Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
120 check-ins using file src/checkin.c version e913153109
|
2022-03-05
| ||
| 10:08 | Added protection against a corner-case condition to the previous checkin. check-in: fdebf174bb user: stephan tags: trunk | |
| 10:02 | When a fossil-managed db handle does a rollback, reload the cached ckout version state, else it can get out of sync. This is an age-old bug, but was just discovered when it broke some new f-sanity tests. check-in: a0f3423462 user: stephan tags: trunk | |
|
2022-03-04
| ||
| 16:14 | Ported over Daniel D's root:XYZ tag lookup performance improvements from [fossil:/timeline?r=start-of-branch-cte]. check-in: 51863fd402 user: stephan tags: trunk | |
| 05:37 | Latest upstream fnc. check-in: 1f0c719961 user: stephan tags: trunk | |
|
2022-02-26
| ||
| 22:47 | Latest upstream fnc.c check-in: 8520337095 user: stephan tags: trunk | |
| 21:01 | Re-enabled the historical 'touch' feature of mkcdep.c to work around some otherwise weak deps handling when headers change. On the one hand, this improves rebuilding for header changes. On the other, it causes the deps files to get rebuilt twice in a row at times. check-in: d8a6b8d0c6 user: stephan tags: trunk | |
|
2022-02-24
| ||
| 11:19 | mkcdep.c no longer adds "." to the -I path by default because that backfires with generation of the amalgamation header, causing a circular dep because the amalgamation header shares a name with a non-amalgamation header. check-in: f59415334b user: stephan tags: trunk | |
|
2022-02-18
| ||
| 16:36 | Back-ported [fossil:868d1608382a9bb1] for (cosmetic) diff alignment improvements. check-in: 57f6086d58 user: stephan tags: trunk | |
| 16:25 | Improved the big/little-endian detection in mkcdep.c by taking advantage of a gcc-ism which exists for that purposes. check-in: 59dea46ad9 user: stephan tags: trunk | |
|
2022-02-15
| ||
| 14:46 | As per Stephan's suggestion in /chat, pluralise ignoreRid(s). check-in: 36b2799c8e user: mark tags: dev-blame-ignore | |
| 14:33 | Enable ignoring specific commits when annotating file. Add flag to annotate options struct to ignore all merge commits. Similarly, provide array to assign rid values of specific versions to ignore specified commits. If either are set, any changed lines from the ignored version(s) will be annotated from the previous commit in the file history. check-in: 79fd090adb user: mark tags: dev-blame-ignore | |
|
2022-02-14
| ||
| 10:06 | Replaced broken big-/little-endian determination in mkcdep.c. check-in: e71174b745 user: stephan tags: trunk | |
|
2022-02-05
| ||
| 05:37 | fsl_foci() now uses the fsl_cx instance associated with its db handle instead of fetching it from fcli. check-in: 746d6bb38a user: stephan tags: trunk | |
| 05:04 | Expanded the foci docs a bit. check-in: df374a82ef user: stephan tags: trunk | |
| 04:51 | Merged in wip-foci branch, a straight port of fossil's files_of_checkin sqlite virtual table. check-in: b78e11a588 user: stephan tags: trunk | |
| 04:45 | Got foci working and added a basic test for it in f-sanity.c. Closed-Leaf check-in: c33ddf9c91 user: stephan tags: wip-foci | |
|
2022-02-04
| ||
| 14:22 | Initial port of Fossil's files-of-checkin (foci) vtable. First pass of porting fossil(1) src/foci.c to libfossil; it's pretty much a 1:1 port using libf's ADTs. This builds, but the implementation isn't used anywhere so that's not a surprise. There is no test yet either, so branching this for further review. check-in: 677d51061d user: mark tags: wip-foci | |
|
2022-01-30
| ||
| 13:50 | Documented a bug/shortcoming in f-vdiff. check-in: 725861fdf6 user: stephan tags: trunk | |
|
2022-01-27
| ||
| 14:42 | Re-did C dependencies generation to use tools/mkcdep.c instead of GCC. MUCH faster and less kludgy. check-in: 236c0f8864 user: stephan tags: trunk | |
| 13:02 | Latest upstream fnc. check-in: e7795f680c user: stephan tags: trunk | |
| 10:08 | Removed some dead code from wh-common.tcl. check-in: 74d87e19f4 user: stephan tags: trunk | |
|
2022-01-26
| ||
| 18:04 | Corrected an argument type in fsl_branch_of_rid() which broke variadic arg handling in that function on platforms where fsl_int_t and fsl_id_t are different sizes. check-in: bcbfb21894 user: stephan tags: trunk | |
|
2022-01-24
| ||
| 07:17 | Corrected f-vdiff's file-mtime checks of local-checkout files to canonicalize the filename so that it works when called from a subdir of the checkout. check-in: 4da4e6a3f4 user: stephan tags: trunk | |
| 03:44 | Ported in upstream [fossil:8cd73dda3d10ff143740] diff generation tweaks. check-in: 75349fc193 user: stephan tags: trunk | |
|
2022-01-23
| ||
| 18:58 | Removed duplicate docs for fsl__diff_optimize(). check-in: d8d1bc92a4 user: stephan tags: trunk | |
| 18:57 | Now that the v1 diff code is gone, consolidated diff.c/diff2.c into diff.c and moved the dibu-specific code into the new dibu.c. That required exposing fsl__diff_optimize() via internal.h so that fnc.c can see it. (Edit: in hindsight, should have renamed diff2.c to dibu.c since most of diff2.c was dibu stuff. Oh, well.) check-in: 2cca86d9a0 user: stephan tags: trunk | |
| 13:47 | Fixed a segfault introduced two checkins back in [5e12e2134b76]. check-in: a8853a173b user: stephan tags: trunk | |
| 13:43 | Minor cleanup to the previous checkin. Fixed TK-format output to be syntactically correct for a no-diff run. check-in: 354f81fd5c user: stephan tags: trunk | |
| 13:19 | f-vdiff can now catch and report renames when the rename is part of the local checkout's state. Code-adjacent constness tweaks. Edit: FIXME: discovered later that this fails when the diff is run from anywhere but the top checkout dir. check-in: 5e12e2134b user: stephan tags: trunk | |
| 12:03 | Ported in the non-debug-output part of upstream [fossil:9aaefcfd0a8746b63bad]. check-in: d58a4b6237 user: stephan tags: trunk | |
| 11:24 | Minor internal doc tweaks. check-in: 12c9022cbb user: stephan tags: trunk | |
| 08:57 | Ported in upstream diff changes [fossil:fbdbc09b402b19de] and [fossil:c311efef078c8733]. Diffing still seems to work. check-in: 93f03cf9ee user: stephan tags: trunk | |
|
2022-01-22
| ||
| 18:08 | Upstream diff fix [fossil:44b6f128453b35039bc2]. check-in: 6725e728d4 user: stephan tags: trunk | |
| 18:02 | Ported in upstream incremental diff tweaks from [fossil:f6112b93e9bbb7e37f24] except that many of the const removal changes are unnecessary and weren't ported. Made fsl_dibu::opt const. Probably should have waited to see if these will be trunked in fossil before going down this rabbit hole. check-in: 01225604ab user: stephan tags: trunk | |
| 14:27 | Removed the disabled, and fundamentally flawed, attempt at catching renames in f-vdiff. check-in: 4866db6ef5 user: stephan tags: trunk | |
| 14:26 | Ported in upstream diff patch [fossil:2dad4158dbf0d92b]. check-in: 5492222b8e user: stephan tags: trunk | |
|
2022-01-20
| ||
| 17:47 | Workaround for the Excessively Massive Diff bug (in f-vdiff, not the library) by reporting renames as an add/remove pair. Make the default split-mode diff content column widths wide enough for f-vdiff's current handling of how adds/removes are reported. check-in: 1960fea37a user: stephan tags: trunk | |
| 16:34 | Latest upstream fnc, which moves the former v1 libf diff code into fnc. check-in: ed78994d64 user: stephan tags: trunk | |
| 15:55 | Merged in rm-diff-v1. This will currently only build when using the --no-fnc configure flag, pending Mark trunking his related fnc changes, at which point that updated fnc will be pulled into this tree. Recursive relationships are tricky sometimes. check-in: 74d90fcdd8 user: stephan tags: trunk | |
| 15:53 | auto.def: distinguish between the --no-fnc flag and "don't use ncurses at all" so that the ncurses-based diff view can be built when fnc is disabled. check-in: eb5e8319b3 user: stephan tags: trunk | |
| 09:38 | Replace calls to the deprecated fsl_ckout_close(f) routine. Replace with fsl_close_scm_dbs(f). check-in: f876a65bef user: mark tags: trunk | |
| 08:54 | Remove deprecated diff v1 code as per /chat discussion. This removes all exclusive diff v1 library routines and data types in preparation for importing this implementation into fnc as per chat with Stephan. It is on a branch pending review in case any bits were missed or incorrectly purged. Closed-Leaf check-in: dbf0fb30dd user: mark tags: rm-diff-v1 | |
|
2022-01-16
| ||
| 07:40 | speeling check-in: a9b499a8de user: bch tags: trunk | |
| 06:44 | f-vdiff now actively checks whether it's being asked to diff ostensibly binary content and refuses to do so. check-in: 47413486d0 user: stephan tags: trunk | |
| 05:04 | Re-added subdir-inc.make (it's used by doxygen build). Updated a path to account for new hoster. check-in: 12e08213af user: stephan tags: trunk | |
|
2022-01-12
| ||
| 04:46 | Added another test for fsl_tkt_id_to_rids(). check-in: a8d5615853 user: stephan tags: trunk | |
| 03:36 | Initialized a couple vars which gcc 11 now incorrectly reports as possibly uninitialized. Sigh. check-in: 19607914ca user: stephan tags: trunk | |
|
2022-01-08
| ||
| 05:45 | Per /chat discussion, changed semantics of fsl_tkt_id_to_rids() to be able to handle an unambiguous ticket ID prefix. check-in: 2bdb6c668f user: stephan tags: trunk | |
| 03:56 | Replaced the fundamentally flawed fsl_tkt_id_to_blob() with fsl_tkt_id_to_rids(). check-in: 7b2cbb1669 user: stephan tags: trunk | |
|
2022-01-07
| ||
| 23:40 | Added fsl_deck_foreach(). check-in: 0741f4d194 user: stephan tags: trunk | |
| 19:06 | Added fsl_tkt_id_to_blob() for Brad. check-in: 7d64642d36 user: stephan tags: trunk | |
| 10:47 | difftk.tcl: increased default width of line number columns. Inexplicably, using this exact same code from fossil expands the columns dynamically but this implementation is not doing so. Disabled {} around f-vdiff TCL output except in tk mode, for compatibility with fossil's (diff -tcl) output. check-in: 37b6d8e11f user: stephan tags: trunk | |
| 06:28 | typo check-in: d069d8d19f user: bch tags: trunk | |
| 06:28 | reference fsl_deck_C_set() in comment-notes check-in: 70abc44543 user: bch tags: trunk | |
|
2022-01-04
| ||
| 22:02 | Added fsl_randomness(). Changed fsl_deck_K_set() semantics such that a NULL ticket ID is interpreted as a new (random) ID. check-in: 896ea2a01b user: stephan tags: trunk | |
| 21:38 | Replaced a couple remaining char-as-bool references with bool. check-in: 1ac20eeb41 user: stephan tags: trunk | |
| 17:15 | Merged in inadvertent fork [fc8fb3a392]. check-in: 9259168d71 user: stephan tags: trunk | |
| 17:01 | Latest upstream fnc. check-in: 1bdf40e928 user: stephan tags: trunk | |
| 17:00 | Experimentally applied the "small gap patch" to eliminate the so-called "lineno behavior" in the diff engine. See the copious comments in diff2.c for details. check-in: 2872db4659 user: stephan tags: trunk | |
| 04:54 | Corrected a newly-broken dependency on a newly-private API. check-in: fc8fb3a392 user: stephan tags: trunk | |
| 03:51 | Corrected the no-diff display case in the ncurses diff view. Minor doc and constness tweaks. check-in: e358ca5350 user: stephan tags: trunk | |
| 03:08 | ncurses diff view: add keybindings to jump to next/previous file in the diff list. check-in: aa723e9a15 user: stephan tags: trunk | |
| 00:44 | Minor doc tweaks. check-in: a7861ad0da user: stephan tags: trunk | |
|
2022-01-03
| ||
| 22:46 | More minor cleanups in the ncurses-related bits. check-in: 1332923f62 user: stephan tags: trunk | |
| 22:27 | Moved ncurses screen init/shutdown out of f-vdiff and into the the ncurses diff builder's start()/finally() methods. Cleaned up the ncurses-related public API, making most of it private. check-in: 1c19687346 user: stephan tags: trunk | |
| 21:30 | Renamed fsl_diff_opt to fsl_dibu_opt in fnc.c. check-in: ef858ebe4f user: stephan tags: trunk | |
| 21:16 | Renamed fsl_diff_opt to fsl_dibu_opt. Moved dibu-type-specific flags out of the FSL_DIFF2_ flags and into the new FSL_DIBU_ flags. Added fsl_dibu::pimplFlags, which are implementation-private flags distinct from fsl_dibu::implFlags, the latter now being the place to apply FSL_DIBU_ flags. check-in: be968bbac7 user: stephan tags: trunk | |
| 19:42 | Improved, but not perfected, f-vdiff's ability to report add/remove/rename in a dibu-friendly way (instead of via f_out(), which is incompatible with the ncurses and tcl dibus). check-in: 096b759623 user: stephan tags: trunk | |
| 19:41 | Added file hashes to the file index lines in the ncurses unified-ish diff view. check-in: 6e33c8e83c user: stephan tags: trunk | |
| 19:39 | Minor tinkering in as-yet-unused fsl_popen2(). check-in: 03868ff9d2 user: stephan tags: trunk | |
| 19:38 | Added a "hard cached" fsl_stmt for fsl_content_blob(), shaving a few allocs and a small bit of memory from a rebuild. check-in: a7de183cd0 user: stephan tags: trunk | |
| 17:21 | Ported in fossil's diff -tk script bits and implemented (f-vdiff --format tk) but the output currently goes to stdout instead of launching tclsh. The TCL dibu can optionally include the TK bits or not, depending on its flags. check-in: 82bf08414f user: stephan tags: trunk | |
| 16:39 | Minor build reorgs as part of integrating (fossil diff -tk)-ish feature. check-in: 574b521a2b user: stephan tags: trunk | |
|
2022-01-02
| ||
| 16:10 | Minor doc updates and code reformatting. check-in: 707fecabb9 user: stephan tags: trunk | |
| 15:50 | Removed a stray failed experiment from the previous checkin. check-in: dd432df69f user: stephan tags: trunk | |
| 15:18 | ncurses unified-ish diff view: implemented delete/insertion pair collapsing and made the colors configurable (more or less). check-in: ca112741c9 user: stephan tags: trunk | |
| 01:19 | Removed long-unmaintained f-apps/test.c. check-in: 72b398779b user: stephan tags: trunk | |
| 00:32 | Added FSL_OMIT_DEPRECATED #define to cause the deprecated APIs to be elided. Got core lib building that way. Removed FSL_EXPORT from functions declared in deprecated.h. check-in: b7046bf39b user: stephan tags: trunk | |
|
2022-01-01
| ||
| 15:07 | Latest upstream fnc client. check-in: cdbcddb1a1 user: stephan tags: trunk | |
| 15:05 | Reorganized (again) fsl_cx db handle management to eliminate the middle-man temp/in-memory db. check-in: a23126bc9f user: stephan tags: trunk | |
| 15:01 | Fixed a bogus assert in f-open.c. Closed-Leaf check-in: dcdaea2f1b user: stephan tags: cx-db-handle-reorg-redux | |
| 14:58 | More work on the db handle redux. Seems to work okay. check-in: 71f5f09817 user: stephan tags: cx-db-handle-reorg-redux | |
| 14:15 | Fixed a require.s2 module broken by db handle refactoring. check-in: 44fae8d76f user: stephan tags: cx-db-handle-reorg-redux | |
| 14:12 | s2 bindings: updated a test which can fail when running from a branch. Fixed a broken attempt to copy a NULL buffer trigered by that failing test. check-in: bc3c7b1737 user: stephan tags: cx-db-handle-reorg-redux | |
| 13:34 | fsl_cx db handle reorg redux, step one. Significantly simplifies how dbs are closed, always treating them as an inseparable pair. Basic sanity checks pass, but more testing is needed and lots of docs to update. check-in: fa7c56d0ca user: stephan tags: cx-db-handle-reorg-redux | |
| 11:15 | Renaming of some internal fsl_xyz... APIs to fsl__xyz... (two underscores). Minor constness tweaks. check-in: 736abef2cc user: stephan tags: trunk | |
| 10:53 | fsl_db_open() no longer automatically applies the create-if-not-exists flag to in-memory or anonymous temp dbs, as this was at the root of a great deal of confusion the past 24 hours. Documented the propagation behaviour of the CREATE flag from an OPEN'd db to other dbs ATTACHed via that handle. See [https://sqlite.org/forum/forumpost/76963f0026b6b12a], and its top-level post, for more details. check-in: ac15c8d1d4 user: stephan tags: trunk | |
| 02:03 | Minor cleanups in the fsl_db API. Updated fsl_db SQL tracing to the sqlite3_trace_v2() API. check-in: 89e932ce1a user: stephan tags: trunk | |
| 01:13 | More cleanups moving towards taking another whack at the fsl_cx db handle reorg. Added a test case to f-sanity.c demonstrating the ATTACH-cannot-create error (whereas the sqlite3 shell can create new files that way). check-in: 51d4b05f5c user: stephan tags: trunk | |
|
2021-12-31
| ||
| 22:19 | Ported over more general-purposes pieces from the cx-db-handle-reorg branch in prep for taking a fresh whack at that. check-in: 39645644a7 user: stephan tags: trunk | |
| 21:54 | Cherrypicked [6f2027aeb4], [74964a09a5], and part of [d290f6375c]. check-in: 211eaecfcf user: stephan tags: trunk | |
| 21:36 | Cherrypicked [74964a09a5] (doxygen tag fix). check-in: 326482e519 user: stephan tags: trunk | |
| 21:34 | Cherrypicked [6e8b5b6677] (f-test-ciwoco Makefile.in reference fix). <s>(But, interestingly, it didn't show up as a cherrypick merge because i made multiple individual commits and the first one cleared the vmerge state. Hmmm.)</s> Edit: actually, the cherrypick happened, and was correctly recorded, in [f3a4ba9472]. This checkin was completely empty and should not have gone through. Hmm. check-in: 31cc119d75 user: stephan tags: trunk | |
| 21:33 | Removed old/unused bindings/Makefile.in. check-in: 8632822dd0 user: stephan tags: trunk | |
| 21:33 | f-merge now defaults to wet-run mode. check-in: f3a4ba9472 user: stephan tags: trunk | |
| 21:30 | Extended fsl__cx_attach_role() to be able to create a new db if needed (required for fsl_repo_create()) but (regardless of this change) discovered that ATTACH, when run through the new fsl_cx::dbMain handle, can no longer create new db files, which means that f-open cannot create a checkout. COMPLETELY stumped as to why that is. Edit: closing because this branch is failing in inexplicable ways and has a horrid mix of branch-relevant and branch-irrelevant changes. Pulling in the branch-irrelevant ones separately into trunk before taking another whack at the branch-relevant ones. Closed-Leaf check-in: a67b757dc0 user: stephan tags: cx-db-handle-reorg | |
| 21:26 | Corrected recently-broken Makefile.in references in f-test-ciwoco. check-in: 6e8b5b6677 user: stephan tags: cx-db-handle-reorg | |
| 15:57 | Reorganized fsl_cx-internal DB handles, eliminating the in-memory/temp main db and (once again) using the first of the repo/checkout which is opened as the main db. Seems to work okay but some fallout would not be terribly surprising. (Edit: moving back to cx-db-handle-reorg because of new, inexplicable inability to create checkout db via ATTACH.) check-in: 46656edc7e user: stephan tags: cx-db-handle-reorg | |
| 15:50 | Corrected names of several @file doxygen tags. check-in: 74964a09a5 user: stephan tags: cx-db-handle-reorg | |
| 15:48 | Doc updates. Moved all @deprecated functions into fossil-scm/deprecated.h for later weeding-out. check-in: 6f2027aeb4 user: stephan tags: cx-db-handle-reorg | |
| 15:10 | Renamed fsl_db_role_label() to fsl_db_role_name(). Doc touchups related to the db handle restructuring. check-in: d290f6375c user: stephan tags: cx-db-handle-reorg | |
| 15:08 | Removed old/unused bindings/Makefile.in. check-in: 6409509264 user: stephan tags: cx-db-handle-reorg | |
| 15:00 | fsl_ckout_open_dir() now treats a NULL dir name as ".". Updated C++ and s2 bindings for the new db changes. check-in: 05c1da0fb0 user: stephan tags: cx-db-handle-reorg | |
| 13:47 | fsl_buffer_external() may now be passed a populated buffer, clearing it before re-initializing it. check-in: 9626c2acf6 user: stephan tags: trunk | |
| 13:34 | Removed hard-coding of main db journal_mode=WAL for the time being. We might want to set it depending on which db is main (WAL for repo, TRUNCATE or DELETE for checkout?). Or we might want to leave it alone entirely. check-in: 09287e3d13 user: stephan tags: cx-db-handle-reorg | |
| 13:29 | Eliminated fsl_cx::dbMem in-memory/temp db and (again) use the first of the repo/ckout dbs which is opened as the MAIN db (aliasing it to its well-defined libf schema name). This has a high potential for fallout, so it's in a branch pending more thorough testing. check-in: aebf241ab5 user: stephan tags: cx-db-handle-reorg | |
|
2021-12-30
| ||
| 13:52 | Removed some debug output added by the previous checkin. check-in: 75eab853b5 user: stephan tags: trunk | |
| 13:49 | Global config db now uses its own fsl_db handle, no longer ATTACHed with the repo/checkout handle. fsl_ckout_open_dir() now canonicalizes the passed-in directory name (simplifies usage). Added f-query -g/--config flag to query against the global config db. Various code-adjacent tweaks. check-in: 1d7a6d2b96 user: stephan tags: trunk | |
| 05:37 | Replaced f-apps fcli_setup() calls with fcli_setup_v2(). check-in: 313b59841e user: stephan tags: trunk | |
| 05:06 | Unified-mode diff builder: reformat runs of delete/insert pairs from a series of D/I/D/I... lines to DD.../II..., like fossil does. check-in: 8752aca1b7 user: stephan tags: trunk | |
|
2021-12-29
| ||
| 23:08 | Added missing commas in JSON diff builder Edit record output. check-in: 257ae0b684 user: stephan tags: trunk | |
| 22:58 | Add backspace and form-feed to the characters escaped in the previous commit. check-in: ae8e038bdb user: stephan tags: trunk | |
| 22:49 | appendf json formatting: emit tabs as \t instead of \u0009. check-in: 50ebe24a14 user: stephan tags: trunk | |
| 03:08 | f-viff --save-format now also saves/restores the --line-numbers flag, if set. check-in: fbbbdfe556 user: stephan tags: trunk | |
|
2021-12-28
| ||
| 20:32 | ncurses unified diff: line number columns now size dynamically to fit their digits. check-in: 4883bff8f7 user: stephan tags: trunk | |
| 14:48 | Partial workaroud for the adjacent-code-block cosmetic issue in the unified text diff builder, eliminating an extraneous separator line. It only applies when using line numbers (so not technically unified diff), but the problem is less glaring now. check-in: 08fdbacefa user: stephan tags: trunk | |
|
2021-12-27
| ||
| 22:34 | Added line number support to the ncurses unified-ish diff view. check-in: 962200c936 user: stephan tags: trunk | |
| 21:07 | Corrected a rare corner case error check in the previous checkin. check-in: 5cccd60c7f user: stephan tags: trunk | |
| 15:00 | When performing a fsl_db rollback, clear the associated fsl_cx's manifest- and blob-content caches in order to ensure that they do not contain any RIDs which are part of the being-rolled-back transaction. Remove fsl_cx::cache::deltaIds as it was determined that the recursion involving that cache could have broken things (but, rather curiously, had not done so thus far). Might want to revisit that cache someday and account for the new effective tail of the array on each recursion. check-in: 35dca27909 user: stephan tags: trunk | |
|
2021-12-26
| ||
| 13:45 | Constness tweaks. check-in: 006fbfb455 user: stephan tags: trunk | |