Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins
2023-12-08
| ||
15:30 | Limit the scope to the error condition to clone operations only for now. Leaf check-in: 29e9e84a1e user: andybradford tags: clone-resume | |
14:33 | Only output clone retry warning if there is another round to go. check-in: bdd22578fa user: andybradford tags: clone-resume | |
14:20 | The rebuilt string is only regenerated if the manifest changes, so use the mtime of when the rebuild is done instead. check-in: a92fddc023 user: andybradford tags: clone-resume | |
13:52 | If the clone is interrupted by the user this should constitute an error condition. check-in: 2c8ad9a417 user: andybradford tags: clone-resume | |
10:20 | In the HTML diff output, honour the user's light/dark mode browser setting. Leaf check-in: a2e6b31aa1 user: danield tags: trunk | |
2023-12-07
| ||
20:06 | Add the --dark option to "fossil diff --tk" check-in: 5a8516df59 user: drh tags: trunk | |
19:57 | Improvements to the diff algorithm. See [forum:/forumpost/515e0d43425d7164|forum thread 515e0d43425d7164] for discussion. check-in: a45c7f0a56 user: drh tags: trunk | |
10:39 | Make the dark mode also available for the diff HTML output (--webpage, -b). Closed-Leaf check-in: 6b7fe7fa95 user: danield tags: tk-diff-darkmode | |
2023-12-05
| ||
16:02 | Provide a dark mode for the TCL/TK-based diff GUI. check-in: b59a42f1f6 user: danield tags: tk-diff-darkmode | |
14:12 | Clean up signal includes. check-in: e6da7f08c8 user: andybradford tags: clone-resume | |
2023-12-03
| ||
21:36 | The th1-expr-3 test revealed that Th_SetResultInt() could produce incorrect result for INT_MIN. Ensure that -1*iVal is positive for all values and remove unsiged casts. Seems a common optimisation problem with gcc on various verisons from 10 through 13 on Linux, NetBSD and OmniOS on various architectures. Leaf check-in: 4a98d2aae1 user: preben tags: th-int-min-errors | |
20:28 | Include termio.h to determine terminal width on OmniOS. Uses __EXTENSIONS__ in line with detection of solaris like systems in auto.def. Leaf check-in: 32428f126f user: preben tags: omnios-terminal-width | |
14:06 | In auto.def, set CFLAGS and LDFLAGS with LIBs as further tests may need them. Leaf check-in: adcbb25256 user: preben tags: auto-def-compilation-flags | |
2023-12-02
| ||
18:30 | Use a new pragma requested only in the first clone cycle to track the rebuilt time on the remote server to be able to error out if the rids are rebuilt before resuming a clone. This is currently just a fossilized string but it could instead be a hash of the components to make it more like a "code". Requesting the pragma only at the beginning does not account for the possibility that a rebuild will happen after the clone begins. I find that scenario highly unlikely, but grant that it might be possible. If the remote server does not yet support the pragma the client falls back to using the remote server-code. check-in: 7f116a437c user: andybradford tags: clone-resume | |
2023-12-01
| ||
22:37 | Typo corrections in comments only. No change in functionality. check-in: a8b6fdc83e user: andybradford tags: trunk | |
21:22 | Just write the value the first time it is seen since there is no need to have the value during the sync (yet). check-in: 15b166df9b user: andybradford tags: clone-resume | |
21:14 | Prevent resuming a clone from a different server by preserving the remote server-code and comparing on resume. This is a crude heuristic to determine whether or not it is likely that the remote server still has the same rids. A better heuristic may be to probe the server for known rids that have been received (e.g. each clone seqno that is sent by the server could be recorded and then probed with a pragma request by the client later on). check-in: aa54ea4116 user: andybradford tags: clone-resume | |
20:31 | Prevent the resuming of a clone that does not have the product code. check-in: 63af8da355 user: andybradford tags: clone-resume | |
20:17 | Minor style change. check-in: d82079546d user: andybradford tags: clone-resume | |
20:12 | Skip all finalizing until the clone is complete. check-in: 9fa25cd746 user: andybradford tags: clone-resume | |
17:55 | Try to handle some error conditions better. First, if the clone's first sync didn't bring down any artifacts there isn't likely any point in continuing. Second, if on resuming a clone no artifacts were received, abort. Third, check if the project-code has changed earlier (this might be detected even earlier when the client receives the "push" card). check-in: 17f3408f6b user: andybradford tags: clone-resume | |
04:12 | Minor correction in comment. No change in functionality. check-in: 4fdc41bd05 user: andybradford tags: clone-resume | |
03:58 | Shorten warning message to fit in a terminal without wrapping. check-in: 5bb6d26187 user: andybradford tags: clone-resume | |
03:53 | Verify that the project-code on resume hasn't changed per discussion in [forum:d1a2ed56271602fa|forum post d1a2ed56271602fa] and rollback changes if they are different. check-in: ee5b942080 user: andybradford tags: clone-resume | |
2023-11-30
| ||
06:26 | Defer rebuilding the database until the clone is complete. This prevents a lot of unnecessary rebuilding while there aren't enough artifacts to make it worth it. check-in: 333a89bf96 user: andybradford tags: clone-resume | |
04:01 | Catch SIGINT during sync so that when pressing ctrl-c during a clone of a large repository it doesn't dispose of all the progress that was made during, thus enabling resume from an user interrupt. When the handler returns, the last sync will complete and then exit the loop. check-in: ad2e148541 user: andybradford tags: clone-resume | |
2023-11-29
| ||
04:32 | Merge in latest from trunk. check-in: 5d12d42680 user: andybradford tags: clone-resume | |
2023-11-28
| ||
22:05 | Tiny documentation fix: an extra space before argument to the '--base-rcvid' hook test option made the argument appear on the line describing the option. check-in: cb283ca505 user: km tags: trunk | |
19:31 | Use signed char for two json timeline functions that may return -1. Leaf check-in: c42dee8634 user: preben tags: signed-chars | |
14:58 | Fix sort order in json_page_finfo() on systems where char is unsigned. check-in: fd288be356 user: preben tags: signed-chars | |
2023-11-27
| ||
10:28 | Fix typos in unvers.wiki. check-in: 22ee43db43 user: danield tags: trunk | |
06:19 | Take advantage of the fact that the seqno is recorded and automatically resume up to three times if failure is encountered. check-in: bc0a4c60c0 user: andybradford tags: clone-resume | |
06:14 | Check for valid value from return of function. check-in: 2f0aac8c01 user: andybradford tags: clone-resume | |
2023-11-26
| ||
19:07 | Fix previous commit: Just fixing amend breaks subsequent tests. As tests should not need input, move the code to the test/tester.tcl. Leaf check-in: 3b37f7d97b user: preben tags: test-fixes-2.24 | |
18:18 | The amend tests may fail on narrow terminals on systems where the terminal size can be determined when tclsh is running (for me, all but Windows). Closing stdin causes fossil not to wrap output. To avoid fossil complaining about file descriptor zero (which is now used on some systems), temproarily open the script readonly. check-in: 1cb0d57db1 user: preben tags: test-fixes-2.24 | |
17:52 | Fix merge_renames-13-3 on macOs as the expected a filename under /tmp but the result was under the normalized /private/tmp directory. check-in: f044437634 user: preben tags: test-fixes-2.24 | |
16:48 | Fix json tests that no longer reflects default capabilities for anonymous. check-in: 9ba74ae04e user: preben tags: test-fixes-2.24 | |
2023-11-25
| ||
16:42 | Update [4f9cfebf18]: indent the verbatim merge conflict markers in the merge1.test script to avoid check-in warnings for this branch. check-in: 3fc08b7633 user: florian tags: test-fixes-2.24 | |
15:56 | Remove description of unsupported option that was removed with [298ccff06c487d035a7] from "fossil all rebuild" command. No change in functionality. check-in: 33fa915508 user: andybradford tags: trunk | |
2023-11-24
| ||
23:51 | Do away with the --resume flag as it is mostly redundant and can be assumed by the presence of the recorded clone_seqno. Suggested by Warren Young in [forum:4bcbb41aab2538d4|forum post 4bcbb41aab2538d4]. check-in: b0a60d8f1d user: andybradford tags: clone-resume | |
22:14 | Need to reset the server-code so that the user password is correct. Might as well also reset the project-code. check-in: 61e0ced9bf user: andybradford tags: clone-resume | |
21:39 | Implement the ability to resume a clone that has failed. This is a variation on the attempt made in [ec26471439] that was never completed. The significant difference in the use of the "clone protocol" which uses cfile cards to complete the synchronization rather than the "sync protocol" which takes much longer using the file card. check-in: ee710cc171 user: andybradford tags: clone-resume | |
12:59 | Update the built-in SQLite to version 3.44.2. check-in: 3f97785608 user: drh tags: trunk | |
2023-11-22
| ||
20:22 | Update the built-in SQLite to version 3.44.1. check-in: 2edeeee4a8 user: drh tags: trunk | |
2023-11-20
| ||
19:16 | On the activity report page, make sure the graph bar with the expected changes for the current week/month is not displayed on a separate line. check-in: 698dc59ec1 user: florian tags: trunk | |
18:57 | Do not separate timeline entries by an empty line when using custom formatting strings. check-in: 256983fdd5 user: danield tags: trunk | |
17:07 | Add a button to review a list of hashes to shun. As discussed in [forum:1d54d9965c7e4c00], fossil may accept 40 byte abbreviated hashes from command output, when the intention is to shun the real 64 byte hash. Leaf check-in: d5c8235cdc user: preben tags: shun-list-review | |
2023-11-19
| ||
11:42 | Improved description on /timeline pages that make use of tl= or rl=. check-in: e03200e545 user: drh tags: trunk | |
2023-11-18
| ||
20:41 | Add new tl= and rl= query parameters to /timeline check-in: 2288b74257 user: drh tags: trunk | |
2023-11-17
| ||
18:05 | Update th1-tcl tests to reflect change from XML style closing hr/ tag to just hr in HTML from change [f5482a0a79]. check-in: 1c637a4985 user: preben tags: test-fixes-2.24 | |