ||
21:15 | Clarified in the docs that cwal_props_copy() and friends do not traverse/copy hidden properties. file: [c2b260d748] check-in: [dbc89b0be6] user: stephan branch: trunk, size: 479235 | |
2020-02-07
| ||
13:00 | cwal_parse_argv_flags() now optionally accepts an array as its target, and appends all raw/unparsed flags to that array. s2sh/s2sh2 now create s2.ARGV as an array. file: [cff9756592] check-in: [058c6f6ae8] user: stephan branch: trunk, size: 479324 | |
2020-02-02
| ||
06:41 | Minor error check reordering so that hash insert matches its docs. file: [b7269b7914] check-in: [ce136e8b11] user: stephan branch: trunk, size: 478690 | |
2020-01-18
| ||
08:43 | Changed cwal_obase_search() and cwal_obase_search_v() to stop looking if any entry in their prototype chain is locked (previously only their initial object was checked for a lock). No APIs currently lock these, so this doesn't break anything. file: [50fcb12aef] check-in: [89b61168d2] user: stephan branch: trunk, size: 478690 | |
05:14 | Fixed some test code which was broken by the recent cycle-detection/iteration changes. Patched Object.eachProperty() impl to handle the iteration-related result codes (transforming them to exceptions). Changed cwal_props_visit_kvp() to *not* return CWAL_RC_CYCLES_DETECTED for concurrent visiting (which is now legal) and fixed some ancient incorrect docs in nearby functions. file: [cbd3c6ad7b] check-in: [9dfc28ffaa] user: stephan branch: trunk, size: 478803 | |
2020-01-11
| ||
21:45 | Minor internal cleanups, mostly removal of dead code. file: [b61a7009be] check-in: [0a6549cae8] user: stephan branch: trunk, size: 478727 | |
2019-12-20
| ||
17:56 | Minor internal doc addition. file: [550a38bcdb] check-in: [e2f34d234d] user: stephan branch: trunk, size: 481351 | |
17:43 | s2 array.sort() now internally uses floating-point comparison so that sort routines which do things like (return 0.1-0.3) can DTRT. Prompted by a bug report files against mujs: https://github.com/ccxvii/mujs/issues/122 file: [48c1c332e6] check-in: [4f0d98c329] user: stephan branch: trunk, size: 481195 | |
2019-12-17
| ||
02:38 | Style change: globally removed commas after 'i.e.,'. file: [0bd779b42c] check-in: [6100e95be2] user: stephan branch: trunk, size: 480788 | |
02:36 | Global style change: removed commas from 'e.g.,'. Style guides disagree on whether that comma is 'required'. IMO it just adds noise and its intent is strongly implied without requiring the comma. file: [3c2e30831b] check-in: [040a1c6ac4] user: stephan branch: trunk, size: 480789 | |
2019-12-13
| ||
01:07 | Added a few missing is-visiting checks on property-modifying functions. file: [48234297be] check-in: [81abb1596a] user: stephan branch: trunk, size: 480791 | |
2019-12-12
| ||
12:52 | Reinstated cyclic valid checks in the core JSON API via addition of another visitation-related RCFLAG (so we're now stealing 8 refcount bits for flags). Simplified and cleaned up the is-traversing flagging APIs. file: [4d264ab99c] check-in: [0cf849b260] user: stephan branch: multi-iteration, size: 480602 | |
02:04 | MOAR multi-iteration-related work. We broke the cycle-detection in the JSON bits with this undertaking, and now have to fall back to the configured max output depth to catch them. file: [3d3023cf6b] check-in: [4fc57dd5d5] user: stephan branch: multi-iteration, size: 478338 | |
00:32 | More cleanups, some relevant to this branch, some only remotely related. This unfortunately increases the sizeof(cwal_obase) on 32-bit builds by 4 (2?) bytes, but we desperately needed more flag bits for container flags. Forgot to mention that the previous commit stole another refcount bit (7 flag bits in total). file: [2458b3a572] check-in: [db9f4b10f3] user: stephan branch: multi-iteration, size: 477740 | |
2019-12-11
| ||
23:06 | Initial work on making multiple concurrent iteration possible/legal, and disallowing it at the right times (e.g., we cannot sanely iterate and sort an array concurrently). The s2 unit tests pass, after a couple minor modifications, but there is still lots of testing and documentation patching to be done. file: [788875fcf3] check-in: [e21f0af7f4] user: stephan branch: multi-iteration, size: 478029 | |
2019-12-10
| ||
19:27 | Added an as-yet-unused flags member on cwal_tuple (no sizeof() increase, due to spare padding). Added some tuple comparison tests. file: [d899642e98] check-in: [1f42218ded] user: stephan branch: trunk, size: 482667 | |
17:33 | Flags-related refactoring in prep for an attempt to increase some flag sizes without increasing sizeof() (with the help of existing padding). file: [58d3ac2d05] check-in: [cdd0f4d93f] user: stephan branch: trunk, size: 482604 | |
14:11 | Added CWAL_CONTAINER_DISALLOW_PROTOTYPE_SET flag and CWAL_RC_DISALLOW_PROTOTYPE_SET result code, which are the prototype-set counterparts of CWAL_CONTAINER/RC_DISALLOW_PROP_SET, and applied them to s2_seal_container()-sealed objects (which includes the new s2out). Reworked s2 enum construction so that setting the prototype property there is well-defined (though probably ill-advised, and replacing enum prototypes may yet be removed as a feature). Removed some dead code. file: [03a71f267e] check-in: [9484a7e93e] user: stephan branch: trunk, size: 482416 | |
2019-11-18
| ||
08:08 | Fixed a typo and reordered a couple if/else conditions. No functional change. file: [64c9932337] check-in: [bb6aa1c237] user: stephan branch: trunk, size: 483860 | |
2019-09-11
| ||
20:31 | Incidental internal doc touchups and removal of one unnecessary #if block. No functional changes. file: [840dc2cc0d] check-in: [e15d5c7b72] user: stephan branch: trunk, size: 483860 | |
2019-08-07
| ||
13:30 | Doc typo fix. file: [8d53f57a50] check-in: [97f7fa77c8] user: stephan branch: trunk, size: 483826 | |
2019-08-04
| ||
12:46 | Insignificant internal docs tweaks and factored out one superfluous variable. file: [d5f9479ce1] check-in: [832b98af5a] user: stephan branch: trunk, size: 483827 | |
2019-07-17
| ||
19:16 | Exposed the cwal-internal hashing function via the public API, as cwal_hash_bytes(), because i have a good use for it in the UUID module. file: [9f80fc7f2e] check-in: [34e21ebeca] user: stephan branch: trunk, size: 483828 | |
2019-07-13
| ||
01:19 | Fixed an ancient string ref mishandling in the JSON parser. No idea why it waited until now to show up :/, but that code certainly dates back to the time where i thought it was generally okay to temporarily have no refs. file: [906cd372db] check-in: [ac444d043c] user: stephan branch: trunk, size: 483766 | |
2019-07-06
| ||
07:21 | Fixed core property lookup to always perform type-strict comparison if either the lookup key or property key is of type CWAL_TYPE_BOOL. Without this, X[true] was resolving to the first truthy property key, and X[false] to the first falsy one. file: [e2e0b8d744] check-in: [a32fcc4268] user: stephan branch: trunk, size: 483688 | |
2019-03-26
| ||
07:33 | Fixed cwal_cstr_to_int() to not succeed if it sees no digits, e.g. a standalone minus sign. file: [4254b1c318] check-in: [5fc02ab14d] user: stephan branch: trunk, size: 482373 | |
2018-12-04
| ||
14:04 | Minor errno-handling tweak. file: [d1449e8cef] check-in: [a1dc0941fb] user: stephan branch: trunk, size: 482281 | |
2018-12-01
| ||
16:32 | Minor error handling tweak which might or might not do anything. file: [f46de3ae86] check-in: [978c1c04d2] user: stephan branch: trunk, size: 482255 | |
2018-11-30
| ||
18:11 | Minor tweaks/cleanups. file: [2b189c233a] check-in: [05e167e2eb] user: stephan branch: trunk, size: 482062 | |
2018-11-29
| ||
14:43 | Minor tweaks caught while porting some of this code into another tree. file: [d7504f5cba] check-in: [4398c5bec4] user: stephan branch: trunk, size: 482712 | |
2018-11-28
| ||
14:43 | Removed a stray comma from the new metrics dump output. file: [e628e9ee1f] check-in: [40423dbcc0] user: stephan branch: trunk, size: 482968 | |
14:33 | Core lib doc updates/additions. Added cwal_engine::metrics::len1StringsSaved to record how many length-1 strings are saved via that optimization. Moved the length-1 ASCII optimization flag into cwal_config.h. file: [1e347945e7] check-in: [39016bac11] user: stephan branch: trunk, size: 482970 | |
2018-11-27
| ||
11:12 | Several deeply internal assert()/fatal condition check improvements and cleanups. file: [100bb09b99] check-in: [af11b6ac1f] user: stephan branch: trunk, size: 480206 | |
09:25 | cwal_function_call_in_scope() now calls cwal_refunref() on the callback's result if its own caller passes NULL for the rv pointer (which, in practice, is never done). Fixed a missing cwal_strlen() in cwal_prop_take() (a holdover from when the API treated an explicit length of 0 to mean 'count it yourself'). Removed the last of the functions which were marked as @deprecated. file: [f8dcafc3d7] check-in: [7b251481bd] user: stephan branch: trunk, size: 480956 | |
07:38 | Added doxygen support, adjusted some docs to squelch doxygen warnings, removed several deprecated functions. Edit: doh, it seems we already had the doxygen files, they were just long forgotten and i overwrote them with copies from another source tree. file: [29695c0220] check-in: [73251240a6] user: stephan branch: trunk, size: 485534 | |
01:34 | Patch previous commit to politely overlook the NaN/Inf special values. The new s2 math module can emit those. file: [871ac39fca] check-in: [d951557a0d] user: stephan branch: trunk, size: 488046 | |
01:05 | Swapped out cwal_double_to_cstr() with an impl which doesn't write e-notation (because s2 can't read that (and i don't want to teach it to), so it shouldn't generate it). This caused several hours more hassle than it really should have, in the form of broken comparison operations and weird stringified output, but floating point numbers are strange, strange beasts. file: [8452e0c547] check-in: [6fcff747c1] user: stephan branch: trunk, size: 487895 | |
2018-11-26
| ||
16:31 | Improved/hardened the cwal_buffer_value() is-this-really-a-Value check. file: [818a61062c] check-in: [7db9dc8d92] user: stephan branch: trunk, size: 486698 | |
14:21 | removed some dead code. file: [d0c130b33c] check-in: [c2c5f3b1fd] user: stephan branch: trunk, size: 486510 | |
11:45 | Added cwal_scope_vacuum() in order to play with recursive vacuuming (which still breaks s2 because there are many places where we recurse into the eval engine). file: [d7d7b1429a] check-in: [014b06ff6d] user: stephan branch: trunk, size: 489451 | |
10:24 | Minor cost-correctness improvements. file: [0c1b9f6f58] check-in: [dea8b92b24] user: stephan branch: trunk, size: 489105 | |
2018-11-24
| ||
21:15 | cwal_value_unhand() now returns the value passed to it, instead of avoid, in the interest of making it easier to use for the most common usage. file: [b8007edc24] check-in: [7aefa0fc0e] user: stephan branch: trunk, size: 489093 | |
2018-11-23
| ||
14:43 | Snapshot. Lots of cleanup of the new s2_scope mechanism. Removed some now-obsolete s2 scope-handling functions. file: [196751fdd6] check-in: [b7de2a8e38] user: stephan branch: scope-push-pop-hooks, size: 489112 | |
07:48 | Initial work on scope push/pop hooks. The intent is so that clients (s2) can keep their own scope state in sync with cwal's, but whether that will really be possible is as yet uncertain because of engine init/startup timing vis-a-vis access to the allocator. file: [68d2da497b] check-in: [a1472de17e] user: stephan branch: scope-push-pop-hooks, size: 488914 | |
2018-11-22
| ||
15:10 | Removed some dead code: cwal_var_unset_s(). file: [3dad773c85] check-in: [b9357f7de4] user: stephan branch: trunk, size: 488434 | |
10:58 | cwal core: changed the string hashing algorithm yet again. Seems to perform better (fewer collisions). file: [36a5d2743e] check-in: [e0524c4ed3] user: stephan branch: trunk, size: 488574 | |
02:41 | minor constness improvements. file: [ff8baf4800] check-in: [028f3467cf] user: stephan branch: trunk, size: 488166 | |
02:29 | Another tweak to that just-fixed (but really re-broken) assert. file: [f3ff83c103] check-in: [59288dfffb] user: stephan branch: trunk, size: 488118 | |
02:07 | Fixed a bogus assert() in cwal_function_call_in_scope2() which was triggered by old C++ test code. file: [441edf3fe5] check-in: [c04f1f3314] user: stephan branch: trunk, size: 487677 | |
2018-11-20
| ||
09:16 | Squelched a couple non-debug build warnings. file: [224e91e676] check-in: [2656a0cb3c] user: stephan branch: trunk, size: 487633 | |
09:05 | Squelched dozens of unused parameter warnings (removing their names causes a completely different warning) triggered by work on the build process. Fixed cwal_scope_chain_set_with_flags() to not ignore the flags it was given and to honor various KVP/container flags which its _v() counterpart was honoring. file: [3ad79c9cdb] check-in: [211af69f84] user: stephan branch: trunk, size: 487617 | |
2018-11-07
| ||
18:19 | Enhanced cwal_value_from_arg() a bit. file: [5f68b30dea] check-in: [3fafa494c7] user: stephan branch: trunk, size: 486284 | |
2018-11-04
| ||
11:22 | Cleaned up ref/unref handling in cwal_parse_argv_flags(). Its previous behavour could have caused Grief with interned strings. file: [ca1c31625f] check-in: [1a2383df89] user: stephan branch: trunk, size: 485809 | |
2018-10-16
| ||
05:12 | Internal doc/comment touchups. Changed the range of builtin integer values (again - it was set to some weird range). Removed a couple unused/unnecessary macros. file: [25d55426a8] check-in: [0689e96ea9] user: stephan branch: trunk, size: 485199 | |
04:45 | Typo fixes (future dates). file: [ee54f05e71] check-in: [f6afbb8ba4] user: stephan branch: trunk, size: 484526 | |
2018-09-19
| ||
17:36 | Fixed a bug in the low-level object cleanup which caused refcount points on prototypes to not be properly removed. Discovered by accident while working on a new module. file: [82f1e8c3ea] check-in: [5e7ebb45df] user: stephan branch: trunk, size: 484526 | |
2018-08-12
| ||
10:07 | Fixed gcc7 warnings, a couple of which were valid and many of which were from its new, stunningly braindeaded behaviour of warning about switch fallthrough (which of course fails because we build with -Wall -Wpedantic). file: [bf0a251c1f] check-in: [6b2c51d9c8] user: stephan branch: trunk, size: 484964 | |
2018-06-20
| ||
14:32 | Stateful array qsort no longer uses a random pivot because checkin [f40e1afaed]'s removal of a call to srand() revealed a bug with that approach (and i'm not up for chasing down exactly what the problem is). file: [cf90bebaa9] check-in: [8c4d44300e] user: stephan branch: trunk, size: 484980 | |
2018-05-23
| ||
15:54 | Array sorting with a custom callback now sets the is-visiting flag for the duration of the sort, to prohibit iteration during the sorting process. file: [6cab7ac28c] check-in: [2fa3e2286d] user: stephan branch: trunk, size: 484104 | |
15:09 | Replaced the broken internal cwal_array_sort_stateful() qsort with a working one and added a way for non-exception errors to be propagated back out of cwal_array_sort_stateful() callbacks. file: [91d3580ddd] check-in: [75cd61d03c] user: stephan branch: trunk, size: 483718 | |
13:09 | added a todo about broken error propagation in array.sort(callback). Also discovered a sorting bug in that context which was yet to be properly identified. file: [f07072cad5] check-in: [b90276c8b6] user: stephan branch: trunk, size: 484593 | |
2018-05-08
| ||
13:29 | Changed how cwal_next_prime() works, utilizing the pre-existing 1000 primes list. file: [ee4324059f] check-in: [a02432c152] user: stephan branch: trunk, size: 484076 | |
2018-05-07
| ||
23:31 | minor internal tinkering. file: [c35b5753da] check-in: [9211afe535] user: stephan branch: trunk, size: 483812 | |
22:38 | minor metrics formatting tweaks and one pedantic micro-optimization. file: [8d6be43c0d] check-in: [301300b9e7] user: stephan branch: trunk, size: 483903 | |
2018-04-11
| ||
00:34 | typo fix, removed some duplicate/unused code. file: [00786f8652] check-in: [3b3f00851a] user: stephan branch: trunk, size: 483921 | |
2018-01-05
| ||
14:33 | Extended the Unique-type cleanup checks ([74e71fae100d44]) to Tuples, which hypothetically have the same cleanup error case (but it's never been witnessed). Edit: and accidentally checked in unrelated termbox tinkering at the same time :/. file: [2e0b9129e8] check-in: [eae27e464a] user: stephan branch: trunk, size: 483921 | |
13:40 | cwal core now gc-queues all values (not just containers) during scope cleanup. This fixes a difficult-to-trigger cleanup-time problem involving Unique-type values and their non-container wrapped values. file: [8934a4a74c] check-in: [74e71fae10] user: stephan branch: trunk, size: 481338 | |
2018-01-03
| ||
16:04 | consistency: changed signature of cwal_prop_has() and cwal_prop_has_v(), removed cwal_prop_has_s(). file: [0200054e04] check-in: [db1a5c3526] user: stephan branch: trunk, size: 480520 | |
2018-01-02
| ||
04:28 | generic tinkering. file: [f22bfa0754] check-in: [bbba146715] user: stephan branch: trunk, size: 480769 | |
2017-12-27
| ||
12:36 | removed some dead code, minor internal touchups. file: [b97eb58387] check-in: [e06340a5d2] user: stephan branch: trunk, size: 486740 | |
11:41 | removed some dead code. file: [fcc7438c4e] check-in: [0be8130cf2] user: stephan branch: trunk, size: 487648 | |
11:30 | Minor refactoring in the scope of interceptor support. Changed the cwal-internal string hash routine again. Added some metrics output. file: [54fbd62e38] check-in: [47e9b8e8a6] user: stephan branch: trunk, size: 488671 | |
10:10 | Fixed some ref/unref boneheadedness in cwal_json which prematurely killed values (bug introduced in [ad93dd9b93e37763]). Added missing ref/unhand in cwal_scope_pop2() which killed propagating values. file: [05435a3424] check-in: [957e87d7f8] user: stephan branch: trunk, size: 487181 | |
2017-12-26
| ||
16:29 | Didn't mean to hit enter there. file: [f663e92da1] check-in: [3af17e6d5d] user: stephan branch: trunk, size: 485889 | |
16:28 | Cleanups and preliminary work for adding property interceptors to the core via alternative getter/setter APIs. file: [c23f4f296e] check-in: [ed485b7b8b] user: stephan branch: trunk, size: 485889 | |
10:24 | Removed more superfluous funcs. Added one unit test. file: [ca9db3fdcd] check-in: [ad93dd9b93] user: stephan branch: trunk, size: 479421 | |
09:59 | removed some old/unused functions. file: [f9abc39165] check-in: [f5570c4333] user: stephan branch: trunk, size: 480827 | |
2017-12-25
| ||
12:04 | Removed deprecated cwal_args_callee_state(). Minor doc tweaks. file: [67a8de7fa9] check-in: [56e46c52bd] user: stephan branch: trunk, size: 481232 | |
10:42 | minor tweak to the string is-ascii determination. file: [84f2296458] check-in: [817bebac7a] user: stephan branch: trunk, size: 481563 | |
2017-12-08
| ||
14:32 | Added cwal_scope_pop2(), analog to s2_scope_pop2(). file: [1597c2fb5c] check-in: [09c10d90c0] user: stephan branch: s2-interceptors, size: 481087 | |
2017-12-07
| ||
00:10 | Minor proactive rescoping fixes. file: [212d07decf] check-in: [6848b566bc] user: stephan branch: trunk, size: 480698 | |
2017-12-06
| ||
19:13 | Apparently fixed an old bug in cwal_prop_get_kvp(). A slight semantics change to cwal_kvp_flags_set(). Tweaks to how cwal_function_call_in_scope() handles references. Fixed an old missing result code check which somehow never caused grief (or did it?). file: [bed819cb5d] check-in: [e491db6af4] user: stephan branch: trunk, size: 480550 | |
2017-12-03
| ||
21:09 | Tiny random cleanups. file: [23890ef119] check-in: [ed33242e31] user: stephan branch: trunk, size: 480709 | |
15:16 | Added cwal_build_info() to make various build-time config options available to client code. Stole another bit from the maximum string length to add an is-ascii flag to strings, which we can use to speed up many algorithms for the most common cases (but it slows down string creation for the common case). file: [a290ef81d3] check-in: [72236d95aa] user: stephan branch: trunk, size: 480709 | |
2017-12-02
| ||
14:33 | Made the range of built-in integers build-time configurable and removed the shared-instance integer cwal_weak_refs. Removed some cruft from CWAL_BUILTIN_VALS. file: [1f8410ec99] check-in: [79988ded9c] user: stephan branch: trunk, size: 475325 | |
2017-12-01
| ||
19:26 | Added cwal_engine_outputer_get/set() to simplify swapping out the outputer. Added cwal_output_f_cwal_outputer(). file: [5a3982e9ba] check-in: [2a21245d11] user: stephan branch: trunk, size: 475160 | |
2017-11-30
| ||
18:32 | a minor const correctness fix in cwal_unique_wrapped_get(). file: [a9b4f29890] check-in: [38620308fe] user: stephan branch: trunk, size: 474543 | |
2017-11-21
| ||
17:05 | added a reminder to self. file: [a552b2c671] check-in: [35baac6ef3] user: stephan branch: trunk, size: 474537 | |
15:57 | Fixed the length-1 string handling for X- and Z-strings. file: [16cc18f176] check-in: [717abf2614] user: stephan branch: trunk, size: 474347 | |
12:43 | cwal core now includes all byte-length-1 ASCII strings in its builtin Values. In the various s2 unit tests this cuts alloc counts by anywhere from ~1% to ~12% with recycling disabled and ~1% to ~2.5% with recycling enabled. Total memory use without recycling was cut by ~1-10% and is highly script-dependent (those with many length-1 variable names save more). file: [8ed4bb8a13] check-in: [8bef0b5c74] user: stephan branch: trunk, size: 473234 | |
2017-11-13
| ||
12:26 | minor utility API renaming and additions. file: [23a10e314e] check-in: [d8df88227b] user: stephan branch: trunk, size: 471272 | |
09:38 | Moved ownership of s2's eval-holder array from the eval() routine to s2_scope. This cuts allocations drastically when recycling is disabled, reducing the unit test amalgamation alloc counts by about 1/3rd and saving 1M of total alloc'd memory. The savings when recycling is on are minimal but better than nothing. file: [19db9c811a] check-in: [61f4780ed8] user: stephan branch: trunk, size: 470659 | |
07:08 | Refactored s2's various assignment operators for better ref/unref handling and readability. file: [fb218738fa] check-in: [557510fcb9] user: stephan branch: trunk, size: 470666 | |
2017-11-12
| ||
10:49 | Discovered and applied gcc's -Wextra warning level, plus patched up violations it caught (mostly missing struct member initializers). file: [eafec55b8c] check-in: [f13f6c3c4b] user: stephan branch: trunk, size: 470672 | |
09:06 | Migrated s2.Buffer.replace() implementations into core cwal functions. file: [e500a18593] check-in: [e980d01913] user: stephan branch: trunk, size: 470640 | |
2017-11-10
| ||
20:49 | cwal_value_from_arg() now trims starting/ending quotes from the value part of --key="value". file: [3ae29f2b97] check-in: [f37e0dbf88] user: stephan branch: trunk, size: 467856 | |
2017-11-07
| ||
13:10 | reworked the change to cwal_buffer_fill_from_filename() from the previous commit. file: [c10c2c104f] check-in: [673581aadb] user: stephan branch: trunk, size: 467685 | |
13:06 | imported GLOB/LIKE pattern matching from sqlite3. Added s2.glob(glob,haystack[,globStyle]). file: [19458a8bb5] check-in: [4eac2f17a9] user: stephan branch: trunk, size: 467683 | |
2017-11-03
| ||
11:49 | Minor semantics change in cwal_exception_setfv() and friends (returning CWAL_RC_OOM as-is, rather than trying to create an exception object). file: [0ef9e8c0a3] check-in: [5e957a91c3] user: stephan branch: trunk, size: 467792 | |
00:59 | minor doc tweaks file: [a09efda190] check-in: [38b67c092e] user: stephan branch: trunk, size: 467508 | |
2017-10-13
| ||
12:36 | minor (and arguable) CWAL_TYPE_LISTMEM metrics tweak: no longer count a cwal_list_reserve() of a larger size as an alloc request if we already have enough memory in the list. file: [a0650e6dc3] check-in: [29f0e06050] user: stephan branch: trunk, size: 467410 | |
12:23 | s/\bUNUSED\b/CWAL_UNUSED_VAR/g. Minor s2 unit test tweak. file: [a616f4ddcb] check-in: [6eaa08f972] user: stephan branch: trunk, size: 467415 | |
2017-10-12
| ||
12:11 | Random tiny cleanups, auto-reformatting, and fixed (i think) CWAL_TYPE_LISTMEM allocation metrics tracking. file: [8bfd823449] check-in: [d7ba485ecd] user: stephan branch: trunk, size: 467388 | |
2017-10-07
| ||
13:38 | re-renamed a couple routines left mis-named by the strlen refactoring. file: [f52b103228] check-in: [a78283a20f] user: stephan branch: trunk, size: 466045 | |
2017-10-05
| ||
17:09 | Merged (private) strlen-rewrite branch. This eliminates the 'length of 0 means to use cwal_strlen()' semantics from the major APIs (there may be a couple small outliers). My hands managed to survive, with the left pinky being only ever so slightly sore. file: [4de44d576a] check-in: [ef477c5d1e] user: stephan branch: trunk, size: 466026 | |
2017-03-22
| ||
18:56 | random typos. file: [8302cb02c7] check-in: [390cab8ffd] user: stephan branch: trunk, size: 466333 | |
2017-03-20
| ||
11:56 | cwal core now sorts Object properties internally, making searches faster (on average). file: [562e5f7d62] check-in: [792bbb7ddd] user: stephan branch: trunk, size: 466454 | |
2016-09-09
| ||
19:21 | Adding a preliminary CMakeLists.txt for use in CLion. Also changed the UNUSED_VAR macro to an UNUSED attribute which appears to be more widely adopted by compilers. file: [ee769c5390] check-in: [d834ba36b3] user: calebg branch: s2-ndebug, size: 467519 | |
2016-08-30
| ||
17:54 | Adding some build information to cwal. file: [6a9d41867a] check-in: [a88bdfc8b2] user: calebg branch: cwal-ndebug, size: 467606 | |
16:26 | Removing NDEBUG and just using DEBUG to simplify sources. file: [6331870832] check-in: [a449dc869c] user: calebg branch: cwal-ndebug, size: 466813 | |
13:37 | Fixing up the cwal non-debug build: * Switched around the amal build so that cwal_amalgamation.* files can be used without including other files. * Added ENABLE_SPEED_OPTIMIZATIONS and ENABLE_SIZE_OPTIMIZATIONS to config.make - ENABLE_DEBUG=0 make - ENABLE_DEBUG=0 ENABLE_SPEED_OPTIMIZATIONS=1 make - ENABLE_DEBUG=0 ENABLE_SIZE_OPTIMIZATIONS=1 make * Replaced all debugging double negatives with just a positive expressions: "#if !defined(NDEBUG)" is now "#ifdef DEBUG" * Added the UNUSED_VAR macro to prevent the 'unused variable' compiler warning. * Removed the "force assert() to always work" section in the test.c file. * Added CWAL_AMALGAMATION_BUILD to test.c so that it's possible to easily build the test with the amal build. * Literred UNUSED_VAR everywhere it was needed in the test.c file. file: [400e69b9d2] check-in: [fce3c1d39b] user: calebg branch: cwal-ndebug, size: 466972 | |
2016-08-25
| ||
20:44 | Changed several *CWAL_INT/DBL derefs to memcpy() to avoid a bus fault on ARM. Had to work around gcc boneheadedness to do it :/. file: [049de06050] check-in: [4b1745c43f] user: stephan branch: trunk, size: 466918 | |
2016-08-20
| ||
10:32 | Fixed a not-_quite_-right assertion in the Unique-type's finalizer which could incorrectly trigger during scope cleanup. file: [96a7eecbf9] check-in: [1627a35a2c] user: stephan branch: trunk, size: 465886 | |
2016-03-09
| ||
09:44 | removed scope pointers from cwal_dump_value() and s2_dump_value() (they're just noise, as scopes are always stack allocated). file: [b0a55644a0] check-in: [104193e661] user: stephan branch: trunk, size: 465492 | |
2016-03-07
| ||
17:55 | Ported s2 String.indexOf() to cwal_utf8_indexof(). Fixed a couple wrong no-match return cases in indexOf() (was still returning undefined (historical semantics) in some cases). file: [44fc55d614] check-in: [8b8f0837cd] user: stephan branch: trunk, size: 465524 | |
2016-03-05
| ||
20:52 | Refactored s2's string case folding utilities into the cwal core. Need to eventually do most of the string utilities (indexOf, split, etc.), but i have an immediate use for this one. file: [41863a3ad0] check-in: [6ad30c288e] user: stephan branch: trunk, size: 467331 | |
2016-03-01
| ||
17:57 | corrected an assertion message. file: [e3e2a7fcfb] check-in: [226b9b7f00] user: stephan branch: trunk, size: 465525 | |
17:15 | removed some dead code, added an assertion to try to tickle and s2-level bug earlier on. file: [4b84fe72c8] check-in: [04e628eb7e] user: stephan branch: trunk, size: 465522 | |
16:36 | merged in trunk, resolved conflicts. file: [b19110c11a] check-in: [9128a3b5d8] user: stephan branch: core-prop-interceptors, size: 470343 | |
16:31 | removed an extraneous call which does the same thing as the thing which calls it. file: [dcdd0aa024] check-in: [75f2be92ce] user: stephan branch: trunk, size: 465805 | |
2016-02-29
| ||
11:44 | os-x compile warning fix from Warren (his compiler is lying and/or misbehaving, though). Edit: okay, not lying, but performing an unexpected (to me) conversion from uint16_t to int. file: [c5958941da] check-in: [de4c5e6afe] user: stephan branch: trunk, size: 465857 | |
2016-02-28
| ||
13:55 | Refactored the vacuum-safe flag from cwal_obase::flags to cwal_value::rcflags. s2_eval_ptoker() now uses a Unique instead of an Array for keeping its pending result safe (this is slightly cheaper), but in theory we can factor out that middleman (first attempt failed, so fell back to the Unique as an easy workaround). file: [fb58c30d89] check-in: [15092b9bd0] user: stephan branch: trunk, size: 465861 | |
11:26 | split off cwal_refcount_t from cwal_size_t so that we can eventually (if needed) up the refcounts+flags to 64-bits on 32-bit builds without increasing cwal_size_t (which would have a higher overall cost). Stole another (the 6th) refcount bit to allow us to move the vacuum-safe flag from containers to values. file: [4ea8898003] check-in: [6f574d3f55] user: stephan branch: trunk, size: 465745 | |
10:56 | merged trunk, resolved a conflict. file: [08229c15fc] check-in: [08244059ce] user: stephan branch: core-prop-interceptors, size: 470361 | |
2016-02-26
| ||
15:47 | child-rescoping fix for tuples (assertion triggered). file: [930eab5b90] check-in: [0c897fd853] user: stephan branch: trunk, size: 465718 | |
14:57 | tuple docs and an iteration error check fix. file: [7835437e3f] check-in: [3ceffbf8b9] user: stephan branch: cwal_value_pair, size: 465712 | |
13:12 | omg facepalm. The automatic resizing of hashes was losing their kvp flags. Thus scope storage failed to retain const flags only after having enough vars to cause it to resize. file: [26431203d2] check-in: [2ac9af21ff] user: stephan branch: cwal_value_pair, size: 465705 | |
01:25 | okay, got carried away: discovered that hashes as scope storage fail to trigger const violations on s2 assignments, at least in one of the unit tests (but not in the interactive shell, for reasons unknown). Too tired to debug it further now. file: [9834e2b1fc] check-in: [42ddd6cf8c] user: stephan branch: cwal_value_pair, size: 464526 | |
2016-02-25
| ||
23:54 | minor tuple touchups. good night. file: [8e597b05e3] check-in: [fe58f61202] user: stephan branch: cwal_value_pair, size: 463629 | |
23:38 | changed tuple hashing to just use the ptr-hash instead of hashing all elements. We might want to revision that later for use with tuples as hash keys (for use in simulating sets via hash tables). Edit: nevermind: hash keys' hashes have to be constant, and tuples are not, so this is the right hash algo. file: [99f56a5684] check-in: [b8fcba87c5] user: stephan branch: cwal_value_pair, size: 463541 | |
23:15 | a small start on the CWAL_F_IS_VISITING transition to CWAL_RFC_IS_VISITING. file: [db21713233] check-in: [1006c5a9dd] user: stephan branch: cwal_value_pair, size: 463418 | |
22:25 | cwal_value_may_iterate() now works for tuples. file: [70290a01b5] check-in: [efcb1523c8] user: stephan branch: cwal_value_pair, size: 463122 | |
22:18 | Added foreach() support for tuples, working like arrays. And stole another refcount bit to get the is-visiting flag for cwal_values (needed for tuples and uniques, though uniques don't currently use it (they really should)). So far 5 refcount bits stolen for flags. file: [3c561365c3] check-in: [515942ded1] user: stephan branch: cwal_value_pair, size: 463061 | |
21:42 | refactored pairs as tuples (fixed-length arrays). Still need to add foreach() support. file: [ecbfbcc56d] check-in: [5cc361ae64] user: stephan branch: cwal_value_pair, size: 462101 | |
19:25 | merged in trunk. file: [44efdc5464] check-in: [ec839ef681] user: stephan branch: cwal_value_pair, size: 461977 | |
17:12 | changed semantics of cwal_array_visit/visit2() when it comes to NULL entries (they are no longer skipped over, so that iteration can keep track of the indexes properly if it wants to, without having to jump through hoops to do so). file: [a56bedba9e] check-in: [8ecd168f4b] user: stephan branch: trunk, size: 455466 | |
2016-02-16
| ||
18:28 | i give up for tonight. file: [01122365ea] check-in: [8bbe9797b9] user: stephan branch: core-prop-interceptors, size: 459520 | |
18:09 | i had a property setter interceptor working, then broke it, and am now too tired (and this machine has no debugger). Someone's getting cleaned up when they shouldn't, random crashes. To look at later. file: [584e0051fb] check-in: [ef67b58df2] user: stephan branch: core-prop-interceptors, size: 459364 | |
14:37 | merged in trunk file: [98497073f9] check-in: [e5dfe6a291] user: stephan branch: core-prop-interceptors, size: 456584 | |
14:29 | merged in trunk file: [a6058aa5da] check-in: [7658116a71] user: stephan branch: cwal_value_pair, size: 461953 | |
12:11 | Got rid of cwal_unref_from(). s2 still works. file: [6698bf2fb4] check-in: [78bac73345] user: stephan branch: trunk, size: 455442 | |
10:45 | oops - enabled one of the #if's which allows th1ish to continue to work (it was disabled while testing yesterday). file: [52530de244] check-in: [71969d4b29] user: stephan branch: th1ish-R.I.P., size: 453868 | |
2016-02-15
| ||
19:49 | Added yet another marker (this one in the penultimate place: cwal_unref_from()) for a place which can be removed once we finally decide to get rid of th1ish. file: [a53874a4db] check-in: [aee732756b] user: stephan branch: trunk, size: 453868 | |
19:42 | Added a few markers in places which can be removed once we finally decide to get rid of th1ish. file: [5c9c3c2211] check-in: [555d3732b0] user: stephan branch: trunk, size: 453742 | |
19:37 | wrote what turned out to be a novella on plans for/thoughts on improving s2_estack to notably improve s2's internal value ref/unref handling (now that my understanding of the intricacies ref handling in the context of temp values, in particular in conjunction with string interning (or value interning in general), has finally caught up to the code\!). file: [7b407ba869] check-in: [dcf19ee9c8] user: stephan branch: trunk, size: 453512 | |
15:37 | added single highest-ever refcount to cwal_engine::metrics to help inform decision-making about hijacking more refcount bits for per-value flags. Currently 260-odd refs on s2's Object prototype in the s2 core unit tests. file: [f7790efe5f] check-in: [133d7cfdd9] user: stephan branch: trunk, size: 453340 | |
2016-02-13
| ||
22:54 | added a potential todo. file: [1d4ce4e2be] check-in: [1b4862fe9f] user: stephan branch: trunk, size: 452704 | |
22:23 | typo/doc/capslock fixes. file: [199bfe9966] check-in: [f6dc886276] user: stephan branch: trunk, size: 452588 | |
19:22 | Minor tweaks to get it compiling (again) with CWAL_SIZE_T_BITS=16. file: [54a1e4e3f1] check-in: [6509fa8dd3] user: stephan branch: trunk, size: 452591 | |
18:55 | Replaced the remaining relevant cwal_obase::flags with cwal_value:rcflags (did indeed need all 4 bits, but freed up 3 cwal_obase::flags bits for potental later use). Pseudo-containers (namely CWAL_TYPE_UNIQUE) can behave sanely during cleanup. e.g., they now participate in the gc queue and cycle-breaking during cleanup. file: [e29217b3a4] check-in: [c391d2ccce] user: stephan branch: value-refcount-flags, size: 452592 | |
17:39 | Renamed cwal_value::refcount to rcflags to ensure that there are no stray uses of it. file: [acc2bd6cdf] check-in: [79d6338ab2] user: stephan branch: value-refcount-flags, size: 450983 | |
17:25 | Initial refactoring for encoding a few bits of flags in cwal_value::refcount. Both th1ish and s2 tests pass with these changes, though we're internally encoding the refcount differently now. Initially 4 bits, but that will likely drop to 2. file: [6fd0dda919] check-in: [ac9cf983f2] user: stephan branch: value-refcount-flags, size: 450993 | |
16:00 | merged in trunk. file: [ddc20f1322] check-in: [12de6a8d35] user: stephan branch: core-prop-interceptors, size: 448461 | |
15:59 | Merged in trunk, as it's about to get messy with planned internal refcount refactoring. file: [49bc5426d9] check-in: [da81a83408] user: stephan branch: cwal_value_pair, size: 453803 | |
15:53 | Worked around a scope-cleanup crash caused by a Unique-typed value containing cycles (same problem we saw with cwal_value_pair experimentation). This case is a good argument for stealing bits from the refcount for a couple internal cwal_value flags. file: [0e35174460] check-in: [e1076c4252] user: stephan branch: trunk, size: 447204 | |
05:32 | Minor metrics dump output addition (total refcount of all interned strings). file: [4ef7fcb439] check-in: [745f3a1f2c] user: stephan branch: trunk, size: 445463 | |
04:46 | Fixed an internal misuse of cwal_string::length (failed to strip flag bits) which caused an OOM error (tried to alloc a few gazillion bytes). file: [228fc0aac5] check-in: [93f5e8f467] user: stephan branch: trunk, size: 444729 | |
2016-02-12
| ||
14:04 | minor change to how cwal_weak_ref is freed/allocated, to make use of the chunk recycler. Minor doc tweaks. file: [b7e58f10e6] check-in: [db36c3e32a] user: stephan branch: trunk, size: 444621 | |
01:09 | Added cwal_value_pair type, but shortcomings elsewhere make it clean up improperly when there's recursion involved (because it's not a full-fledged container, and is not privy to the cleanup treatment involved). Not sure if this is yet worth the hassle (and it asserts() now in s2 in very simple cyclic cases). They also can't be made explicitly vacuum-safe with the current code. file: [b577515e0c] check-in: [f505331bdb] user: stephan branch: cwal_value_pair, size: 451369 | |
2016-02-11
| ||
20:02 | merged in trunk. file: [c9c6c4243a] check-in: [66ba5effe5] user: stephan branch: core-prop-interceptors, size: 445853 | |
2016-02-06
| ||
19:07 | More hash experimentation. Found the worst hash ever. file: [7fb08dfbfc] check-in: [655faf1704] user: stephan branch: trunk, size: 444596 | |
18:53 | More tinkering with the core string hashing algo. Found one which works a tiny tick better than FNV-1a in basic s2 tests. Minor metrics output addition. file: [4db9ad8f0d] check-in: [85c4e2a579] user: stephan branch: trunk, size: 443834 | |
16:37 | Internal refactoring to how rescoping is handled to allow us to differentiate between is-rescoping and is-visiting. file: [44d9251d40] check-in: [9339197dd9] user: stephan branch: trunk, size: 441955 | |
12:51 | minor metrics output addition, enabled memchunk padding. file: [ebc37bb086] check-in: [c49259a801] user: stephan branch: trunk, size: 440823 | |
12:06 | Refactored cwal_native_rescoper_f() into cwal_value_rescoper_f() and consolidated the cwal_native/cwal_function APIs for those. It was easier than expected. file: [1b782b9c20] check-in: [599e5b4b81] user: stephan branch: trunk, size: 440207 | |
11:48 | Added a rescoper to cwal_function's state, analog to cwal_native. Now to consolidate them using the newer interface. file: [34803ed642] check-in: [e2d831d607] user: stephan branch: trunk, size: 440243 | |
02:08 | Fixed a param of type char which should have been int (char is unsigned on some platforms). Minor internal commentary and tinkering. file: [7be3fbcbf3] check-in: [4f63014a31] user: stephan branch: trunk, size: 439670 | |
01:10 | Added integer values 2..10 to the built-in constants. The malloc difference in s2's unit tests is absolutely negligible (a small handful). Maybe we'll crank it up to 100 someday. file: [632af8e73e] check-in: [7745e5b370] user: stephan branch: trunk, size: 439019 | |
2016-02-05
| ||
23:43 | Some experimentation with a different internal approach to property setting in some case (avoiding a second lookup). file: [a5b549ddbe] check-in: [09988759e5] user: stephan branch: trunk, size: 437268 | |
21:58 | fixed cwal_new_string() to work as expected when its argument starts with a NUL byte but has a positive length. file: [4128fe3265] check-in: [09c1d58eac] user: stephan branch: trunk, size: 435791 | |
20:17 | finally added String.replace() in s2. file: [85b5725656] check-in: [9254246116] user: stephan branch: trunk, size: 435728 | |
12:40 | Trying out my current understanding of refs vis-a-vis propagating return values. th1ish's ref handling is poor, though, so this breaks it. Maybe time to delete th1ish altogether (i would if it weren't for the demo CGI site). file: [efc92e6980] check-in: [6e466bc44c] user: stephan branch: trunk, size: 435585 | |
2016-02-03
| ||
17:38 | Fixed cwal_hash cleanup to behave like objects/arrays wrt unhand vs unref when cleaning up entries which live in older scopes during the container's scope's finalization. Minor internal refactoring in prep for (potentially) trying (again) to sort cwal_obase-level properties to speed up searches a bit. file: [ee886a40ea] check-in: [f916a73ed7] user: stephan branch: trunk, size: 434992 | |
2016-02-02
| ||
19:29 | merged in trunk. file: [fa11b4e10c] check-in: [7c7c6b1e0f] user: stephan branch: core-prop-interceptors, size: 438663 | |
2016-02-01
| ||
18:33 | insignificant internal touchups. file: [bcac2a6640] check-in: [052b0840a3] user: stephan branch: trunk, size: 437406 | |
18:02 | started adding some infrastructure to allow a cwal_engine to be marked as "dead", which basically means "in a debug build, it would have asserted by now, but this is not a debug build." When an engine is dead, it might have leaked memory to avoid manipulating memory which appears (to cwal) to have been corrupted or misused (99.9% of the time, unreferenced too many times). file: [6ac7dca52a] check-in: [0ca7f73ed8] user: stephan branch: trunk, size: 437392 | |
17:03 | a proactive fix in cwal_unref() for a hypothetical condition involving manipulating/upscoping scope property storage containers from client code (s2 or th1ish). file: [8bf3cc5371] check-in: [b097caff96] user: stephan branch: trunk, size: 437435 | |
12:37 | facepalm. lol. good thing this was not an important piece of code. file: [c470416bc5] check-in: [90e096d6a9] user: stephan branch: trunk, size: 435144 | |
2016-01-26
| ||
17:06 | more interning notes/warnings. file: [a5115f2fd5] check-in: [48bf6eead2] user: stephan branch: trunk, size: 435144 | |
16:54 | found another instance of string interning screwing up client code by wiping values out from under it, added a couple assertions and halfway informative crash messages. file: [09bd91248f] check-in: [1c111ca778] user: stephan branch: trunk, size: 434942 | |
2016-01-25
| ||
20:13 | consolidated some duplicate code. file: [95b17e320f] check-in: [5be320aab0] user: stephan branch: trunk, size: 433786 | |
18:49 | experimenting with cwal scopes copying const-flagged vars from parent scopes when a local lookup fails, but this can change script semantics, namely typeinfo(islocal...) and declaration of locals with the same name after a higher-up one has been copied in. file: [c2a060fbe3] check-in: [cb9e1ef96d] user: stephan branch: scopes-copy-consts-on-lookup, size: 436052 | |
2016-01-23
| ||
19:29 | disable cwal_prop_set() and friends when finalizing (can theoretically only be seen by clients in the context of a cwal_native/Native circular reference manipulated in the Native half's finalizer. file: [ccbf0901a0] check-in: [18143f714e] user: stephan branch: trunk, size: 435239 | |
16:48 | added cwal_var_set_with_flags(). file: [47110fa206] check-in: [b3f8ba341e] user: stephan branch: trunk, size: 434814 | |
12:01 | minor semantics change in cwal_args_state() and cwal_function_state_get(). file: [3b29b6a0bf] check-in: [6bbcd9bcef] user: stephan branch: trunk, size: 434582 | |
2016-01-22
| ||
15:25 | merged in trunk. file: [e3e614e949] check-in: [e75bac64fb] user: stephan branch: core-prop-interceptors, size: 435801 | |
2016-01-21
| ||
21:41 | fixed a regression in cwal_var_decl() which allowed a variable to be declared more than once without error. Caused by the recent addition of hashes as scope-level property storage. file: [29ceb860d7] check-in: [dae15968d8] user: stephan branch: trunk, size: 434544 | |
09:06 | renamed cwal_hash_search_v_kvp() to cwal_hash_search_kvp_v() for consistency with recent similar changes. file: [0245d316c6] check-in: [aae2daf7a3] user: stephan branch: trunk, size: 434629 | |
2016-01-13
| ||
20:38 | some prep work for property interceptors experimentation. file: [442650ac46] check-in: [5983b9b959] user: stephan branch: core-prop-interceptors, size: 435886 | |
17:34 | moved some local vars around. file: [42570727d3] check-in: [69b672efc3] user: stephan branch: trunk, size: 434629 | |
2016-01-12
| ||
19:52 | tweaked cwal_value_hash_ptr(). file: [26945c68f5] check-in: [e49749bbad] user: stephan branch: trunk, size: 434601 | |
2016-01-11
| ||
18:29 | Added CWAL_VERSION_STRING config option and populate it from the fossil info if it is available. added --version option to s2sh. file: [0294337375] check-in: [035cefe320] user: stephan branch: trunk, size: 434530 | |
17:30 | added some comments about the workaround from the previous commit. re-enabled a chunk of s2 which triggered the problem. file: [9058aed737] check-in: [22b1da9ca8] user: stephan branch: trunk, size: 434243 | |
16:53 | added workaround: string interning no longer re-uses an interned string when a copy in the interning table has a refcount of 0. Interned values should now not introduce any weird/hard-to-find differences from other values. Incidentally also undid a recent code tweak which broke the cwal_value_unref() return value expectations. file: [32105cfc5b] check-in: [da9d7bdceb] user: stephan branch: trunk, size: 433766 | |
16:14 | characterized/documented an inherent incompatibility with string interning and the use of temp values (specifically, strings, with a refcount of 0), in which a downstream call can (_seemingly_ legally) nuke a value created higher-up in the call chain, because string interning confuses the "real" ref count of temp values. Aha... there might be a workaround for this in interning itself (don't use interned copies if their refcount is 0). file: [acb1e31cd6] check-in: [970642d496] user: stephan branch: trunk, size: 432996 | |
2016-01-10
| ||
14:37 | added cwal_memcap_config.forceAllocSizeTracking and the -fst/--fst flags to s2sh to toggle it. file: [f4abe09a62] check-in: [d1f8344ad3] user: stephan branch: trunk, size: 432208 | |
05:43 | Changed how a ref is managed in cwal_hash_insert_with_flags(). file: [86f287f83c] check-in: [be54ad2d8b] user: stephan branch: trunk, size: 431427 | |
2016-01-08
| ||
17:09 | large batch of tweaks, largely related to the hash-vs-objects work, but also an attempt at guaranteeing under/overflow behaviour on integer ops. file: [b5464315b2] check-in: [4d688d1ffc] user: stephan branch: s2-scope-hashes, size: 431383 | |
14:55 | large batch of tweaks, largely related to the hash-vs-objects work, but also an attempt at guaranteeing under/overflow behaviour on integer ops. file: [4044c05986] check-in: [b79462d915] user: stephan branch: s2-scope-hashes, size: 430706 | |
01:59 | scopes seems to work with both hash and property storage. still doc cleanups to make and valgrind output to compare. bed time. file: [f945c1524a] check-in: [97b6c34301] user: stephan branch: s2-scope-hashes, size: 431000 | |
00:40 | most/much of the groundwork for making scopes be capable of using either objects or hashtables for property storage. file: [54ff2f7f29] check-in: [99e0b10067] user: stephan branch: s2-scope-hashes, size: 428355 | |
2016-01-07
| ||
19:19 | minor internal refactoring to facilitate scopes which can optionally use hashes instead of objects for property lookup. Also imposed string length limits, as too-big strings would overwrite the 2 bits of encoded data in cwal_string::size. Removed cwal_new_number(). file: [8b5872cc52] check-in: [bedc1cdb2a] user: stephan branch: trunk, size: 426366 | |
17:41 | minor internal commenting. file: [f76cd124eb] check-in: [1a452f999b] user: stephan branch: trunk, size: 426019 | |
14:56 | merged enum-with-value branch to trunk and extended cwal_array_index_of() to take a flag to control type strict/non-strict comparison. file: [b3578d1f37] check-in: [1f2e988254] user: stephan branch: trunk, size: 424731 | |
2016-01-06
| ||
15:58 | expanded the cwal_next_prime() list a bit. file: [c90bc93e0a] check-in: [24483b6c83] user: stephan branch: enum-with-value, size: 424494 | |
15:15 | port cwal_hash_grow_if_loaded() and cwal_next_prime() from s2 into the core. file: [b44659900b] check-in: [cb58cd5a46] user: stephan branch: enum-with-value, size: 424447 | |
2014-12-20
| ||
10:44 | minor heuristic tweak in cwal_malloc2(). file: [22799607d4] check-in: [50fbcf8cc5] user: stephan branch: trunk, size: 422719 | |
2014-12-19
| ||
02:24 | added cwal_malloc2(), which first checks the chunk recycler, to enable a recycling optimization in Native implementations. file: [0939c48df5] check-in: [e58b15dc3f] user: stephan branch: trunk, size: 422189 | |
2014-12-17
| ||
22:12 | Minor const-related refactoring related to, and notes regarding, property accessor interceptors, which i would like to explore once the hand allows it. file: [bea1f67823] check-in: [063462ebcd] user: stephan branch: trunk, size: 422098 | |
04:32 | core: internal refactoring to make Buffer Values containers, so they may now hold properties and be prototypes. It was much easier than expected to implement and the cost increase is negligible (48 bytes) in the amalgamated s2 unit tests. file: [13648a7025] check-in: [73df044fc1] user: stephan branch: trunk, size: 420119 | |
2014-12-16
| ||
20:19 | refactored cwal_value_native_part() to take an optional type ID, and (if not 0) return only on a match. Fingers have stopped working again, so this is it for a while :/. file: [48a71c396e] check-in: [a4ab970b93] user: stephan branch: trunk, size: 420598 | |
19:22 | added missing total memory use constraint checks. file: [22960c9ec4] check-in: [33d4a0dd78] user: stephan branch: trunk, size: 420473 | |
17:40 | Added metrics for "slack" bytes recovered by the chunk recycler. file: [97461c082b] check-in: [dc4ebce93d] user: stephan branch: trunk, size: 420043 | |
16:14 | more metrics dump prettification. file: [cc6f9b1d2d] check-in: [e0149376a0] user: stephan branch: trunk, size: 419586 | |
15:04 | core: when memory size tracking is on, make use of it in the chunk recycler to reclaim bytes which get sometimes get "truncated" during recycling of oversized chunks. file: [d3ade57491] check-in: [9c7e086e74] user: stephan branch: trunk, size: 418792 | |
2014-12-15
| ||
18:43 | When running in over-allocating mode, disallow allocations >32 bits on 64-bit builds b/c the size counter is internally uint32_t for the time being (for the sake of 32-bit builds). file: [54dfd97875] check-in: [cc38a84bf8] user: stephan branch: trunk, size: 416891 | |
2014-12-14
| ||
20:57 | minor metrics fix. file: [59ca6dcd64] check-in: [8fcec39269] user: stephan branch: trunk, size: 416561 | |
15:12 | core: added optional allocation count/amount capping, in support of Caleb's site. Refactored the Unique type to act like a wrapper for a single value. My hand has had it - enough hacking for a good while :(. file: [b5104b1af1] check-in: [8cb797fc88] user: stephan branch: trunk, size: 416560 | |
01:14 | minor refactoring of/additions to the hash API in prep for eventual additions in s2. file: [f5f559d58a] check-in: [e6829b71af] user: stephan branch: trunk, size: 411474 | |
2014-12-13
| ||
23:14 | Minor hash code calc change in the core, removed the comma from typeinfo(TAG EXPR), added unit tests for typeinfo. file: [a83c88ebf4] check-in: [663f4dff66] user: stephan branch: trunk, size: 411052 | |
22:20 | After a 3.5-day Typing-Verbot, i exercised my hand implementing some low-hanging fruit in the s2 typeinfo keyword. Fixed a return type inconsistency in cwal_value_exception_part(). Some minor efficiency hacks in the cwal_value_XXX_part() routines. Added the missing cwal_value_object_part(). Now back to the Typing-Verbot. file: [ae4d8a3835] check-in: [339cea5e4d] user: stephan branch: trunk, size: 411025 | |
2014-12-09
| ||
18:42 | Reverted the buffer realloc part of [8a4bd944bdca], as it causes +50-60 allocs in the s2 amalgamated unit tests. Once again, a hard lesson in over-optimization. file: [b3f9eed84b] check-in: [38f581b85a] user: stephan branch: trunk, size: 410762 | |
2014-12-08
| ||
20:40 | moved s2_refunref() to the core cwal_refunref() and reimplemented it there to be notably more efficient (since it has direct access to the internals). s2: caught another formerly stray temporary value in the short-circuiting logical/ternary ops. Allocs in the s2 amalgamated unit tests are up notably since yesterday (but still lower than last week) but my fingers are killing me so investigating why will have to wait. file: [be09b68336] check-in: [5ebcca7477] user: stephan branch: trunk, size: 410570 | |
2014-12-07
| ||
09:35 | core: accounted for a NUL byte in z-string recycling, cwal_buffer_to_zstring() now shrinks its memory if it is "too big" (buffer recycling could give us a 20k chunk for what turns out to be a 50-byte z-string). file: [bf474cf25c] check-in: [8a4bd944bd] user: stephan branch: trunk, size: 410377 | |
04:29 | Migrated from using the result code CWAL_RC_ACCESS to CWAL_RC_IS_VISITING in many cases to avoid ambiguities in their interpretation in higher-level code. Cached a few more commonly-used strings. file: [a56f26f632] check-in: [52824f4adc] user: stephan branch: trunk, size: 409884 | |
00:24 | cwal_props_clear() now honors CWAL_CONTAINER_DISALLOW_PROP_SET. Started marking places where CWAL_RC_ACCESS needs to be replaced with CWAL_RC_CONCURRENT_VISITS. file: [de800d07a0] check-in: [aabeccc821] user: stephan branch: trunk, size: 410045 | |
2014-12-06
| ||
22:58 | metrics-related tweaks. file: [926732eecd] check-in: [780d6a11c1] user: stephan branch: trunk, size: 409898 | |
20:33 | core: went ahead and added a numeric value to "unique" values, but its value is unspecified (except that it will be unique amongst all live unqique-type values). Change how uniques are compared (using their numeric value). file: [279e747fae] check-in: [8d5f65f33a] user: stephan branch: trunk, size: 409943 | |
18:46 | Refactored cwal_printf_cstr() to use the cwal_engine allocator. file: [211f3b980c] check-in: [9f0c85b080] user: stephan branch: trunk, size: 409230 | |
16:30 | core: more work on the various container flags, s2: more work on the enum keyword, other minor cleanups. file: [dd78189b4a] check-in: [8c61a703a0] user: stephan branch: trunk, size: 406894 | |
00:55 | minor chunk recycler optimization experiment. file: [804a904540] check-in: [42701dac86] user: stephan branch: trunk, size: 406404 | |
2014-12-05
| ||
22:44 | more work on the container flags, added a couple missing cwal_value_is_XXX() variants, minor API renamings. file: [155d604028] check-in: [38060dc5e9] user: stephan branch: trunk, size: 406132 | |
22:22 | core: removed obsolete/duplicated function, doc updates. file: [41b45798e7] check-in: [104ffccfd9] user: stephan branch: trunk, size: 406203 | |
21:47 | core: added an internal flag to cap single max alloc size. file: [2699691b8d] check-in: [d850e44911] user: stephan branch: trunk, size: 406302 | |
19:49 | core: reimplemented cwal_value_upscope() with new semantics. Added the seemingly missing upscoping of prototype pointers. file: [2aa61e3a27] check-in: [3b90d0a7c8] user: stephan branch: trunk, size: 406373 | |
18:38 | compile fixes for non-debug builds. file: [9c2bd0de2d] check-in: [6b170364c7] user: stephan branch: trunk, size: 405681 | |
18:11 | minor tweak to the chunk allocator (now allows chunks of one byte smaller). file: [c9572cf7d6] check-in: [a0e3a37b18] user: stephan branch: trunk, size: 405390 | |
17:33 | Changes in semantics how cwal_prop_set() and friends report various cases, so that we can move some constraint checks into cwal, so that s2 does not have to do duplicate lookups to get the same feature. file: [f91c5fb0e1] check-in: [6dbf89b73c] user: stephan branch: trunk, size: 405391 | |
13:05 | core: refactored the chunk allocator to remove all dynamic memory overhead plus all overhead related to the tracking of each chunk (it is now stored in the chunk itself). This means we can only recycle stuff larger than (sizeof(void*)+sizeof(cwal_size_t)), but we were doing that, anyway, except perhaps in some tiny buffer cases. file: [6a9b20c674] check-in: [435ccd17b7] user: stephan branch: trunk, size: 404253 | |
11:22 | metrics dump header fix for recycling bins. file: [44e782aa93] check-in: [6c301265d9] user: stephan branch: trunk, size: 409013 | |
01:31 | core: metrics tweaks. Fixed mis-recycling of cwal_weak_ref. Added new "unique" type for use as opaque sentry values and a sort of enum type. file: [e535975c01] check-in: [d23084f162] user: stephan branch: trunk, size: 408979 | |
2014-12-04
| ||
21:55 | core: platform workaround for my ARM, which is bus-faulting on a particular deref of a static/const pointer. Accessing the same pointer via anoether route does not bus-fault it. file: [27874eafc2] check-in: [319b16b847] user: stephan branch: trunk, size: 407669 | |
19:31 | removed two now-redundant calls to the chunk recycler. file: [0b9d182bfd] check-in: [ccab3e2a09] user: stephan branch: trunk, size: 407323 | |
17:47 | Minor internal cwal_array_set() optimization. file: [6cba82e0df] check-in: [5d80eb7a5f] user: stephan branch: trunk, size: 407380 | |
15:42 | core: added a couple new memchunk metrics, disabled the memchunk recycler as the string recycling fallback because it is triggering a valgrind warning via memset() via cwal_value_new() - to be investigated later. file: [c224374adb] check-in: [0687ce7bc3] user: stephan branch: trunk, size: 407298 | |
2014-12-03
| ||
16:59 | Added cwal_hash_take_props() and s2.Hash.takeProperties(). s2.Hash.new(N) no longer modifies N to ensure it is prime. Fixed a redundant call to the chunk allocator. file: [f7ceaaea0b] check-in: [15810833d0] user: stephan branch: trunk, size: 406299 | |
2014-12-02
| ||
21:21 | core: added a todo. file: [857e64d001] check-in: [c4c02a716f] user: stephan branch: trunk, size: 402558 | |
18:36 | Added cwal_container_flags_set(), cwal_container_flags_get(), made some experimental tweaks to when containers use cwal_unref_from() during cleanup (only when their owning scope is being finalized). file: [098c7f8a12] check-in: [71b228faa9] user: stephan branch: trunk, size: 402129 | |
2014-12-01
| ||
21:48 | core: string allocation will now fall back to the memory chunk handler in some cases. file: [0e8714d522] check-in: [e87f73283e] user: stephan branch: trunk, size: 401217 | |
18:42 | checking in an experiment in sizeof(cwal_obase) to see how it affects ARM. file: [e104adde3a] check-in: [9e3cef307f] user: stephan branch: trunk, size: 399677 | |
2014-11-30
| ||
15:54 | Added cwal_free2() to potentially send memory to the recycler. file: [986f52cc9b] check-in: [058951d311] user: stephan branch: trunk, size: 399448 | |
11:31 | added type information to the recycle bin metrics output, s/cwal_value_type_id_name/cwal_type_id_name/g file: [8bc7336f3e] check-in: [5411fd86ba] user: stephan branch: trunk, size: 399243 | |
2014-11-29
| ||
20:59 | Factored out cwal_memchunk::left. 90% of the code was either setting it or just asserting that it was set. file: [e162139af5] check-in: [3fc1976b1d] user: stephan branch: trunk, size: 396911 | |
20:31 | Moved cwal_engine::eWeak into cwal_engine::recycler, fixed consolidation of Value- with non-Value types in a single recycler: each recycler must be specific to a certain base type so that we can use the list-linking members to build the lists. file: [16051a4db4] check-in: [8ba45681a0] user: stephan branch: trunk, size: 398245 | |
17:17 | Added per-recycling-bin hit/miss counters. file: [bb9141233c] check-in: [520ed62f5d] user: stephan branch: trunk, size: 396779 | |
16:33 | more metrics-related tinkering. file: [8538ca43ee] check-in: [624ffc5926] user: stephan branch: trunk, size: 396471 | |
15:11 | more work on the metrics, now recording value recycling hits/misses. file: [73811e51ae] check-in: [61ab49b29e] user: stephan branch: trunk, size: 395065 | |
14:13 | core: Value recycling bins are now allotted by the size of the Value, so like-sized Value types get grouped. i.e. more effective use of the recycler. However, this does not jive with the per-type-based configuration of bin sizes, but is an overall improvement with no losers, so we'll find a different configuration mechanism. file: [8d1428f372] check-in: [97b8ce3372] user: stephan branch: trunk, size: 393914 | |
2014-11-28
| ||
20:11 | more minor metrics tweaks. file: [f6c93deefe] check-in: [03503303ef] user: stephan branch: trunk, size: 389031 | |
19:30 | more tweaking related to the new recycler bits. file: [d162dffbcf] check-in: [06df42034a] user: stephan branch: trunk, size: 388672 | |
15:28 | core: improvements in the chunk recycler, and more memory gets recycled. file: [26be69ed4f] check-in: [a9a7aa2637] user: stephan branch: trunk, size: 387905 | |
01:07 | core: consolidated array/buffer memory recycling and made it usable for other purposes. Some types (hashes, z-strings) contribute to the recycling pool but do not yet draw from it. This cuts mallocs by just over 60 in the amalgamated unit tests, while costing minimal mangement memory (1-2k, allocated once). There is tweaking to be done and more stuff which can be recycled this way, but it's a promising start. Valgrind is happy. file: [3b9c035494] check-in: [94bd6ae976] user: stephan branch: trunk, size: 380348 | |
2014-11-25
| ||
16:29 | Minor but subtle semantics changes in cwal_function_call() and friends ("self" is now made vacuum-safe) and cwal_exception/propagating_set() (now uses unref instead of unhand when reseting/clearing the value). Accommodated that in s2/th1ish. Added cwal_utf8_char_next() utility function. In s2: started experimentation with the "new" keyword, but it is disabled and incomplete. Yeah, that is more than i wanted in this commit. file: [9b01d648ba] check-in: [bd8ecee444] user: stephan branch: trunk, size: 369629 | |
2014-11-24
| ||
21:32 | core: fixed a bug in waiting in the handling of vacuum-proofing functions for the duration of their call. Cleanups related to the previous commit. file: [3f3edec55b] check-in: [fd27a99b8a] user: stephan branch: trunk, size: 368838 | |
20:46 | core: minor internal tweaks in prep for some experimentation in s2 which will expose scope variable storage Objects to script code. cwal_function_call() now refs/unhands its result value (currently same effect as before, but to account for potential upcoming experimentation). file: [b693662825] check-in: [d9d4be0daa] user: stephan branch: trunk, size: 368788 | |
16:33 | minor doc touchups. file: [165ada42b8] check-in: [cace68b3b6] user: stephan branch: trunk, size: 364269 | |
2014-11-21
| ||
12:36 | hashtable metrics tweaks. file: [c7c2c58bca] check-in: [a6ccf5847d] user: stephan branch: trunk, size: 364073 | |
2014-11-20
| ||
20:19 | Added cwal_first_1000_primes(). file: [d3c63a084e] check-in: [1ff9355f0d] user: stephan branch: trunk, size: 363557 | |
19:43 | Added cwal_hash_resize() and s2.Hash.resize(). file: [f7b9550470] check-in: [75730700a6] user: stephan branch: trunk, size: 357237 | |
18:20 | trivial internal refactoring and docs. file: [890b1b54a4] check-in: [f31a261570] user: stephan branch: trunk, size: 354834 | |
18:13 | minor return-path reduction tweak. file: [1b899723f9] check-in: [c0051d3d47] user: stephan branch: trunk, size: 354582 | |
18:10 | const correctness tweaks. file: [4378fadf7a] check-in: [43b97ba497] user: stephan branch: trunk, size: 354513 | |
18:06 | core: refactored cwal_hash so that we can resize hashtables (APIs for that still pending). Negatively affects alloc counts (currently +1 per table) but the Value part of the hash is now recyclable and the hash table memory is currently reusable as array list memory. file: [61d50c2e3c] check-in: [310ba44db8] user: stephan branch: trunk, size: 354501 | |
2014-11-15
| ||
18:47 | core: 2nd half of the prop-search-while-visiting change. file: [9c367cc594] check-in: [1e84023da9] user: stephan branch: trunk, size: 353567 | |
18:10 | core: cwal_obase_search() now allows searching on a being-traversed container. Not sure why it was prohibited before. file: [101167de5f] check-in: [3402366053] user: stephan branch: trunk, size: 353536 | |
2014-11-14
| ||
18:13 | added cwal_rc_str2(). file: [313f454032] check-in: [1d4b45a91f] user: stephan branch: trunk, size: 353505 | |
2014-11-01
| ||
14:36 | minor internal cleanups. file: [dcc168c423] check-in: [6c80b0f90c] user: stephan branch: trunk, size: 353399 | |
2014-10-31
| ||
09:54 | clang warning cleanups. file: [0b572925e3] check-in: [e43eef1211] user: stephan branch: trunk, size: 353072 | |
2014-10-28
| ||
00:05 | minor internal cleanups and docs. file: [8b96214aa0] check-in: [bfd5737474] user: stephan branch: trunk, size: 353146 | |
2014-10-11
| ||
14:18 | inor portability fixes for building with CWAL_SIZE_T_BITS=16. file: [67c3731581] check-in: [8aead931ee] user: stephan branch: trunk, size: 352358 | |
11:43 | Tweaks to cwal_value_compare() when both values are numeric, so that (3<3.3) and (3!=3.3) are true even in an integer context. Changed a th1ish test broken by this. file: [0d8021325c] check-in: [915dfc2d84] user: stephan branch: trunk, size: 351934 | |
08:47 | added cwal_engine_adjust_client_mem() so that clients can account (in some minimal way) for memory via cwal metrics. s2 now uses this to report stack token and script function memory. file: [41d040824b] check-in: [8a484ca98c] user: stephan branch: trunk, size: 351623 | |
2014-08-31
| ||
10:56 | core: another string recycling size tweak, gives a tiny (under 1%) alloc reduction in s2 unit tests as well as saving about 600 bytes total mem. file: [a268893800] check-in: [e154af2b91] user: stephan branch: trunk, size: 350893 | |
2014-08-30
| ||
05:28 | cwal_function_call_array() now allows a NULL array. file: [71de732954] check-in: [ea21a5da3e] user: stephan branch: trunk, size: 350834 | |
2014-08-29
| ||
12:08 | minor tweak to max recyclable string size, gives slightly better results on 64-bit, tiny peak increase, marginal malloc reduction. file: [6541f690bb] check-in: [7c13a15c82] user: stephan branch: trunk, size: 350789 | |
2014-08-28
| ||
21:35 | minor doc and comment tweaks. file: [47f028194d] check-in: [9cc9a32d91] user: stephan branch: trunk, size: 350414 | |
2014-08-27
| ||
07:27 | minor tweak to yesterday's fix, honoring the initial list size minimum (helps the recycler out). file: [540cc3ae1d] check-in: [446b495cfc] user: stephan branch: trunk, size: 350399 | |
2014-08-26
| ||
20:54 | removed an assert added in the last round of bug-chasing. file: [f8b7ef03c0] check-in: [e9a5b4f51f] user: stephan branch: trunk, size: 350323 | |
20:38 | Fixed an inexplicable brain-burp in cwal_array_set() which triggered a bogus OOM error when setting an array index >= CwalConsts.InitialArrayLength on an empty array, a corner case which was only found via dumb luck after hours of debugging, trial, and error. file: [57966020f0] check-in: [20f7ea2175] user: stephan branch: trunk, size: 350366 | |
2014-08-25
| ||
16:42 | minor cleanup and semantics change (extension) in cwal_value_prototype_set(). file: [1c3c1d2ca4] check-in: [de3c61a486] user: stephan branch: trunk, size: 350443 | |
15:30 | Deprecated cwal_args_callee_state(), as it's a functional dupe of (yet less efficient than) cwal_args_state(). file: [0bd2c5f454] check-in: [26f947ec9d] user: stephan branch: trunk, size: 350380 | |
2014-08-24
| ||
14:40 | minor informational tweak in cwal_dump_allocation_metrics(). file: [1ed653a971] check-in: [1da45cb231] user: stephan branch: trunk, size: 350287 | |
2014-08-23
| ||
15:51 | core: consolidate integers and doubles into one recycling bin where there sizeofs are the same. Have to check it in to try it out on the 32-bit box (where they differ). file: [ee3e2dd059] check-in: [06a8a0df82] user: stephan branch: trunk, size: 350240 | |
2014-08-22
| ||
13:09 | typo file: [e2f9c93710] check-in: [5972de16f6] user: stephan branch: trunk, size: 349586 | |
12:43 | core: tinkering with the buffer recycling algo. file: [a8016a93c5] check-in: [3208b69997] user: stephan branch: trunk, size: 349586 | |
2014-08-20
| ||
21:14 | memory misuse fix for the buffer caching, caught by the libfossil bindings (and pinpointed by gcc's stack-smash protector). file: [a254961b7d] check-in: [9dba38f602] user: stephan branch: trunk, size: 349258 | |
18:08 | Minor touchups to the previous commit. file: [c4a6404327] check-in: [35a020bb78] user: stephan branch: trunk, size: 348954 | |
16:13 | core: found another optimization, recycling cwal_buffer memory (5 buffers' worth) similar to how cwal_array list memory is recycled. In the s2 unit tests this reduces the aggregate alloc count by 3.5-9%, depending on their buffer usage, with the _potential_ for a higher peak usage of (sizeof recycled buffers), but that depends on how the app uses buffers. file: [e1e0ad0d84] check-in: [1648e371e7] user: stephan branch: trunk, size: 348535 | |
2014-08-18
| ||
20:38 | minor internal tweak to cwal_array_shift(), doc touchups. file: [49ed8e7d46] check-in: [fcd9591170] user: stephan branch: trunk, size: 345164 | |
18:16 | core: swapped out string hash algo again. FNV-1a seems to give the best overall results so far, at least on 64-bit. file: [18f2473d5a] check-in: [b86d09b1cf] user: stephan branch: trunk, size: 345066 | |
2014-08-17
| ||
17:09 | typo file: [d5681d319f] check-in: [f6f070b0e0] user: stephan branch: trunk, size: 344764 | |
11:26 | experimental tweak to kvp-set which appears to make no functional difference. file: [2921a22a89] check-in: [3edf547f3a] user: stephan branch: trunk, size: 344767 | |
2014-08-16
| ||
11:33 | Added cwal_props_count() and cwal_props_has_any(), plus s2 Object.isEmpty() and propertyCount(). file: [313d005334] check-in: [c95efcf86c] user: stephan branch: trunk, size: 344243 | |
2014-08-15
| ||
13:16 | added cwal_buffer_fill_from_filename2(), patched s2_eval_filename() to use it (to avoid hypothetical corner cases involving non-NUL-terminated X-/Z-strings). file: [b1edcda697] check-in: [4c8879eff0] user: stephan branch: trunk, size: 343844 | |
2014-08-10
| ||
18:53 | Minor cleanups in error propagation in cwal_array_sort_stateful(). More work to do there and it seems to sort incorrectly at times. file: [7974a08be5] check-in: [2df8b807e0] user: stephan branch: trunk, size: 342923 | |
11:27 | doc updates. file: [3189eb49ca] check-in: [70cfc14407] user: stephan branch: trunk, size: 342477 | |
2014-08-06
| ||
17:10 | fixed return index in cwal_array_index_of() when value==NULL. file: [ce2f11972c] check-in: [ee436d7317] user: stephan branch: trunk, size: 342257 | |
2014-08-05
| ||
17:10 | fixed a missing clearing of cwal_native::rescope_children which (justifiably) triggered an assertion in s2 code when a native was rescoped but it could not find its bound client-side part. file: [3c3f22bff4] check-in: [85b1c7157e] user: stephan branch: trunk, size: 342217 | |
2014-07-22
| ||
17:57 | minor internal efficiency tweak. file: [6f9bb6e331] check-in: [f541cd4d5e] user: stephan branch: trunk, size: 342113 | |
2014-07-18
| ||
13:44 | added cwal_buffer_to_zstring_value(), due to popular demand in client code. file: [18ef8ccba2] check-in: [f4705fd709] user: stephan branch: trunk, size: 342069 | |
2014-07-16
| ||
22:22 | core: popping the top-most scope now properly disconnects cwal_engine::prototypes, and will re-created it as needed. file: [da526e97e9] check-in: [30fcf92b1d] user: stephan branch: trunk, size: 341931 | |
2014-07-09
| ||
15:48 | core: minor cosmetic metrics dump tweak. file: [b8ad7111fc] check-in: [ea1e56e3c6] user: stephan branch: trunk, size: 341369 | |
2014-07-07
| ||
21:22 | core: minor metrics dumps tweak. file: [d8ca967606] check-in: [aaf8a5df0d] user: stephan branch: trunk, size: 341324 | |
2014-07-06
| ||
15:53 | core: minor metrics dump tweaks. file: [90f02995f0] check-in: [c913c665d7] user: stephan branch: trunk, size: 341296 | |
2014-07-05
| ||
10:59 | core: cwal_array_take(): replaced an old ugly refcount abuse the the formal mechanism for such abuse. file: [1989f22645] check-in: [d82ad21d69] user: stephan branch: trunk, size: 341189 | |
2014-07-04
| ||
20:51 | core: a minor optimization in one of the kvp search routines. file: [a881595cd5] check-in: [a8897043c6] user: stephan branch: trunk, size: 341280 | |
15:13 | minor assertion improvement after triggering it via an s2 bug. file: [c5260ca4eb] check-in: [7407fd839c] user: stephan branch: trunk, size: 341196 | |
2014-07-02
| ||
22:24 | tweaked weak ptr table size again. file: [588d740dfe] check-in: [cb2ccce9d9] user: stephan branch: trunk, size: 341170 | |
18:06 | core: added missing x/z-string to cwal_engine_recycle_max(). Comsetic metrics output tweaks. file: [96b78297f3] check-in: [464babc10e] user: stephan branch: trunk, size: 341169 | |
17:48 | Added cwal_hash_size(). file: [072b32b92d] check-in: [3fd4fe9001] user: stephan branch: trunk, size: 340824 | |
2014-06-30
| ||
15:57 | Added cwal_prop_set_XXX_with_flags(), to support setting hidden properties from s2. Fixed a pointer misuse in cwal_value_may_iterate() which caused it to return unpredictable results. file: [0915092e70] check-in: [4434f6db64] user: stephan branch: trunk, size: 340744 | |
2014-06-29
| ||
14:14 | Tweaking of cwal_weak_ptr page sizes. Proactive range fix weak ref recycling. file: [b58f42db33] check-in: [c12e7f860a] user: stephan branch: trunk, size: 340170 | |
13:33 | Appear to have fixed a ptr_table-remove bug caught by an assertion in a mis-cleanup situation. file: [6c6234f30d] check-in: [b89eeb0831] user: stephan branch: trunk, size: 340141 | |
2014-06-28
| ||
15:32 | fixed a test broken by a recent change. Minor internal fix in cwal weak ref cleanup. file: [049fee7a40] check-in: [7f98a79b52] user: stephan branch: trunk, size: 340339 | |
13:47 | merged in the s2-semicolons branch. file: [81549c590b] check-in: [9af1212295] user: stephan branch: trunk, size: 340312 | |
12:31 | minor vacuum tweak. file: [af1df9ea21] check-in: [ed74bc8c9f] user: stephan branch: s2-semicolons, size: 340165 | |
10:59 | A fix for the two specially-propagating values in the face of a vacuum. file: [b7290d74df] check-in: [0362c958b3] user: stephan branch: s2-semicolons, size: 340063 | |
04:40 | Added cwal_json_output_opt::cyclesAsStrings, which causes cycles to be output as some debugging-only string representation, rather than causing an error. file: [c8601b3adf] check-in: [640531cd49] user: stephan branch: s2-semicolons, size: 339855 | |
02:20 | Fixed a major blunder in cwal_engine_vacuum(), which did not start gc-queuing mode when cleaning up the scope. 7 hours debugging that. Amazing this has not backfired until now, but it requires certain conditions before it trips up. file: [a18e5772bd] check-in: [e377a0525b] user: stephan branch: s2-semicolons, size: 339854 | |
2014-06-27
| ||
19:00 | Added cwal_kvp_flags_set(), for use in some upcoming s2 bits. file: [9c8226967a] check-in: [14d18baad5] user: stephan branch: trunk, size: 339496 | |
2014-06-26
| ||
19:56 | improved an assertion triggered by an invalid unref in s2 code. file: [74ec443e13] check-in: [c993c23530] user: stephan branch: trunk, size: 339349 | |
2014-06-24
| ||
22:16 | re-introduced cwal_propagating_set() and friends, as a place to store a single auto-propagating value (both th1ish and s2 implement this themselves, but that's no longer needed). file: [a4c13df62b] check-in: [6a720a9358] user: stephan branch: trunk, size: 339231 | |
19:55 | minor doc/comment touchups. file: [21d936edf7] check-in: [ae9fdce53a] user: stephan branch: trunk, size: 338954 | |
2014-06-23
| ||
21:46 | minor cleanups file: [259241b670] check-in: [219ac05917] user: stephan branch: trunk, size: 338851 | |
19:53 | minor z-string metrics tweak: now assumes (and counts) a NUL byte (only for metrics purposes). file: [a3cbce8d53] check-in: [df4c4776d1] user: stephan branch: trunk, size: 338749 | |
19:45 | cwal core: X- and Z-strings now participate in value recycling (in the same pool, b/c they have the same sizeof()). Only their "shells" recycle, not their string bytes. Fixed a char-should-be-int case which had not yet caused a problem. file: [b37d96e995] check-in: [ec0d73eba5] user: stephan branch: trunk, size: 338564 | |
2014-06-22
| ||
08:51 | Added cwal_function_call_array(). file: [5a8e4f5e0a] check-in: [519baf9738] user: stephan branch: trunk, size: 337598 | |
2014-06-21
| ||
21:19 | cwal_buffer_fill_from_FILE/filename() now try to pre-allocate the buffer based on the guessed size of the file (based on seek-probing), falling back to the default strategy. file: [d7803f063b] check-in: [5085bfa2c5] user: stephan branch: trunk, size: 337095 | |
20:05 | Added cwal_scope_search_kvp(). file: [ab46c4047d] check-in: [76860bb343] user: stephan branch: trunk, size: 336381 | |
14:46 | Removed a redundant cannot-happen block. file: [06c86180a7] check-in: [63a980a029] user: stephan branch: trunk, size: 335376 | |
2014-06-20
| ||
17:39 | Fixed string-request metrics miscounting when string interning is enabled. file: [0765dfbd4d] check-in: [33cbbb8fbc] user: stephan branch: trunk, size: 335508 | |
2014-06-19
| ||
13:51 | Changed how cwal_engine manages its exception handle - it now takes uses ref/unhand, instead of not referencing at all. file: [1c5f7c8ceb] check-in: [e801bfc408] user: stephan branch: trunk, size: 335404 | |
10:10 | bending around the s2 exception/error infrastructure a bit, in prep for throwing exceptions. file: [f476ec7c70] check-in: [d7399dd09b] user: stephan branch: trunk, size: 336098 | |
2014-06-18
| ||
21:35 | fixed a thinking error in that last commit. file: [5f6a5e3975] check-in: [d9bc552399] user: stephan branch: trunk, size: 336038 | |
21:12 | changed semantics of explicit bool/string comparison so that (false==""). Does not seem to break any existing code. file: [832a61d795] check-in: [a3ad2000cf] user: stephan branch: trunk, size: 336005 | |
2014-06-15
| ||
19:41 | Seem to have gotten sweep-up working how i want it in s2. file: [3e51ccfc07] check-in: [2239164d8f] user: stephan branch: trunk, size: 335902 | |
2014-06-14
| ||
14:52 | changed the cwal_value_compare() semantics for NULL and UNDEFINED. Surprisingly, this does not break th1ish. file: [d88500e40b] check-in: [cd2897a4e5] user: stephan branch: trunk, size: 335897 | |
2014-06-13
| ||
18:59 | added cwal_container_part() file: [6c2d9666ef] check-in: [d6def05713] user: stephan branch: trunk, size: 335400 | |
2014-06-11
| ||
20:27 | added cwal_value_unhand(). file: [9b85873797] check-in: [86085394ee] user: stephan branch: trunk, size: 335210 | |
2014-06-05
| ||
18:26 | minor internal cleanup. file: [6ce187830d] check-in: [99b9b8dd8e] user: stephan branch: trunk, size: 334954 | |
2014-06-03
| ||
22:32 | Got some basic PoC in place for s2. file: [cf41a6a7a7] check-in: [0e11ae0e16] user: stephan branch: trunk, size: 334947 | |
2014-05-16
| ||
21:22 | cwal_function_call_in_scope() now vaccuum- and sweep-proofs the function for the duration of the call, to avoid that a recursive cleanup potentially nukes an temp function called at its decl point. file: [9eced6b08a] check-in: [25ee81e2ed] user: stephan branch: trunk, size: 334834 | |
19:00 | Fixed broken cleanup ordering in Natives (which i recently swapped), as this breaks natives which use their own properties in their cleanup (namely linenoiseish's auto-save-at-finalize). file: [8657a3c712] check-in: [242ab76d5d] user: stephan branch: trunk, size: 332677 | |
2014-05-14
| ||
20:41 | A round of cleanups, tweaks, and API renamings. Improvements in the ++/-- operators. file: [d045851350] check-in: [6fca161343] user: stephan branch: trunk, size: 332303 | |
2014-05-12
| ||
21:00 | Minor semantics change for cwal_value_get_cstr() when value is a completely empty (no memory) buffer - now returns 0 instead of an empty string. file: [cb5b730024] check-in: [1db9511921] user: stephan branch: trunk, size: 329922 | |
2014-05-11
| ||
15:22 | Minor internal touchups. file: [26c77941a0] check-in: [fb42f4b960] user: stephan branch: trunk, size: 329937 | |
12:05 | minor cleanups in the vacuum code. file: [2ae9c9fc77] check-in: [d37e7de872] user: stephan branch: trunk, size: 329868 | |
2014-05-10
| ||
19:24 | A round of doc additions and minor API renaming. file: [cc4ba6877b] check-in: [d4a56b1fb6] user: stephan branch: trunk, size: 329845 | |
19:07 | Added getter/setter for cwal_engine client data. Found, quite by accident, the 'dangling native' which was breaking vacuuming in th1ish (cwal_engine::prototypes). Vacuum now works in th1ish's global scope :-D. Valgrind agrees. file: [96146116a6] check-in: [961ce26847] user: stephan branch: trunk, size: 329841 | |
17:05 | And now the lower-level bits for Buffer.resize(). file: [7f29563b17] check-in: [a698f5f7b6] user: stephan branch: trunk, size: 329221 | |
2014-05-09
| ||
18:24 | Minor cleanups. Increased StringPadSize to 8, as this gives us a big boost in efficiency on large scripts. Once again removed the 'phantom ref' on interned strings. Improved api.ob API a bit by returning api.ob from the void-returning methods. file: [77b52d12dc] check-in: [2c2187749f] user: stephan branch: trunk, size: 328569 | |
15:09 | reenabled an old assertion. file: [845b590af6] check-in: [05885cd8d9] user: stephan branch: trunk, size: 328187 | |
15:04 | minor internal tinkering. file: [28d6f2cbdf] check-in: [98dee75be2] user: stephan branch: trunk, size: 328193 | |
11:31 | Found the hashing (int range) culprit which broke interning lookups. Other tweaking with vacuuming and interning. Added a kludge ++refcount on interned strings upon insertion to save them from vacuuming. i don't like that, but it does save memory. Switched string hash algos again. file: [702f22aa50] check-in: [a6668115e9] user: stephan branch: trunk, size: 327928 | |
10:38 | More work on vacuuming and related code. Discovered in the process that string interning is not hitting anywhere nearly as often as i really expect it to. file: [79e49019bc] check-in: [a24cf62be2] user: stephan branch: trunk, size: 327734 | |
2014-05-08
| ||
23:41 | Refactored cwal_engine_vacuum() to... drum roll, please... no longer copy anything. It now only rescopes stuff, so no allocation needed, which means no error cases unless it detects memory corruption (assertion). file: [3c53218b22] check-in: [ae032f2d59] user: stephan branch: trunk, size: 328360 | |
23:13 | minor cleanups before larger changes. file: [5c6115eac4] check-in: [65535570a0] user: stephan branch: trunk, size: 329551 | |
23:04 | Started adding (untested, except that existing code/tests still work) infrastructure for making individual values, and those they reference, safe from vacuuming up. Intention: a way to make non-script-visible values safe from vacuum. file: [315f1a0050] check-in: [f5ae60cad0] user: stephan branch: trunk, size: 329580 | |
2014-05-07
| ||
19:29 | metrics output is now less cramped. file: [ff1f0abada] check-in: [2a5a4c7ff8] user: stephan branch: trunk, size: 326586 | |
19:01 | cwal_dump_allocation_metrics() now counts scopes differently (only allocated ones) because they skew the stats unduly (and unfairly ;). file: [44269054e7] check-in: [e4e693eb9f] user: stephan branch: trunk, size: 326574 | |
2014-05-06
| ||
20:54 | minor cleanups. removed some dead code. added some comments. file: [43cdf2badc] check-in: [571cccbd37] user: stephan branch: trunk, size: 326089 | |
2014-05-05
| ||
22:23 | minor tracing message tweak. file: [3f02ae6eba] check-in: [f97d4ed9e2] user: stephan branch: trunk, size: 325982 | |
19:38 | Vacuum now works as a drop-in replacement for sweep in th1ish :-D meaning that, in principle, it is (with proper script structuring) possible to create a leak-free, really-long-running scope, regardless of cycles. :-D! Interestingly, vacuum does not often catch more values than sweep, so it will be optimized to normally sweep, and replace every Nth sweep-up with a vacuum. file: [771bd0a749] check-in: [4ce80bab79] user: stephan branch: trunk, size: 325981 | |
18:06 | this variant of cwal_engine_vacuum() works for the first 3 th1ish unit test scripts, then falls apart somewhere, but is certainly better than the previous iteration. file: [07759d8973] check-in: [7a0a0b3e7a] user: stephan branch: trunk, size: 325350 | |
2014-05-04
| ||
12:36 | minor internal cleanups. file: [46787b7173] check-in: [4123cb68de] user: stephan branch: trunk, size: 325354 | |
11:07 | Added long-overdue cwal_buffer_clear() convenience function file: [192b593caf] check-in: [4dd397b5b4] user: stephan branch: trunk, size: 325140 | |
2014-05-03
| ||
23:51 | more tinkering with cwal_engine_vacuum(). file: [12ca405d45] check-in: [0cad6273e0] user: stephan branch: trunk, size: 325037 | |
20:07 | Added cwal_prop_flags_v(), admittedly a quick-hack measure for th1ish which needs to be improved upon at some point. file: [3c1c567878] check-in: [73df633530] user: stephan branch: trunk, size: 325058 | |
19:35 | Added cwal_prop_take_v(), though it has kinda funny semantics for the key part. Was necessary for postfix obj.prop++/-- operators. file: [b8c65e0614] check-in: [01628d5381] user: stephan branch: trunk, size: 324692 | |
16:28 | More tinkering with vacuum. It does horrible things to th1ish. file: [26a8da30c9] check-in: [93310a20c0] user: stephan branch: trunk, size: 323145 | |
15:35 | more work on vacuum. Seems it cannot be dropped right in to th1ish for reasons as yet determined but very possibly incompatibilities with vacuuming assumptions. file: [c17ae319f8] check-in: [e276ed0f1d] user: stephan branch: trunk, size: 324073 | |
14:33 | Introducing cwal_engine_vacuum(), an experimental algo which cleans up all non-variable values which are themselves not reachable via variable refs/containers. Fixed two bogus assertions while doing so. file: [8f72c11b15] check-in: [8f455e260b] user: stephan branch: trunk, size: 323324 | |
2014-05-02
| ||
20:55 | minor change to cwal_array_index_of() semantics. file: [15799f9582] check-in: [761fa1bdb2] user: stephan branch: trunk, size: 320417 | |
17:41 | minor tweaks to the cwal_native bindings. file: [52240b6680] check-in: [7411d5518f] user: stephan branch: trunk, size: 320314 | |
13:21 | Added cwal_buffer_to_zstring(). Minor string hash tinkering. file: [8b1a401aff] check-in: [88d8f96564] user: stephan branch: trunk, size: 320264 | |
09:38 | Added array list peak memory to the allocation metrics. Removed an old/arguable assertion which, for reasons completely inexplicable, is trigged only with this change. file: [b298b12c90] check-in: [bc61ad3eb4] user: stephan branch: trunk, size: 318934 | |
2014-05-01
| ||
16:49 | Switched to FNV-1a hash for string hashing. file: [43c1cb19c5] check-in: [e3fa91abaf] user: stephan branch: trunk, size: 318169 | |
16:33 | minor metrics-related tweaks. Now count scope_push() as a scope alloction request. file: [fce8315f74] check-in: [1d9683743e] user: stephan branch: trunk, size: 317492 | |
14:21 | Now tracks cwal_buffer_reserve()d memory via the metrics counter, recording only the peak capacity of any given buffer. file: [4d1d8b93a1] check-in: [7349e44714] user: stephan branch: trunk, size: 315799 | |
14:05 | minor internal cleanups. file: [67393b3b07] check-in: [b869fcd158] user: stephan branch: trunk, size: 315735 | |
2014-04-30
| ||
20:51 | improved the allocation metrics dump output a bit, adding percentages. file: [4fee6c1ae3] check-in: [cedeb7b5b1] user: stephan branch: trunk, size: 315666 | |
20:06 | Tweaked string interning table size. Does not affect memory usage of the th1ish unit tests in 64-bit mode and is now the same value on both 32/64-bit platforms. file: [f484f8851b] check-in: [880482bdb9] user: stephan branch: trunk, size: 314644 | |
2014-04-27
| ||
10:40 | Expanded the allocation metrics dump to include interned string and weak ref tables. file: [4ce4a8b950] check-in: [0b3097482a] user: stephan branch: trunk, size: 314539 | |
2014-04-26
| ||
20:23 | minor tweaks to allocation report output. file: [e1312d104b] check-in: [6309f987ed] user: stephan branch: trunk, size: 313580 | |
19:42 | improved allocation metrics a bit. file: [d906ff7e9f] check-in: [36fb4ce9a3] user: stephan branch: trunk, size: 313267 | |
2014-04-25
| ||
23:12 | added some internal comments. file: [4a4fc5519a] check-in: [bced5153b5] user: stephan branch: trunk, size: 310240 | |
19:57 | A fix for the Array.sort(Function) two-entry array case. file: [d64fb64a86] check-in: [c1024efd40] user: stephan branch: trunk, size: 310016 | |
2014-04-20
| ||
18:59 | okay, this branch has broken th1ish in ways i cannot explain and i am too tired to fight with it. Shelving this for the time being. file: [bf9759b72a] check-in: [aacb2e833c] user: stephan branch: new-string-length, size: 310850 | |
13:50 | Started re-doing the semantics for the length param to cwal_new_string() and friends, to something easier to use downstream. Breaks semantics in lots of th1ish code. file: [6f4a03c8d5] check-in: [b0dee8928f] user: stephan branch: new-string-length, size: 310492 | |
13:05 | A big around of minor tinkering and cleanups. file: [c8a31b4e06] check-in: [41d834342c] user: stephan branch: trunk, size: 309740 | |
2014-04-19
| ||
12:39 | added long-missing cwal_strlen(). file: [9b9546bd2e] check-in: [0222186bbf] user: stephan branch: trunk, size: 310460 | |
2014-04-17
| ||
19:09 | Added allocation metrics tracking cwal_engine. Added -metrics flag to the th1ish shell to dump them out. They do not tell 100% of the story but are interesting nonetheless. file: [c707fa49e1] check-in: [65cbcfd22a] user: stephan branch: trunk, size: 310364 | |
16:56 | Another round of minor cleanups and doc improvements. file: [69060c79b2] check-in: [66b65cb7ff] user: stephan branch: trunk, size: 308381 | |
2014-04-13
| ||
15:31 | minor internal cleanups. file: [76022b9234] check-in: [54972639a8] user: stephan branch: trunk, size: 306067 | |
2014-04-10
| ||
16:23 | Changed cwal_engine_init() semantics to require client to call cwal_engine_destroy() regardless of success or failure so errors info can be propagated. Another round of minor cleanups, docs, and commenting. file: [10fc08415d] check-in: [4f0da73e06] user: stephan branch: trunk, size: 305453 | |
2014-04-04
| ||
10:34 | Fixed signed/unsigned comparison warnings, which apparently do not show up without -Wsign-compare. file: [f581caf450] check-in: [c256618440] user: stephan branch: trunk, size: 304309 | |
2014-03-07
| ||
18:56 | Changed hashes again. file: [aa4d44568f] check-in: [e421f08ab1] user: stephan branch: trunk, size: 304294 | |
18:13 | Changed hash routine used for string interning. file: [77942908dd] check-in: [0801afff42] user: stephan branch: trunk, size: 304149 | |
16:59 | Removed some dead code. Added some more. file: [eee181adf4] check-in: [44bd21d624] user: stephan branch: trunk, size: 304149 | |
15:57 | Another tweak to the refcounting (or not) of interned strings. Moved determination of whether a given string is internalizable to a predicate function the cwal_engine_vtab interface. The default impl behaves as before. th1ish adds its own impl which basically limits interning to very short strings an arbitrary identifiers. file: [803e881182] check-in: [7588890cb3] user: stephan branch: trunk, size: 303907 | |
2014-03-04
| ||
13:51 | Added cwal_array_index_of() and Array.indexOf(). file: [e7c8482027] check-in: [54039517f2] user: stephan branch: trunk, size: 302967 | |
2014-03-02
| ||
10:16 | minor internal cleanups. file: [c406a1b616] check-in: [8d9f762d8f] user: stephan branch: trunk, size: 302459 | |
2014-02-28
| ||
20:41 | minor optimizations and tweaks. print() and api.io.output() now outputs Buffers as their content instead of a type@ptr string. file: [6d84083b81] check-in: [9cc51aba6e] user: stephan branch: trunk, size: 302455 | |
2014-02-27
| ||
21:22 | fixed an unused var for non-tracing builds. file: [2c1841ac7e] check-in: [5dfd7a164d] user: stephan branch: trunk, size: 302416 | |
20:40 | minor internal cleanups and one optimization which cut 5.thish's mallocs down by almost two order of magnitude :-D. Happy accident, actually, but the reason for it is clear - mallocs are down notably across the board for all test scripts. file: [a69000a4d3] check-in: [6321c72aba] user: stephan branch: trunk, size: 302305 | |
18:35 | Added cwal_callback_hook and friends, and refactored th1ish to inject its argv/this vars from there. This generically fixes the problem of this/argv/argv.callee resolving to one from a higher-up callback. Removed the workaround added for this yesterday. file: [5e7d0d356e] check-in: [4ea181bb5d] user: stephan branch: trunk, size: 302257 | |
2014-02-26
| ||
23:07 | fixed a slightly wrong assertion. file: [7f9f30f8e8] check-in: [db8d1e8273] user: stephan branch: trunk, size: 301486 | |
22:45 | moved th1ish_inject_callback_argv() into public th1ish API (will need it elsewhere), some const correctness fixes in the core API. file: [8b39fd2d1e] check-in: [f71379525b] user: stephan branch: trunk, size: 301465 | |
2014-02-25
| ||
18:10 | Fixed a problem in cwal_array_take() which caused it to not take built-in values. Doc update. file: [a0f0c3f794] check-in: [c59879843c] user: stephan branch: trunk, size: 301451 | |
15:58 | Added cwal_props_sort() and Object.sortProperties() binding. file: [35731ca641] check-in: [bc29f58ba4] user: stephan branch: trunk, size: 301119 | |
2014-02-24
| ||
22:20 | fixed an off-by-one in the cwal_array_copy_range() fix. file: [d428c81415] check-in: [41b5535a45] user: stephan branch: trunk, size: 299553 | |
20:59 | Fixed cwal_array_copy_range() (and thus Array.slice()) to work as advertized. Fixed a wrong assumption (assertion) in Array.shift() binding. file: [051e95dae0] check-in: [02490653cf] user: stephan branch: trunk, size: 299556 | |
20:00 | Added cwal_array_take() and changed cwal_object_take() to reprobate (instead of transfer reference ownership) if needed because that's the only way to provide the client with sane return mechanism. Added Array.pop() binding. file: [18ce6225e6] check-in: [d1ed6fc7da] user: stephan branch: trunk, size: 299447 | |
2014-02-19
| ||
18:31 | Removed an extraneous rescoping of a re-used interened strings. Added yet more comments to the interned strings notes. file: [1653582f70] check-in: [a98e648aaf] user: stephan branch: trunk, size: 298591 | |
2014-02-15
| ||
18:37 | cwal_new_zstring() now no longer clones empty strings. file: [b30716df72] check-in: [2a469fd15d] user: stephan branch: trunk, size: 298135 | |
2013-09-09
| ||
17:12 | minor doc update file: [29d0c0c18c] check-in: [db4ff0d9fb] user: stephan branch: trunk, size: 297952 | |
2013-09-03
| ||
09:08 | cwal_value_compare() for two buffers now uses memcmp() semantics. file: [dd11fe4cb7] check-in: [825c349517] user: stephan branch: trunk, size: 298182 | |
2013-08-07
| ||
22:20 | another round of warnings which refuse to show up in this tree. file: [44a4eeb7ba] check-in: [1f300807c9] user: stephan branch: trunk, size: 297346 | |
22:14 | Fixed compile warnings which for some reason do not show up when i compile in this tree, but do in another. Weird. file: [7e60a7e2b3] check-in: [591d051369] user: stephan branch: trunk, size: 297339 | |
2013-07-27
| ||
15:06 | cwal_value_clone2() now deeply clones Buffers. file: [b80e86a6ec] check-in: [e201868761] user: stephan branch: trunk, size: 297328 | |
2013-07-22
| ||
21:15 | fix an endless loop case in cwal_value_derives_from(). file: [1174fd7f75] check-in: [d9008c02a7] user: stephan branch: trunk, size: 296953 | |
2013-07-21
| ||
18:26 | Added Object.toJSONString(). Still missing overridable toJSON() support. Moved th1ish_value_XXX_part() to cwal_value_XXX_part(). Added String.applyFormat(). file: [6cd1479eb6] check-in: [77d1757a8b] user: stephan branch: trunk, size: 296782 | |
2013-07-20
| ||
14:36 | Optimized string recycling to re-use strings based on a first-close-fit algorithm (was exact-fit). This cuts mallocs in the th1ish test scripts anywhere from a few percept, up to 36%. file: [5b6d890ef3] check-in: [9ee156625d] user: stephan branch: trunk, size: 294345 | |
12:42 | Minor refactoring of cwal_value_prototype_get() and friends. file: [9efa1ce3ae] check-in: [8bf2f71a4e] user: stephan branch: trunk, size: 291992 | |
11:04 | cwal_value_get_cstr() now also treats buffers as strings. file: [2784b63e3d] check-in: [89039d20a7] user: stephan branch: trunk, size: 291272 | |
2013-07-19
| ||
11:41 | Added cwal_engine_type_name_proxy() and friends and tweaked th1ish to use it. file: [2615c349eb] check-in: [aa092439d4] user: stephan branch: trunk, size: 290956 | |
11:09 | More work on buffer formatting. cwal_prop_xxx() and friends now have a const-correct container parameter. file: [f486c1a1c5] check-in: [9557396db3] user: stephan branch: trunk, size: 290308 | |
2013-07-14
| ||
01:29 | Ported in some utility code from cson (namely cwal_prop_get_sub() and friends). file: [7fbb5b15f0] check-in: [ee67dca8e2] user: stephan branch: trunk, size: 290191 | |
2013-07-12
| ||
18:42 | Ported in basic JSON parsing functionality from libcson. Added th1ish api.json.parse(). file: [db41caa530] check-in: [5b1b80dda0] user: stephan branch: trunk, size: 287294 | |
2013-07-11
| ||
15:51 | Added very primitive recycling of the lists which back Array values. This cuts alloc count anywhere from about 1 to 7 percent in the current th1ish test scripts. file: [8b94e389a8] check-in: [491ad36c59] user: stephan branch: trunk, size: 286895 | |
2013-07-07
| ||
16:22 | Changed signature of cwal_buffer_fill(). file: [a43db64a24] check-in: [6bd4212d3f] user: stephan branch: trunk, size: 285550 | |
2013-07-05
| ||
23:23 | Much tinkering with the weak references support. Refcounted them and share all instances pointing to the same memory. Most of this is untested, though (no use case yet :/). Added APIs for (un)registering client-side memory for weak referencing. file: [dbfb68b9d4] check-in: [e0e531af6d] user: stephan branch: trunk, size: 285541 | |
18:50 | Fixed a potential orphaning corner case in cwal_array_shift(). file: [492a0a81d1] check-in: [34cd72d7e5] user: stephan branch: trunk, size: 283213 | |
17:43 | Truly pedantic cleanup. file: [f031657138] check-in: [191b50dc48] user: stephan branch: trunk, size: 282695 | |
16:29 | Added cwal_new_number(). file: [0bd40b3447] check-in: [270172fac8] user: stephan branch: trunk, size: 282641 | |
2013-07-04
| ||
18:30 | Experimenting with accessor interceptors, but this is not quite the way there. file: [b01641a96c] check-in: [5299a543a6] user: stephan branch: accessor-interceptors, size: 284323 | |
2013-07-03
| ||
19:30 | added experimental/broken cwal_engine_sweep_non_vars(). Does not work right but want to play with it later. file: [af063a804f] check-in: [cdec608b94] user: stephan branch: trunk, size: 282393 | |
2013-07-02
| ||
17:06 | doc/assertion additions. file: [a8af4350f9] check-in: [6963074e47] user: stephan branch: trunk, size: 280679 | |
16:38 | Implemented recycling of cwal_weak_ref. file: [0d6fef6a75] check-in: [51dff05fca] user: stephan branch: trunk, size: 279976 | |
15:14 | Tweaking the weak ref support. file: [7599b8c18b] check-in: [6c85df8cb2] user: stephan branch: trunk, size: 278478 | |
14:04 | Added weak value reference support. This allows, e.g., N bound natives with mutual relationships to safely figure out if the other endpoint has been destroyed. file: [8e26f22498] check-in: [fab8ad55e3] user: stephan branch: trunk, size: 276070 | |
2013-07-01
| ||
22:11 | Introducing z-strings (similar to x-strings except that cwal owns their memory). This cuts the cost of th1ish's slurpFile() function by half because it removes an extra copy. Increased cwal_buffer_fill_from() buffer to 4kb to reduce the number of reallocs(). file: [a909e307ad] check-in: [8c04f5388e] user: stephan branch: trunk, size: 270043 | |
17:31 | Added (several prototypes).compare(). Added shell_extend.c as an easy way for clients to install their own script-side functionality into the default shell app. file: [163aca8282] check-in: [91250213b2] user: stephan branch: trunk, size: 268882 | |
16:24 | Expanded the function/args-state APIs a bit. Used them to implement Function.sourceInfo() type-safely. file: [7e70bc005c] check-in: [5f7fe4b6bc] user: stephan branch: trunk, size: 268833 | |
15:37 | Revamped cwal_new_function() and friends to take a type ID tag for the state parameter, to safify some of the current casting, enable 'multiple-context' callbacks, and a couple other minor things i would like to try out. file: [524e1a8d18] check-in: [cb4072639e] user: stephan branch: trunk, size: 268391 | |
2013-06-30
| ||
16:05 | Minor code reorg. file: [2448913634] check-in: [aa478564b2] user: stephan branch: trunk, size: 267917 | |
15:45 | Scope property objects now get their prototype removed so that they do not resolve members of the prototype via variable resolution. file: [2c4c22e188] check-in: [56974219ac] user: stephan branch: trunk, size: 267876 | |
12:24 | Changed return type of cwal_scope_properties(). file: [e3add04dcf] check-in: [fcdf9c23b7] user: stephan branch: trunk, size: 267187 | |
2013-06-29
| ||
14:19 | Added cwal_outputer::flush() and cwal_output_flush(). file: [4f68b89dc9] check-in: [ece45b44c4] user: stephan branch: trunk, size: 267218 | |
11:34 | Added cwal_parse_argv_flags(). file: [331b45b592] check-in: [882dfc6d7c] user: stephan branch: trunk, size: 266789 | |
2013-06-28
| ||
23:48 | moved an assert. file: [c6374be9f1] check-in: [66c27e2e54] user: stephan branch: trunk, size: 263479 | |
2013-06-27
| ||
17:56 | Added cwal_prototype_base_set/get() and started rewiring th1ish to use it. file: [c3cd4972aa] check-in: [1771a58344] user: stephan branch: trunk, size: 263479 | |
2013-06-24
| ||
10:35 | Moved the generated cwal_list code into the main API so i can re-use it in th1ish. file: [8fe5988919] check-in: [fb3224f152] user: stephan branch: trunk, size: 260980 | |
2013-06-22
| ||
23:16 | Added cwal_native_clear(). file: [2e5f551fdc] check-in: [8333211602] user: stephan branch: trunk, size: 258552 | |
2013-06-21
| ||
20:31 | fixed c99ism. file: [1cdb40ae17] check-in: [7c4fbde061] user: stephan branch: trunk, size: 258166 | |
18:51 | insignificant internal cleanups. file: [8873e01061] check-in: [66a69588a5] user: stephan branch: trunk, size: 258155 | |
2013-06-20
| ||
16:21 | More of the container functions now fail with CWAL_RC_ACCESS when called during traversal of the container. file: [4383a1fad4] check-in: [801932bb85] user: stephan branch: trunk, size: 257979 | |
15:21 | const-declared scope vars now ignore an assignment-to-self to simplify some client-side code. file: [ea327d9d7a] check-in: [f4f9b57191] user: stephan branch: trunk, size: 257630 | |
2013-06-19
| ||
16:38 | Fixed an argument-handling-related memory corruption case caused by a well-intentioned optimization (which was just removed). This also fixes the bug that returning argv from a function lead to unexpected results (empty array). file: [2717721d93] check-in: [90aad2442f] user: stephan branch: trunk, size: 257062 | |
2013-06-18
| ||
15:38 | cwal_value_prototype_set() no longer allows (what it can recognize as) cycles between the child and its prototype. file: [e176ebf1a5] check-in: [69ab011467] user: stephan branch: trunk, size: 256933 | |
11:43 | More re-working of cwal_value_clone() and friends. file: [1d89ccbeb3] check-in: [9fd33ca240] user: stephan branch: trunk, size: 256677 | |
2013-06-17
| ||
20:27 | Minor internal defensive hack. file: [5ef7f5deda] check-in: [29aac3c628] user: stephan branch: trunk, size: 256422 | |
16:42 | Started work on adding an option to disable doubles in cwal. It basically just changes cwal_double_t to be cwal_int_t. file: [97fa6e14df] check-in: [264317163d] user: stephan branch: trunk, size: 256387 | |
2013-06-16
| ||
12:53 | Added cwal_hash_clear(). file: [a0b90b1ef6] check-in: [b1cc0ee1ce] user: stephan branch: trunk, size: 255632 | |
11:57 | Fixed: buffers now also evaluate to true in a boolean context. file: [22f6beb886] check-in: [00d2d3851b] user: stephan branch: trunk, size: 255142 | |
11:39 | Fixed: the newer container now evaluate to boolean true. file: [1bd80876c0] check-in: [278f121614] user: stephan branch: trunk, size: 255067 | |
06:18 | Added cwal_array_copy_range() (a.k.a. slice). file: [1fb046c06f] check-in: [071c1cef8d] user: stephan branch: trunk, size: 255257 | |
04:15 | Added cwal_array_shift(). file: [91ffa7176b] check-in: [8e4d80789d] user: stephan branch: trunk, size: 254359 | |
02:32 | Fixed an inconsistency of cwal_scope_properties() vs cwal_scope_init_props(). Minor doc tweaks. file: [d6a4ae3ba1] check-in: [63caeada6c] user: stephan branch: trunk, size: 252684 | |
2013-06-15
| ||
15:25 | minor cleanups file: [ae69872c9d] check-in: [6465552570] user: stephan branch: scope-hash, size: 255327 | |
15:18 | Switched cwal_scope::props from object to hash. Not sure i am willing to pay the memory cost, though. file: [88818a0d97] check-in: [931fbcca46] user: stephan branch: scope-hash, size: 255172 | |
12:41 | cwal_hash_insert/remove() now fail with CWAL_RC_ACCESS if called while iterating over the table. file: [8a07d2b48c] check-in: [9e0415bcd6] user: stephan branch: trunk, size: 253131 | |
11:58 | Fixed bugs in cwal_hash search which caused it to not communicate left-hand chain member, which lead to broken linking on hash collisions. file: [94b840447f] check-in: [7374119631] user: stephan branch: trunk, size: 252709 | |
2013-06-14
| ||
17:21 | added cwal_value_may_iterate(). file: [792dd4c884] check-in: [292db68d9a] user: stephan branch: trunk, size: 253041 | |
14:27 | And more tinkering with hashes. file: [0627687e82] check-in: [f7e6535b53] user: stephan branch: trunk, size: 252756 | |
13:35 | More work onthe hashtable. file: [d65ede6cf1] check-in: [77a8222358] user: stephan branch: trunk, size: 252617 | |
12:09 | More tinkering with the hash table. file: [a3889c1491] check-in: [fc25158582] user: stephan branch: trunk, size: 248687 | |
2013-06-13
| ||
23:54 | Added prototype/experimental hashtable support to th1ish. Had to uplift some static cwal funcs into cwal_internal.h to do it. Finally tested ability for Natives to rescope their children (and discovered the problems involved). Hashtables still triggering at least one outstanding re-parenting/lifetime-related assertion in some script constructs. file: [2a3ddfe6d7] check-in: [670a0ad9c5] user: stephan branch: trunk, size: 240828 | |
2013-06-12
| ||
16:04 | Added cwal_array_sort_stateful() and friends. file: [66af6ebdb8] check-in: [3927173367] user: stephan branch: trunk, size: 241242 | |
2013-06-11
| ||
12:38 | Sometimes i hate printf. 32/64-bit compatibility fix. file: [60cb0d0e31] check-in: [7d9b4f2ee4] user: stephan branch: trunk, size: 237389 | |
2013-06-10
| ||
21:24 | Added integer -1 and double -1.0 to the list of built-in constant values because they appear so often in script code. file: [45e7d2e146] check-in: [54989ac4fa] user: stephan branch: trunk, size: 237396 | |
20:46 | Removed some old comments and a stray debug message. Tweaked size of string interning table size again. Cleaned up the main test app a bit. file: [eb6a4baf0b] check-in: [0c0a865f24] user: stephan branch: trunk, size: 237004 | |
20:03 | Fixed inconsistent orderings in cross-type value comparisons. Fixed a huge bug in the new cwal_ptr_table linked list which caused several megabytes of tables to be allocated (one per string). file: [85e51f5047] check-in: [f4773da793] user: stephan branch: trunk, size: 237265 | |
17:25 | Fixed a bug in the ptr_table remove-entry op which traversed tables unnecessarily. Renamed cwal_interned_search_v() to cwal_interned_search_val() file: [8ec4e2df6d] check-in: [11ea684831] user: stephan branch: trunk, size: 236296 | |
17:14 | Refactored cwal_ptr_page to a linked list and replaced cwal_ptr_table::pages (type=cwal_list) with that linked list. Saves a small tick of memory when interning strings, removes several potential malloc error cases (on OOM), and has the same performance characteristics as before. file: [46bebcdf9a] check-in: [823220ffed] user: stephan branch: trunk, size: 236326 | |
15:24 | Added cwal_array_reverse(). cwal_array_sort() now fails if called while the array is being visited/traversed. Doc additions. file: [03190aa81f] check-in: [8079c5b03d] user: stephan branch: trunk, size: 238197 | |
2013-06-09
| ||
19:48 | th1ish: minor tinkering. Removed the limit that 'importSymbols' only works for the calling scope (too onerous given th1ish's tendency to create scopes whenever it can get away with it, and a function call's scope starts while the parameters are being processed, as a memory-saving measure). file: [66f6d10df5] check-in: [32806d9602] user: stephan branch: trunk, size: 237778 | |
17:45 | Fixed: prototype values are now re-scoped when child properties are re-scoped, and are unref'd using cwal_unref_from() instead of cwal_value_unref() (so they get re-probated if needed). Fixes a problem in th1ish with returning a new prototype value from a sub-scope. file: [b5b327911f] check-in: [3f002b6bc3] user: stephan branch: trunk, size: 237581 | |
13:00 | Added cwal_engine_recycle_max_get(). file: [f7093e5d1f] check-in: [1c28677e5b] user: stephan branch: trunk, size: 237140 | |
11:24 | Fixed (it seems) cwal_value_clone(). file: [f63ca5a94c] check-in: [c430cfad6c] user: stephan branch: trunk, size: 236779 | |
08:54 | th1ish: minor overhauling of the internal exception-throwing bits. file: [652a663041] check-in: [a7bf69df61] user: stephan branch: trunk, size: 234977 | |
03:46 | Documented an unsightly kludge. file: [b78836c57c] check-in: [47f9c9b09e] user: stephan branch: trunk, size: 234702 | |
2013-06-08
| ||
04:31 | Changed an assertion to a proper return value in cwal_prop_set(), and removed its requirement that its key be non-empty (for consistency). file: [b9896b611b] check-in: [9083d938ec] user: stephan branch: pcre, size: 233994 | |
01:33 | Seem to have found a decent heuristic for working around the obj.prop-gets-upscoped-and-cleaned problem by moving values _back_ into probationary state _if_ they would be cleaned up by a container which lives in a lower scope than the value. Allows removal of the 'extra ref' th1ish had to add before this. file: [74f7323454] check-in: [8f9e77f3ea] user: stephan branch: pcre, size: 233965 | |
2013-06-06
| ||
19:36 | Some rough (completely garbage) cycle-counting code i may want to revisit someday. file: [68e323765f] check-in: [429a1a3dad] user: stephan branch: cycle-counting-revisited, size: 233148 | |
17:28 | Removed an ambiguity in the handling of empty string keys in cwal_var_set() vs cwal_var_set_s() and cwal_var_set_v(). file: [ff3b06491f] check-in: [957ae67c59] user: stephan branch: trunk, size: 229969 | |
2013-06-04
| ||
19:55 | cwal_prototype_set() now ensures that the prototype's scope is at least as old as the child's. file: [84a2cb4d70] check-in: [6cc44d3e3d] user: stephan branch: th1ish_script, size: 229988 | |
18:22 | Added a missing return value check in cwal_array_length_set(). file: [419e1b631e] check-in: [b07032081c] user: stephan branch: th1ish_script, size: 229933 | |
17:54 | Removed some debuggering output. file: [5a45d3144c] check-in: [8711c79b82] user: stephan branch: th1ish_script, size: 229898 | |
16:48 | moved a #define. file: [e9b95c83ae] check-in: [ef558ed5c8] user: stephan branch: th1ish_script, size: 229890 | |
15:24 | Fixed a property lookup bug caused by recent sorting changes. file: [3728170eed] check-in: [ff0e48671c] user: stephan branch: th1ish_script, size: 229858 | |
2013-06-03
| ||
22:58 | Minor internal tinkering of the buffer size in cwal_buffer_fill_from(). file: [747367960c] check-in: [14158c119e] user: stephan branch: th1ish_script, size: 229823 | |
2013-06-02
| ||
17:58 | Added cwal_array_sort(). file: [79e6acc98c] check-in: [d114a07e93] user: stephan branch: trunk, size: 229523 | |
14:23 | Fixed switch-fall-through bug in the new string-to-number code. file: [b7ec830309] check-in: [f4a355ef3a] user: stephan branch: trunk, size: 228800 | |
13:45 | added cwal_cstr_to_int/double(). Documented CWAL_INT_T_MIN/MAX. file: [439d7010d3] check-in: [4b8d1e0a45] user: stephan branch: trunk, size: 228772 | |
11:32 | Added cwal_props_copy() and more docs. file: [8da9cf065f] check-in: [8d5d4a5040] user: stephan branch: trunk, size: 226088 | |
2013-06-01
| ||
15:43 | Fixed a potential buffer overflow in the previous cwal_int_to_cstr() fix. file: [a7e7ca187e] check-in: [dfac0b681b] user: stephan branch: trunk, size: 225520 | |
15:13 | Fixed a bug in cwal_int_to_cstr() - corner case involving INT_MIN. Added CWAL_INT_T_MIN/MAX macros. file: [069650a66b] check-in: [29e01f5d63] user: stephan branch: trunk, size: 225476 | |
06:09 | Added new json formating option (addSpaceAfterComma). Added experimental code to insert object properties in sorted order. Searching is not optimized based on it, though. file: [5b9b7dd8fe] check-in: [76df00ac93] user: stephan branch: trunk, size: 224928 | |
2013-05-31
| ||
23:56 | removed an old ugly hack. file: [62929d1798] check-in: [15506c1773] user: stephan branch: trunk, size: 223350 | |
21:50 | Extended cwal_scope_push() to accept an externally-allocated scope instance (malloc optimization). file: [f65918c43c] check-in: [9c1398a9c1] user: stephan branch: trunk, size: 223344 | |
20:06 | Changed how cwal_scope::mine::r0 is cleaned up. cwal_exception_set() no longer ref/unref's the exception (causes problems downstream). Forced a single refcount increase on shared interned strings the first time they are re-used to work around a lifetime problem. It's not a 100% solution but it's close. file: [86828d9f01] check-in: [63ade6e28a] user: stephan branch: trunk, size: 223213 | |
17:09 | Changes signatures of cwal_new_function() and friends. Started adding toString support to th1ish.. file: [2cdaa3fa82] check-in: [0ac710027a] user: stephan branch: trunk, size: 222356 | |
15:57 | Simplified the func_call() signatures a bit. file: [1e54339c72] check-in: [dc614cb0ac] user: stephan branch: trunk, size: 222268 | |
14:56 | Fixed swapped undef/null value definitions (broken in last round of refctoring). file: [6ab66e7414] check-in: [8f0b505002] user: stephan branch: trunk, size: 222123 | |
14:53 | factored out cwal_value::value ptr, saving sizeof(void ptr) bytes per value instance. This has broken something in th1ish, though. file: [09105e3d13] check-in: [e7cf5db3b5] user: stephan branch: no-value, size: 222123 | |
12:32 | tinkinger, trying to find the cause of the errant recycled exception... file: [987a50663c] check-in: [ffecd3a6f0] user: stephan branch: trunk, size: 223752 | |
2013-05-30
| ||
12:38 | Added an assertion to help catch a bug triggered by certain piece of th1ish code. file: [4962e66e5e] check-in: [39ab96b342] user: stephan branch: trunk, size: 223679 | |
2013-05-29
| ||
21:57 | Minor refactoring for... Added an amalgamation build to the makefile. file: [58b60f86ef] check-in: [f0740b8f7d] user: stephan branch: trunk, size: 223386 | |
20:54 | Fixed cwal_value_prototype_set() to return CWAL_RC_MISUSE if the given prototype is-not-a obase. file: [fc22d9c4d7] check-in: [b4b8e2115d] user: stephan branch: trunk, size: 223471 | |
20:06 | Fixed a serious referencing error in cwal_prototype_set(). Make cwal_compare_cstr() NULL-safe. file: [96b1cba1db] check-in: [1049202a82] user: stephan branch: trunk, size: 223448 | |
17:57 | started adding facilities to allow cwal_native types to rescope non-property children. Simplified cwal_value_rescope() signature. file: [2aa3a4108a] check-in: [c2ab18c056] user: stephan branch: trunk, size: 223377 | |
16:29 | potential fix for a slight potential va_start/va_end misuse. file: [6909a02da1] check-in: [559635875a] user: stephan branch: trunk, size: 222674 | |
14:46 | more x-string-related fixes and some prep for z-strings (like x-strings, but the cwal_string owns and deallocates the (char*)). file: [f1710e2a14] check-in: [0edafb49ef] user: stephan branch: trunk, size: 222668 | |
13:23 | fixes related to x-string handling. file: [b0bd3dcb10] check-in: [93debd3607] user: stephan branch: trunk, size: 221484 | |
2013-05-28
| ||
19:10 | comment/doc changes only. file: [113e142f96] check-in: [8d31254298] user: stephan branch: trunk, size: 219984 | |
19:04 | Added initial support for x-strings (external strings, not allocated by cwal). file: [97fa374c93] check-in: [76b841bfbd] user: stephan branch: trunk, size: 219950 | |
2013-05-25
| ||
17:16 | minor pedantic corner-case fix for a not-really-a-problem. file: [d7dd88f5b4] check-in: [d0d89d47b2] user: stephan branch: trunk, size: 216971 | |
14:36 | added some (unused) code which keeps insertion order (instead of reverse) for object properties. file: [1f682671dd] check-in: [3c4c0c4ddc] user: stephan branch: trunk, size: 216966 | |
12:04 | Fixed cwal_array_set() to accept NULL values. file: [54acfb21ea] check-in: [1b3c313c6b] user: stephan branch: trunk, size: 216579 | |
02:46 | added a TODO note. file: [1e29f3c1e4] check-in: [a5107abac3] user: stephan branch: trunk, size: 216524 | |
01:00 | changed cwal_kvp_key() to return a cwal_value instead of cwal_string (old behaviour). file: [66d31bb365] check-in: [7f67301752] user: stephan branch: trunk, size: 216413 | |
2013-05-24
| ||
21:15 | experimenting with object/array access. file: [fd3f9804fa] check-in: [ad33b01201] user: stephan branch: trunk, size: 216437 | |
18:12 | cwal_new_string() and friends now no longer ref() a string which came from the internalizer. It _seems_ we don't really need to. file: [275490bcd2] check-in: [f67d411781] user: stephan branch: trunk, size: 215634 | |
15:09 | #if'd out a var which is only used in big-void-ptr mode. file: [932482cd56] check-in: [074279f425] user: stephan branch: trunk, size: 214436 | |
15:02 | pulled in [refcount0] branch, which changes how refcounting working (starts from 0 instead of 1, with 0 having a special "probationary" status). file: [b9432a6579] check-in: [f518911f69] user: stephan branch: trunk, size: 214408 | |
14:59 | internal doc changes - nothing functional. file: [e698cede28] check-in: [a457e7fb69] user: stephan branch: refcount0, size: 214407 | |
14:56 | Disabled the special value 1 when the big-void-ptr optimization is on because i cannot seem to get dereferencing of the int pointer working for that case. file: [1e566ff58c] check-in: [6987fd404a] user: stephan branch: refcount0, size: 214336 | |
11:58 | Weird - i made these changes in [bae0623b6c] and now they are gone. (Edit: DUH - wrong branch, Einstein!) file: [dcfb4860c7] check-in: [bd0c6073a0] user: stephan branch: trunk, size: 208122 | |
2013-05-23
| ||
13:03 | minor efficiency hack in cwal_new_integer(). file: [f1021ecef4] check-in: [d999a437fe] user: stephan branch: refcount0, size: 213825 | |
12:58 | Added integer 1 and double 1.0 to the list of built-in values. file: [ea1591945e] check-in: [925e2bc847] user: stephan branch: refcount0, size: 213779 | |
12:22 | started adding searchPrototype option to functions which can use it. file: [a5e77eecef] check-in: [6e23ecc726] user: stephan branch: refcount0, size: 213161 | |
2013-05-22
| ||
19:28 | Fixed an old null pointer deref bug in the cwal_value compare impl for INT/DBL when the LHS was the special value 0 (or 0.0). file: [cd2a077e6d] check-in: [bae0623b6c] user: stephan branch: refcount0, size: 212746 | |
18:09 | removed some debug code, doc updates. file: [bdb4ea0e48] check-in: [f036d8a611] user: stephan branch: refcount0, size: 212151 | |
17:47 | Tweaking of interned string refcount mechanics, but just cannot seem to get what i need here. Might need to special-case string destruction for this. file: [c3e0eaecec] check-in: [2cbe4802e5] user: stephan branch: refcount0, size: 211788 | |
17:11 | Added cwal_scope::mine::r0 list to keep extra track refcount==0 values. sweep-scope is now much faster and safer. file: [cbf72050c4] check-in: [ef4c8f9e67] user: stephan branch: refcount0, size: 210735 | |
2013-05-21
| ||
22:53 | minor tinkering before bed. Seems to work so far. file: [d03de23e08] check-in: [6be90896fc] user: stephan branch: refcount0, size: 210022 | |
22:30 | changed value refcount to start at 0, meaning no my ref held by the scope. This _might_ give me semantics off of which i can vastly improve the memory usage in th1ish, by proving a "purgatory" state where a value has not yet gotten an explicit reference. Previously we had start refcount==1, with the 1 being justified as the holding scope, but that causes us grief in script implementations (orphaned values held too long) and an adds ambiguity between "purgatory" and "refcount was _reduced_ to 1". file: [4ea2ceb93e] check-in: [7ca566b740] user: stephan branch: refcount0, size: 210003 | |
2013-05-20
| ||
05:58 | Changed string tail-chopping algo to chop off the oldest 25%. Removed some dead comments. Extended cwal_recycler_get() to support the string recycler. file: [6166a61594] check-in: [88ba6547de] user: stephan branch: trunk, size: 208050 | |
04:14 | debug message changes only. file: [382fa8f3f4] check-in: [97e65d9d22] user: stephan branch: trunk, size: 206253 | |
04:06 | Added basic string value recycling. file: [3ae68ac328] check-in: [32a78562b9] user: stephan branch: trunk, size: 206229 | |
00:15 | Added the long-missing cwal_array_length_set(). file: [4e1b74e5c6] check-in: [cde01729a4] user: stephan branch: trunk, size: 202631 | |
2013-05-19
| ||
21:21 | Missing file from previous commit. file: [b4677a1fcb] check-in: [47140c2f78] user: stephan branch: trunk, size: 202094 | |
19:33 | Some disabled code saved for future consideration. file: [9f97ecb1c5] check-in: [03a2e937be] user: stephan branch: trunk, size: 201702 | |
18:22 | minor internal tweaking. file: [1811f4e6b0] check-in: [cb592daeae] user: stephan branch: trunk, size: 201283 | |
15:13 | Changed how cwal_function_call() manages the lifetime of the result value. Removed cwal_result_xxx() family of APIs. file: [4cad3185a8] check-in: [d7b822f53b] user: stephan branch: trunk, size: 201161 | |
14:28 | Added a new cwal_scope_prop_XXX() API to supplement or replace the cwal_var...() bits. Changed how Function call()s return their value, to make cwal_result_{s,g}et() unnecessary and handle result value lifetime better from the client. file: [74429ffaf1] check-in: [0e89645697] user: stephan branch: trunk, size: 203956 | |
2013-05-18
| ||
12:10 | Added miss types to cwal_value_type_id_name(). file: [8fe84623ca] check-in: [f8f041d801] user: stephan branch: trunk, size: 200058 | |
11:29 | added cwal_value_type_id_name() plus some notes about value interning. file: [6f72485be4] check-in: [325709de43] user: stephan branch: trunk, size: 199873 | |
04:03 | Simplified cwal_unref_string() signature. file: [87e95dc7b7] check-in: [8d3223b9b2] user: stephan branch: trunk, size: 198762 | |
2013-05-17
| ||
17:44 | started, the commented out, beginnings of cwal_output_buffer(). file: [279edad41f] check-in: [82c4d5df43] user: stephan branch: trunk, size: 198702 | |
2013-05-15
| ||
19:49 | more tinkering with th1ish. Can now evaluate basic math. file: [988d46bd81] check-in: [2c33797d53] user: stephan branch: trunk, size: 198153 | |
2013-05-11
| ||
03:57 | too tired to diff. file: [791db89377] check-in: [fdaaf81724] user: stephan branch: trunk, size: 198441 | |
02:16 | fixed var-set(NULL) return value. Changed semantics of sweep() return value. Other minor tweaking. file: [b218e19a85] check-in: [1eab8d6ab8] user: stephan branch: trunk, size: 198351 | |
2013-05-10
| ||
21:18 | Fixed a buffer appending problem. Added CWAL_RC_ASSERT. file: [3f3d55aa9f] check-in: [6782d563eb] user: stephan branch: trunk, size: 195729 | |
20:02 | Added string concat ops. file: [6994c2ac80] check-in: [ae73cb0299] user: stephan branch: trunk, size: 195674 | |
17:15 | more code consolidation. Changed some internal hash values. file: [e9dfee4de4] check-in: [a8307e7444] user: stephan branch: trunk, size: 194028 | |
16:31 | Consolidated some code. Added "catch" support to s1. file: [cf62a124a9] check-in: [7ee91d6574] user: stephan branch: trunk, size: 195737 | |
15:01 | fixes in the exception bubbling. file: [56ac859c63] check-in: [a324911a06] user: stephan branch: trunk, size: 193570 | |
10:52 | Added cwal_var_get_v(). Updated s1 to compile against recent changes. file: [f7a1090df2] check-in: [bcdc8f9f8f] user: stephan branch: trunk, size: 193215 | |
09:48 | Simplified more of the public API by removing more extraneous cwal_engine parameters. file: [6d26c13d6c] check-in: [17395f36ae] user: stephan branch: trunk, size: 191973 | |
09:02 | more tinkering. Added a missing s1 test script. file: [019a8f51f6] check-in: [3386bd8e93] user: stephan branch: trunk, size: 190933 | |
2013-05-09
| ||
14:47 | more minor tinkering. file: [e21355af13] check-in: [6522247bd2] user: stephan branch: basic-prototypes, size: 191472 | |
10:17 | minor internal tweaking and refactoring. file: [11aaa6927a] check-in: [4ba195c6b8] user: stephan branch: basic-prototypes, size: 190151 | |
09:39 | more object parameter simplifications file: [c528f64153] check-in: [57af8b5b6a] user: stephan branch: basic-prototypes, size: 189703 | |
03:30 | more removal of extraneous cwal_engine parameters. file: [0bafe777a5] check-in: [977eb9cd6e] user: stephan branch: basic-prototypes, size: 189011 | |
03:08 | Fixed the unset leak. Was an old, old bug which i am surprised i have no triggered yet. file: [f0718cfda6] check-in: [6fc95ba5ab] user: stephan branch: basic-prototypes, size: 189084 | |
02:58 | more work on values-as-keys. Found unset-prop cases for non-strings which cause leaks, though. file: [4ca333b5ce] check-in: [51fddf67c7] user: stephan branch: basic-prototypes, size: 189085 | |
01:29 | obase cleanup now unrefs the object's prototype. file: [21abf7b64b] check-in: [4d13c51957] user: stephan branch: basic-prototypes, size: 187522 | |
01:14 | more work on getting non-string key infrastructure in place. file: [52f5fae1b5] check-in: [630cba347c] user: stephan branch: basic-prototypes, size: 187199 | |
00:02 | Started adding experimental "prototypes" support (similar to JavaScript). More work on making key/value pairs able to use non-string keys. file: [7de39c7234] check-in: [743b712cda] user: stephan branch: basic-prototypes, size: 185493 | |
2012-08-20
| ||
22:28 | Added cwal_value_clear_mutable_state(). file: [dbb328644c] check-in: [01481544c5] user: stephan branch: trunk, size: 182207 | |
20:47 | Fixed a bug in kvp recycling (counter was wrong). Added scope recycling, which shaves a significant number of mallocs off of my test code and in s1. file: [4e6955a791] check-in: [24a0e9a8f2] user: stephan branch: trunk, size: 181031 | |
17:28 | Added an id field to cwal_recycler to tag the type associated with the recycler, and some additional sanity-checking to use it. file: [1a36426ebc] check-in: [d06acf53ad] user: stephan branch: trunk, size: 175151 | |
17:16 | moved the main rescope-children operation into cwal_value_vtab to make it easier to re-use that code for new classes without touching the core for it. file: [3071ad0968] check-in: [47e431d767] user: stephan branch: trunk, size: 174123 | |
2012-08-19
| ||
16:14 | A number of fixes to the interpreter and incidental tweaks to the core to support them. s1 scope unit tests fail but scoping from a script file works fine, so i will chase that later. file: [49b44dad10] check-in: [f2a666d73e] user: stephan branch: trunk, size: 173999 | |
14:14 | Function values now compare equal if they have the same callback and state data pointers. file: [c7e2b65245] check-in: [cef477d430] user: stephan branch: trunk, size: 173949 | |
05:11 | Changed cwal_exception_setXXX() to treat a NULL format string as an empty one. file: [7bc5109e19] check-in: [ba78c2784e] user: stephan branch: trunk, size: 173178 | |
03:23 | Tweaked s1 test code to read input from a file. Seem to have gotten the func decl-vs-call ambiguities beaten (finally!). Added var/const decl support. file: [f796f241a7] check-in: [558d586769] user: stephan branch: trunk, size: 173177 | |
2012-08-18
| ||
22:46 | removed a dead line. file: [cae776201f] check-in: [2f2a687550] user: stephan branch: trunk, size: 172052 | |
05:07 | Got execution of custom script-side functions working, but still missing keywords to allow one to declare such a thing. Minor API renamings. file: [312acc0b2c] check-in: [e744cdc31e] user: stephan branch: trunk, size: 172129 | |
01:20 | got first scripted function call working. file: [61b9437bb2] check-in: [6eec35f8e9] user: stephan branch: trunk, size: 171718 | |
2012-08-17
| ||
23:23 | Fixed var_set behaviour to set the var in the scope it is found in. Need to optimize this, though (currently 2 lookups). file: [bd918596a9] check-in: [baa62c845c] user: stephan branch: trunk, size: 171685 | |
20:45 | more touch-ups and testing of the exception handling. file: [125de84335] check-in: [668442941e] user: stephan branch: trunk, size: 170676 | |
19:36 | renamed cwal_error to cwal_exception. Fixed lifetime/ownership problems in exception message handling by moving into a normal property. file: [e6c6cf31a0] check-in: [6d6374aa15] user: stephan branch: trunk, size: 169289 | |
15:48 | Internal optimization to cwal_value_is_builtin() which shaves a percent or so of the runtime from my test app. file: [a76d66ce66] check-in: [a54b88187c] user: stephan branch: trunk, size: 161089 | |
15:27 | Added missing cwal_value_get_native() impl. file: [1ac1b7c35e] check-in: [b73394c788] user: stephan branch: trunk, size: 160162 | |
15:13 | Some cleanups to the property consolidation. Removed some duplicate code. Added cwal_value_can_have_properties() (but do not like the name much!). file: [8e6a217e73] check-in: [9cd5ecdaba] user: stephan branch: trunk, size: 160008 | |
14:48 | oops - fixed memleaks caused by Function/Native/Array cleanup routines which did not honor their (new) property state. file: [78844d7a1a] check-in: [6f3110ea33] user: stephan branch: trunk, size: 160345 | |
14:34 | Refactored the key/value pair support to work with any cwal_obase type (object, array, function, native). Functions and Natives may now contribute to cycles. file: [616f871090] check-in: [c111db9842] user: stephan branch: trunk, size: 160119 | |
2012-08-15
| ||
14:22 | Finally pulled in the good old printf() implementation. Added cwal_outputf(), cwal_new_stringf(), cwal_buffer_printf(), and a few others. file: [c5ba82dde7] check-in: [07c25cf985] user: stephan branch: trunk, size: 157240 | |
03:32 | too many changes to remember. Started work on ternary if, but am having token-skipping-without-executing related problems. file: [c4d62652e0] check-in: [92c2a21a66] user: stephan branch: trunk, size: 153959 | |
2012-08-14
| ||
21:49 | more work on the kvp-related internal refactoring (to consolidate key/value pairs support for higher-level types). file: [68ddd22f56] check-in: [15a61ebbf2] user: stephan branch: trunk, size: 153906 | |
2012-08-12
| ||
11:14 | Removed braindeaded auto-set-from-parent in cwal_var_set(). file: [da2c4c3527] check-in: [37ea7e0876] user: stephan branch: trunk, size: 150902 | |
2012-08-11
| ||
22:17 | cwal_value_is_number() now also returns true for booleans. file: [26ea2e861d] check-in: [98559140e8] user: stephan branch: trunk, size: 150740 | |
19:57 | Got first assignment working in interpreter. file: [489e61bf26] check-in: [35c7283fc3] user: stephan branch: trunk, size: 150569 | |
18:07 | Added cwal_array_clear(). Tinkering with the interpreter. file: [9853b1652c] check-in: [d7a2ab0fa8] user: stephan branch: trunk, size: 150533 | |
15:21 | More work on the script interpreter. file: [785173c061] check-in: [ed917d6e49] user: stephan branch: trunk, size: 150186 | |
13:47 | Fixed a bug which causes cwal_value_undefined() to not be recognized as a builtin value. file: [ede073d2cf] check-in: [d46319a52a] user: stephan branch: trunk, size: 149888 | |
12:45 | Removed cwal_engine::resultScope b/c it causes new lifetime issues. Result value is now simply moved into the parent scope for ownership purposes (nothing an occassional cwal_scope_sweep() won't take care of). file: [bbdcfb93b1] check-in: [e99d61e441] user: stephan branch: trunk, size: 149848 | |
11:27 | Got first prototype script token eval working. file: [298c2c9876] check-in: [c74131e774] user: stephan branch: trunk, size: 151036 | |
09:30 | removed resultScope.level adjustment from call-in-scope op. file: [f303c2a56b] check-in: [f1d704b33e] user: stephan branch: trunk, size: 149871 | |
05:36 | Minor fixes(?) to the resultScope handling. Started some refactoring to consolidate property handling of Object and friends. file: [433803b981] check-in: [a9ff58bb0e] user: stephan branch: trunk, size: 149754 | |
04:07 | Seem to have gotten a reasonable return-value mechanism in place. The JSON output API now fails gracefully when faced with cycles. file: [9c8fff7948] check-in: [027070161a] user: stephan branch: trunk, size: 148597 | |
00:37 | minor internal cleanups file: [89b1c51eab] check-in: [74e03b3df5] user: stephan branch: trunk, size: 145389 | |
2012-08-10
| ||
21:51 | Had to add a second link to cwal_value. i cannot believe it even worked this far with one link. file: [5b49db070f] check-in: [ede1cffe1b] user: stephan branch: trunk, size: 144904 | |
17:10 | oops - moved cwal_list.h to where it belongs. file: [eaa1044c06] check-in: [432f1c4b11] user: stephan branch: trunk, size: 143557 | |
14:56 | "var" API seems to work. file: [f3a5b88301] check-in: [bc65ff9434] user: stephan branch: trunk, size: 143662 | |
2012-08-09
| ||
20:59 | Added the cwal_var_xxx() API, but it is completely untested (and will remain so until at least tomorrow ;). file: [0b408933b5] check-in: [00060a9b01] user: stephan branch: trunk, size: 143027 | |
17:38 | Added cwal_scope_sweep(). Reworked the trace masking (b/c 31 bits really is too few). file: [a572d7675f] check-in: [40e15d31ab] user: stephan branch: trunk, size: 136778 | |
16:05 | s/api/vtab/g for cwal_engine and cwal_value. file: [172c140354] check-in: [922c26c199] user: stephan branch: trunk, size: 135079 | |
15:37 | Fixed memory corruption in the kvp recycling code. Added cwal_engine_vtab::init_engine(). file: [d17c3e1481] check-in: [28ea4d13f9] user: stephan branch: trunk, size: 135026 | |
14:39 | Added basics for the Function type. file: [b5c4dd47d2] check-in: [2506f0d91a] user: stephan branch: trunk, size: 134547 | |
12:31 | Moved cwal_obase::list into cwal_array. Fixed json code to use cwal_object::kvp. Removed the cycle-counting code, since the current API doesn't use it. (We might need it later, though.) file: [fd39af08b2] check-in: [77b2583370] user: stephan branch: obj-kvp-list, size: 129240 | |
12:07 | Ah, it was just the child xscoping :). file: [8f64d09059] check-in: [ee9a15f061] user: stephan branch: obj-kvp-list, size: 134812 | |
11:52 | initial porting of cwal_object properties to a cwal_kvp linked list. Compiles but segfaults predictibly on some test code. file: [1b9c0517a4] check-in: [7c2b40e63a] user: stephan branch: obj-kvp-list, size: 134870 | |
10:30 | started adding hash/compare infrastructure to allow some downstream features like arbitrary keys for object properties and interning of arbitrary POD values. file: [66730832c5] check-in: [03c22e0473] user: stephan branch: trunk, size: 136254 | |
2012-08-06
| ||
22:05 | disabled auto-interning by default, minor tinkering with the VOID_PTR_IS_BIG bits (hope it still builds on x64). file: [ff385a06e8] check-in: [b53e520cfc] user: stephan branch: trunk, size: 129155 | |
20:57 | Minor config-level tweaking. Fixed some signed/unsigned screwyiness (thank you, clang). file: [03fbc9c540] check-in: [51fc55ddf7] user: stephan branch: trunk, size: 128956 | |
20:20 | Tinkering with interned string table size, other minor cleanups. file: [7407e2f067] check-in: [8e6142c151] user: stephan branch: trunk, size: 128757 | |
19:25 | Fixed some bugs in the scope cleanup code triggered by interned strings, and incidentally ended up trimming the algo from about 40 lines and two nested loops to only 1 functional line (and several tracing/checking lines) in a single loop. file: [4eba6f7b1c] check-in: [02af50b668] user: stephan branch: trunk, size: 127692 | |
17:22 | removed a fixme note (fixed!) file: [7917056c62] check-in: [73c853583a] user: stephan branch: trunk, size: 128611 | |
17:02 | added Value binding for cwal_buffer class. file: [246983101c] check-in: [f378d18817] user: stephan branch: trunk, size: 128679 | |
16:22 | fixed up/down-scope terminology use. file: [01edcaabbb] check-in: [782b3781ad] user: stephan branch: trunk, size: 126411 | |
13:47 | dammit, i am simply not going to be able to get that to compile in both 32- and 64-bit modes, am i? file: [66e7dcf161] check-in: [44f5d6f109] user: stephan branch: trunk, size: 126412 | |
2012-08-05
| ||
19:27 | final (i think) tinkering for the night. file: [4216d6a377] check-in: [ee81b02749] user: stephan branch: trunk, size: 126412 | |
19:09 | Tracing tweaks. Fixed a bitmasking range error in the tracing flags. file: [9218bc4f90] check-in: [8414ccbc5b] user: stephan branch: trunk, size: 126230 | |
17:51 | valgrind likes it. time to walk Baako. file: [5b187f0e49] check-in: [0def871fe5] user: stephan branch: trunk, size: 125463 | |
15:59 | minor cleanups, nothing notable. file: [e57b9e9198] check-in: [6295b36c28] user: stephan branch: trunk, size: 124176 | |
13:21 | auto-interning of strings can now be toggled on/off at runtime. file: [fab222ad43] check-in: [3244397263] user: stephan branch: trunk, size: 127394 | |
12:45 | Turned the value linked list from doubly- to singly-linked. Shaves off 4-8 bytes/value (32/64-bit). file: [0b691e08ce] check-in: [0b4ebfbc5b] user: stephan branch: trunk, size: 126550 | |
12:07 | Refactored recycling list to use value linked lists, removing the 5 mallocs needed for the lists and cutting the cost of the recycler a bit. file: [abdbb14249] check-in: [1489228c0d] user: stephan branch: trunk, size: 127022 | |
11:13 | Refactored delayed GC to use the new linked list, cutting its memory cost to a single pointer, removing any allocation it used to do, speeding GC-push up to O(1), and removing some unrecoverable error cases caused by OOM conditions. :-D file: [d049fb1f1f] check-in: [c29b48dd13] user: stephan branch: trunk, size: 128081 | |
10:43 | Refactored scope/value parenting to use a linked list. Costs 2 pointers per value but ends up costing fewer mallocs and less memory overall because some (large) scope-level overhead goes away. This also makes some ops O(1) which were previously O(N). file: [a1eb6b1337] check-in: [1f8844853b] user: stephan branch: trunk, size: 127569 | |
2012-08-04
| ||
20:57 | minor tinkering before bed. nothing functional. file: [fcee0e206c] check-in: [911afc1584] user: stephan branch: trunk, size: 128373 | |
20:21 | Renamed cwal_value_pop() to cwal_value_take(). file: [05be13c31a] check-in: [3e93418fa1] user: stephan branch: trunk, size: 128383 | |
14:18 | minor tinkering. file: [1fea21820b] check-in: [1a9442de79] user: stephan branch: trunk, size: 128320 | |
13:28 | more work on the tracing API. Added an engine param to cwal_value_ref(). file: [7e6df8f6c9] check-in: [4a5609991f] user: stephan branch: trunk, size: 128248 | |
00:23 | Fixed a string-re-use bug in cwal_trace(). file: [fe6ae8bcd4] check-in: [b28f3a9106] user: stephan branch: trunk, size: 125632 | |
2012-08-03
| ||
16:12 | First draft of tracing calls. file: [3df4976231] check-in: [ea25245906] user: stephan branch: trunk, size: 125527 | |
00:13 | Started adding more tracing infrastructure to replace most of the debug output. file: [d6658a7034] check-in: [536bb58c0b] user: stephan branch: scope-parenting, size: 122326 | |
2012-08-02
| ||
21:57 | Had to add auto-down-scoping of container children when a container is moved down, to avoid an up/down cross-scope cleanup problem. Seems to work. file: [c055ed2c3f] check-in: [3b1fc938cf] user: stephan branch: scope-parenting, size: 121419 | |
19:43 | valgrind seems to like this construct. file: [b5ea205768] check-in: [f3b2b88b06] user: stephan branch: scope-parenting, size: 118770 | |
17:43 | still more refactoring and docs. file: [619da76045] check-in: [0fbe8fb142] user: stephan branch: scope-parenting, size: 118310 | |
16:54 | more refactoring and docs. file: [81d795b0cf] check-in: [1053654f66] user: stephan branch: scope-parenting, size: 116853 | |
2012-08-01
| ||
23:28 | started adding infrastructure for tracing engine activity via a callback. file: [e584ae6af4] check-in: [7334a03423] user: stephan branch: scope-parenting, size: 116125 | |
22:08 | Demonstrates a client-supplied native living amongst the cycle gang :). file: [081556f338] check-in: [60d147bb16] user: stephan branch: scope-parenting, size: 115987 | |
21:27 | This one shows a well-knotted array/object graph going down the drain (in a good way). file: [cc51a2263a] check-in: [3cbd9d5079] user: stephan branch: scope-parenting, size: 115862 | |
17:49 | array cycling appears to work after scope-wandering refactoring. file: [8bd1f29ad5] check-in: [fb7610aa39] user: stephan branch: scope-parenting, size: 114891 | |
16:20 | fixed a counting error in value-list-removal. file: [71f3d4e605] check-in: [7c58c67231] user: stephan branch: scope-parenting, size: 113048 | |
2012-07-31
| ||
23:12 | added a todo. file: [f556aea884] check-in: [2de038aa43] user: stephan branch: scope-parenting, size: 112962 | |
22:29 | time for bed before i get carried away with getting the containers in order. file: [ec60a92814] check-in: [be2c482a7f] user: stephan branch: scope-parenting, size: 112839 | |
22:02 | i broke down and added a scope ptr to the values. This costs me but it greatly simplifies some problems and code paths until i can find a way to do it safely without the ptr. file: [ad80ec8c74] check-in: [11a0bb37af] user: stephan branch: scope-parenting, size: 112106 | |
20:21 | Tweaking of the string-interning table size and hash function (found a combination i like better). Initial POD destruction looks good except i am going to run into a problem with lower-level scopes when they get refs to strings from higher-level scopes. i think i can only fix this by adding a scope ptr to each value :/. file: [1930b5fe80] check-in: [f0b3dff0a7] user: stephan branch: scope-parenting, size: 111151 | |
18:29 | Started rewriting test code in baby steps to ensure new tracking/cleanup mechanism is working as desired. Lots not working right now, but the path looks clear :-D. file: [a5392d0ab3] check-in: [7fd3dcf630] user: stephan branch: scope-parenting, size: 109979 | |
14:39 | leaking atoms in some cases, but need to move to another machine... file: [9a9c762341] check-in: [a0ae26ab7c] user: stephan branch: scope-parenting, size: 111105 | |
03:26 | still leaking atoms, but debuggering will have to wait another 13 hours or so... file: [1157cefd30] check-in: [92cd1cd63f] user: stephan branch: scope-parenting, size: 110990 | |
02:34 | ALMOST got scope-parenting-based cleanup working. Leaking primitives now but not composites, it seems. file: [c3646fa10e] check-in: [0da5da5d1a] user: stephan branch: scope-parenting, size: 109045 | |
2012-07-30
| ||
22:51 | More tinkering with the string interning (to find a decent memory balance) and started add cycle-clearing/counting, but it is not working and i am too tired to debugger it. file: [c3615beb39] check-in: [e45735c13d] user: stephan branch: trunk, size: 107016 | |
19:32 | Minor tinkering with the scope cleanup and debug output. file: [f23894ce4f] check-in: [5c4e2e8c47] user: stephan branch: trunk, size: 100364 | |
18:24 | Added cwal_dump_interned_strings_table(), revealing what _appears_ to be a bug in the string de-internalization. Todo... file: [a8ce9d66ad] check-in: [f343473f84] user: stephan branch: trunk, size: 98874 | |
17:37 | Increased default string internalization table size to approx. 2kb/table (exact value depends on sizeof(void*) and a nearby prime number). file: [002152a56b] check-in: [c7ae7457d0] user: stephan branch: trunk, size: 97170 | |
17:13 | Refactored cwal_scope::allocs into two lists, one for containers and one for atomics, to help optimize some operations. file: [e88bbe5c9c] check-in: [de3b8113d7] user: stephan branch: trunk, size: 97149 | |
16:24 | Changed return semantics of cwal_value_unref() to provide precise info about what is does/does not do. file: [dc3bf2f910] check-in: [7319c8feb0] user: stephan branch: trunk, size: 96215 | |
2012-07-29
| ||
20:58 | more internal tinkering, added cwal_engine_recycle_max(). file: [1c9bac64a3] check-in: [bd79d5e6d8] user: stephan branch: trunk, size: 96153 | |
18:34 | Added string-interning support. Now all new strings with the same bytes get auto-interned/re-used. There might be a cross-scope ownership problem here, but only (i think) if we leave the "strict scope ordering" realm (e.g. for closures, which are not a high-prio feature). file: [4eb5b19e73] check-in: [32a0ade6f0] user: stephan branch: trunk, size: 94738 | |
16:10 | Started adding support for interning strings (so we share _all_ strings which have the same content), but it has a serious Denkfehler in how i am using the hashtable and needs to be fixed. file: [ea0cadb4c4] check-in: [19759dcf7b] user: stephan branch: trunk, size: 89823 | |
03:40 | Added initial 'native' (a.k.a. Host Object) binding support. file: [c6e31a9b67] check-in: [c5d9755013] user: stephan branch: trunk, size: 85492 | |
02:07 | more minor refactoring. file: [2595b3e638] check-in: [c0912871aa] user: stephan branch: trunk, size: 83683 | |
2012-07-28
| ||
23:52 | Ported in most of the missing cson docs. Minor refactoring. file: [60efcb8bdf] check-in: [0e4f895e99] user: stephan branch: trunk, size: 80398 | |
21:14 | added cwal_json_output() and friends, ported from libcson. Changed the signature for cwal_output_f as part of that. file: [c89da1b103] check-in: [d16e8ab46a] user: stephan branch: trunk, size: 87153 | |
20:20 | An assertion fix in cwal_value_push(). cwal_value_pop() now manipulates the scope alloc list to avoid endless expansion of it. file: [3fd473f803] check-in: [48fd00e698] user: stephan branch: trunk, size: 86918 | |
18:49 | More private/internal refactoring. Seeing a weird bug in e_such_much_recycling(), but am too tired to chase it down. file: [c94f8e075c] check-in: [fc32638cd1] user: stephan branch: trunk, size: 86849 | |
17:19 | removed cwal_engine::scopes, subsumed by the cwal_engine::current linked list we already had. file: [37c16c95f3] check-in: [0062bd8ecf] user: stephan branch: trunk, size: 87640 | |
16:46 | started refactoring into two headers (public+internal). file: [daa01f2bb7] check-in: [fc05f69b2c] user: stephan branch: trunk, size: 88771 | |
14:12 | merged in cycle-counting branch. Works at least as well as it did before and has basic support for graph destruction for "relatively little" cost. file: [2515356da5] check-in: [e58a47936d] user: stephan branch: trunk, size: 88384 | |
13:51 | Added initial value allocation recyler. Lots of other tweaks, improvements, and debug output. file: [26a86ed437] check-in: [02ab3dc56e] user: stephan branch: cycle-counting, size: 88473 | |
04:56 | too... tired... file: [9f839da2ac] check-in: [73a2e1b526] user: stephan branch: cycle-counting, size: 84234 | |
02:21 | minor tinkering. Fixed, i think, a future bug when cycle counting more than once. file: [8f0109aba0] check-in: [92dd48c7c1] user: stephan branch: cycle-counting, size: 82535 | |
01:17 | more docs and minor refactoring/cleanups. file: [5de7282e11] check-in: [31d896646e] user: stephan branch: cycle-counting, size: 79652 | |
2012-07-27
| ||
23:35 | Added boatloads of API docs. Still far from finished (many still to be copied from cson). file: [57d4e38ab6] check-in: [6c4c514f74] user: stephan branch: cycle-counting, size: 77835 | |
20:18 | this is a beautiful thing when seen through the eyes of valgrind... file: [0beff61a14] check-in: [354aa42b4e] user: stephan branch: cycle-counting, size: 74729 | |
20:13 | this one looks a bit better. Fixed a REALLY dumb flags bitset bug which caused a mis-diagnosed is-destructing check. file: [87f0ae737b] check-in: [aed1986bae] user: stephan branch: cycle-counting, size: 74729 | |
17:42 | minor cleanups before the jackhammering begins. file: [1b9cf7dd6b] check-in: [50037bf0e2] user: stephan branch: cycle-counting, size: 72445 | |
16:54 | Improved the [f9924f2539] workaround (i think). file: [a04be01d44] check-in: [5a0e3b69cf] user: stephan branch: cycle-counting, size: 72100 | |
16:49 | merged in [f9924f2539] compile fix. file: [5d024c8cec] check-in: [e29713f468] user: stephan branch: cycle-counting, size: 72001 | |
16:49 | backed up a bit to remove some dysfunctional complexity. file: [c68e76916f] check-in: [a5ba13ffc8] user: stephan branch: cycle-counting, size: 71979 | |
13:26 | Temporary(?) hack for 32/64-bit compatibility. file: [2072dc7ddd] check-in: [f9924f2539] user: stephan branch: trunk, size: 68479 | |
02:06 | Auto-cycle-count/clean is now done at scope-pop, but this is not a complete solution because values can outlive a scope. file: [df7218aeec] check-in: [303510704d] user: stephan branch: cycle-counting, size: 74136 | |
00:35 | First cyclic free-up demonstrated when the parent scope expired :-D, but requires that one run the cycle-counter first :/. Making that transparent is on the TODO list for the weekend :D. Reminder to self: we might be able to put that in the Scope cleanup. file: [544e547565] check-in: [5b1208c9f8] user: stephan branch: cycle-counting, size: 70855 | |
2012-07-26
| ||
22:51 | Contains an intentional memleak due to incomplete refactoring, but it is time for bed and do not want to risk losing this. file: [b07a73602b] check-in: [d73fbf518a] user: stephan branch: cycle-counting, size: 70010 | |
2012-07-25
| ||
17:33 | Added engine::current scope convenience ptr. file: [ff40907dc2] check-in: [9f145a967c] user: stephan branch: trunk, size: 68457 | |
14:53 | Tinkering with value-scope parenting, but the built-in values just threw a wrench in it. file: [25c2022877] check-in: [c8298bc6d1] user: stephan branch: trunk, size: 68259 | |
13:05 | a number of API tweaks. file: [d2c4d2774a] check-in: [56ac8240e7] user: stephan branch: trunk, size: 66780 | |
10:33 | more minor refactoring. file: [5970577cf1] check-in: [7d83b4f0e9] user: stephan branch: trunk, size: 64446 | |
10:01 | more refactoring and tinkering. file: [c27d0677ad] check-in: [348267e3b3] user: stephan branch: trunk, size: 64817 | |
08:05 | ported in the object API. file: [5dc2b22456] check-in: [e77bdf86b5] user: stephan branch: trunk, size: 58105 | |
06:37 | switched to stdint fixed-size ints throughout. Added cwal_size_t and renamed a few things. file: [7e0e480a50] check-in: [bb5b0db794] user: stephan branch: trunk, size: 51045 | |
05:39 | more refactoring. file: [42407d5c7a] check-in: [89909b7eac] user: stephan branch: trunk, size: 49255 | |
05:05 | got the object skeleton in place but it is missing any useful features. file: [755c438a36] check-in: [84f2fc8b27] user: stephan branch: trunk, size: 49232 | |
04:14 | minor refactoring. file: [321ce52b3d] check-in: [1868791179] user: stephan branch: trunk, size: 47043 | |
03:14 | Tinkering with various loads for the ptr table. 90% only causes 2 tables under the right circumstances, the second one only lightly loaded. file: [289d6d4f5a] check-in: [91da9fef15] user: stephan branch: trunk, size: 46423 | |
02:02 | Got an array containing a self-cycle to free without upsetting valgrind. :) file: [005a3c2310] check-in: [4dcb87c117] user: stephan branch: trunk, size: 44640 | |
01:47 | ptr table is looking good. file: [b13f13170e] check-in: [d88a5fa845] user: stephan branch: trunk, size: 44691 | |
2012-07-24
| ||
21:45 | ptr table almost works right, but page->entryCount seems to be off by one sometimes? To check tomorrow... file: [f20b867f5d] check-in: [0c98670772] user: stephan branch: trunk, size: 43449 | |
20:31 | got arrays ported in. file: [a5a5217a9e] check-in: [c1e7d24d89] user: stephan branch: trunk, size: 41544 | |
19:43 | new "key table" seems to work. Now i am not sure i need it :/. file: [39af67b50f] check-in: [0c7c5ce695] user: stephan branch: trunk, size: 35514 | |
19:19 | insert/search seem to work on the ptr table. file: [48b299747a] check-in: [69c3ef00e5] user: stephan branch: trunk, size: 34242 | |
17:03 | refactored page list to use a cwal_list (simplifies the code notably). file: [b7389b7495] check-in: [fb89dc98ad] user: stephan branch: trunk, size: 32579 | |
16:47 | some more work on the weird hashtable. file: [b7398bcc28] check-in: [33c220873c] user: stephan branch: trunk, size: 32633 | |
00:03 | Started adding special-case ptr-hashtable for tracking cycles and is-var-in-this-scope checks. file: [57de6b423d] check-in: [de91c24c1c] user: stephan branch: trunk, size: 30542 | |
2012-07-23
| ||
21:58 | We can now allocate vars, but scoping is not far enough to add the allocations there. file: [60d3f2cfa2] check-in: [d0cc64e478] user: stephan branch: trunk, size: 28127 | |
20:14 | We can now free values but not allocate them. Object/Array will be missing for a while. file: [6c76bfd0e5] check-in: [13ccbfc670] user: stephan branch: trunk, size: 14136 | |
19:49 | More infrastructure... file: [40c0085bb6] check-in: [25b7b37575] user: stephan branch: trunk, size: 6034 | |
19:26 | More infrastructure added. Still a long way to go. file: [56cbf063f7] check-in: [bff484cc73] user: stephan branch: trunk, size: 5768 | |
18:47 | Added: egg file: [55b5ba2ae3] check-in: [a71b168320] user: stephan branch: trunk, size: 4757 | |