Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
42 check-ins using file client/fnc/fnc.c version 1401b4a29f
|
2021-12-14
| ||
| 15:52 | More work on the merge port and adjacent code. check-in: b9c2a05cfe user: stephan tags: trunk | |
| 11:19 | Added fsl_file_copy(), fsl_symlink_read(), and fsl_file_isexec() as part of the merge op effort. Expanded fsl_file_tempname() to be able to accept a NULL list of directories in order to return just the name without a directory prefix. check-in: 6dca54dfde user: stephan tags: trunk | |
| 09:16 | Minor allocation optimization and tweak to how the filename prefix is applied in fsl_file_tempname(). check-in: b89b9e0058 user: stephan tags: trunk | |
| 08:55 | Ported in a variant of the temp-filename generation bits from fossil because it's needed for the next merge steps. check-in: dd3acff665 user: stephan tags: trunk | |
| 05:23 | More work on the merge API port. Still lots to do. check-in: 7d626b574d user: stephan tags: trunk | |
| 04:17 | Tweaks to the return semantics of fsl_db_exec() and fsl_stmt_step(). Doesn't/shouldn't affect any existing code. Added a test to ensure that fsl_stmt_step() behaves as expected in the face of a RETURNING clause. check-in: ad7ef26207 user: stephan tags: trunk | |
|
2021-12-13
| ||
| 06:15 | Minor code modernization in the still-far-from-complete merge code. check-in: d77e495227 user: stephan tags: trunk | |
| 03:52 | Merge in port-merge branch for further merge development - keeping this in a separate branch is unnecessary and is getting annoying. There's still plenty of work to do on the merge port. check-in: 63fe74d7f3 user: stephan tags: trunk | |
|
2021-12-11
| ||
| 07:34 | Minor internal cleanups in f-repostat.c check-in: 5a99011c5b user: stephan tags: trunk | |
|
2021-12-10
| ||
| 12:11 | As a special case, fsl_file_simplify_name() now always resolves to '/' if its input resolves to that, regardless of whether or not it is told to retain a trailing slash. The alternative would be to resolve to an empty string if it were told to strip the trailing slash for that case. check-in: 3ad71dbc66 user: stephan tags: trunk | |
| 11:35 | Fixed bug which caused fsl_file_simplify_name() to not honor its trailing backslash policy. check-in: 14fffb4e82 user: stephan tags: trunk | |
| 07:01 | Typo fix. check-in: 2b2b8c3ac9 user: stephan tags: trunk | |
| 06:49 | Wrapped f-repostat work in a transaction to avoid a minuscule corner case where another app changes the stats as they're being read. Made use of the percent-z fsl_appendf() specifier to remove a few explicit fsl_free() calls. check-in: 48d30b38a6 user: stephan tags: trunk | |
| 04:32 | Added missing status and quick-check to f-repostat. Missing only full integrity check (requires porting (fossil test-integrity)). check-in: a6ba4b4444 user: stephan tags: trunk | |
| 04:08 | Initial version of f-repostat, analog to (fossil dbstat). Some adjecent constness cleanups. check-in: 4926ae9b31 user: stephan tags: trunk | |
|
2021-12-08
| ||
| 08:17 | Minor doc updates. Changed the main db journal mode to TRUNCATE because OFF apparently leads to undefined behaviour when ROLLBACK is used (something we do often, e.g. to implement dry-run mode). check-in: a1b9594aec user: stephan tags: trunk | |
| 07:04 | Cherrypicked inadvertent checkin to the wrong branch via /fileedit. check-in: 6dddf8537b user: stephan tags: trunk | |
| 07:02 | Minor doc touchups Closed-Leaf check-in: e9c9fccc7f user: stephan tags: port-merge | |
| 06:55 | Merged in trunk. Kinda funny that f-ci can check in a merge but we don't yet have f-merge to actually create a merge. check-in: 49e14934d8 user: stephan tags: port-merge | |
|
2021-12-06
| ||
| 03:42 | Ported in line numbers in merge conflict markers from [fossil:14f44e933c0c0c2c | fossil checkin 14f44e933c0c0c2c]. check-in: 3e58eea1f3 user: stephan tags: trunk | |
| 02:33 | Minor help text and API doc cleanups. One minor allocation optimization in fsl_repo_rebuild(). check-in: 17a7f5cf12 user: stephan tags: trunk | |
|
2021-12-04
| ||
| 07:39 | Internal restructuring of crosslink callback handling, such that registered listeners are called at a later phase for ticket artifacts. This should allow client-side listeners to overwrite timeline entries for tickets without too much hassle (that wasn't possible at all before this change). check-in: 8349911eb6 user: stephan tags: trunk | |
| 06:36 | (Finally) implemented ticket crosslinking. check-in: d8e8c16cc4 user: stephan tags: trunk | |
|
2021-12-03
| ||
| 09:33 | s2 binding: extended Fossil.Db.each() to take a target array as a form of callback, appending each row to the target. check-in: 1c591b1ffc user: stephan tags: trunk | |
| 09:03 | More tinkering with the s2 binding, getting older scripts working again. check-in: 5ea356ab81 user: stephan tags: trunk | |
| 08:36 | Minor internal cleanups and costness tweaks. check-in: 8541d31570 user: stephan tags: trunk | |
| 07:39 | Tinkering with the s2 binding to get tests working after removal of the various db handles. check-in: 93a66791fe user: stephan tags: trunk | |
| 06:52 | SQL UDFs fsl_content() and fsl_sym2rid() now match any blob type, not just checkins. That they did not already do so was a bug. check-in: 624a9a175a user: stephan tags: trunk | |
| 06:23 | Moved most of the fsl_cx-specific code out of fsl_db and into the new src/udf.c. check-in: 84a59c6833 user: stephan tags: trunk | |
| 05:15 | Drastically simplify the s2 fsl_cx db mappings, exposing only the main db handle instead of separate repo/checkout/config handles. It was severely over-engineered and this approach will allow us to more easily eliminate the middle-man main db handle in the lib if we end up choosing to do so. check-in: 8ac418df2e user: stephan tags: trunk | |
| 04:49 | Removed a large swath of assigned-but-unused code in the v1 HTML sbs diff generator, as detected by clang 13. Reported by Mark via /chat. check-in: 5a433bf878 user: stephan tags: trunk | |
| 04:03 | Removed an unused var from fsl_delta_create2(), reported by clang 13 via Mark. Worked around what appears to be a clang 13 bug in (mis)detecting another var as unused. check-in: 351cd03fa3 user: stephan tags: trunk | |
| 03:52 | Experimentally change the journal_mode to OFF for the main (temp) DB, noting that we don't actually store anything in that db handle - it's only used as an ATTACH point for the other dbs. Added a note about being able to rename the MAIN db, which might be useful in eliminating the current middle-man MAIN db (barring one corner case for the global config db). check-in: 76d7f8f4eb user: stephan tags: trunk | |
|
2021-11-30
| ||
| 22:40 | Changed several internal temp tables to use (create if not exists/delete) instead of (create/drop) in the hope of proactively avoiding locking issues. This will slightly increase long-term memory usage for certain long-running apps but should ease locking pains for those same apps. check-in: cbaafebf22 user: stephan tags: trunk | |
| 18:13 | Internal fsl_db cleanups related to [17cc186e16351f5]. No public API changes. check-in: ddc1efaeb8 user: stephan tags: trunk | |
| 17:24 | Added fsl_db-level internal infrastructure to be able to catch the a potential OOM worked around in [11a6a98c81d1] and report it properly. Updated that fix to use the new infrastructure and distinguish between real and fake OOM. check-in: 17cc186e16 user: stephan tags: trunk | |
| 16:35 | Fix bug in annotate logic when file has insufficient history. When the file being annotated has less than the minimum three versions, one of the passes returns a null from the SQL query, which gets falsely reported as an OOM due to an empty strdup(). Check for this case before passing to strdup() and return. check-in: 11a6a98c81 user: mark tags: trunk | |
| 15:43 | Added fsl__cx_clear_cached_stmt() to forcibly clear ALL cached fsl_stmt handles held by a fsl_cx and use that from fsl_repo_rebuild() to ensure that the cache is cleared before beginning work. This is a hypothetical/proactive fix for a problem which has not yet been seen in the wild but seems very likely, given recent locking issues reported by Mark. Branching this for now because i cannot reproduce similar locking problems which fsl_repo_rebuild() should hypothetically trigger without this addition. Leaf check-in: f9d39435cf user: stephan tags: clear-cached-stmt | |
|
2021-11-29
| ||
| 15:06 | Don't drop and create fsl_compute_ancestors if it already exists. To workaround some locking issues in fnc, don't always drop and recreate the temp fsl_compute_ancestors table; instead, check for existence and only create when needed. Also, as per /chat discussion with stephan@, if the table exists, DELETE FROM to ensure we don't mix state between versions. check-in: 009bc1ff55 user: mark tags: trunk | |
| 06:52 | Fix comparison typo that clobbered requested version. The fsl_card_F_ckout_mtime() doc suggests that the current checkout will only be used if the rid of the passed in version is 0; however, the typo forced all rids greater than or equal to 0 to be overridden with the checkout version. check-in: 0031d2f466 user: mark tags: trunk | |
|
2021-11-23
| ||
| 20:10 | f-ciwoco: do not require any F-cards if the deck is explicitly a new root DAG entry. check-in: 35c58dd98d user: stephan tags: trunk | |
| 17:11 | Latest upstream fnc. check-in: cc967df182 user: stephan tags: trunk | |