Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 check-ins by user jan.nijtmans occurring around 1eb438d61a6db40d.
|
2013-10-14
| ||
| 10:54 | Restore some styling improvements. ... (check-in: ee598d69be user: jan.nijtmans tags: tkt-change-hook) | |
| 10:38 | Phase 4 refactoring broke some important requirements: 1) "fossil rebuild" should not run any hook script, nor should it read the hook script code from the database for each manifest it processes. 2) When handling tickets through the UI, the ticket hook should be processed once, just as if it came in through xfer. 3) (less important) Allow "http -asynchronious" to be shortened to "http -async" I apologize for having to revert part of the "Phase 4 refactoring" restoring this. Other than that, it looks good! ... (check-in: b92e111a59 user: jan.nijtmans tags: tkt-change-hook) | |
|
2013-10-13
| ||
| 15:19 | don't hardcode default value of "ticket-change" code ... (check-in: 0e1f94732c user: jan.nijtmans tags: tkt-change-hook) | |
| 15:13 | Put back ticket-change in JSON part as well ... (check-in: 12c53e1cb6 user: jan.nijtmans tags: tkt-change-hook) | |
| 15:03 | One more place where failing run_common_script() should not prevent running manifest_crosslink(). If run_common_script() fails, let the following run_script() output its error-message. ... (check-in: cc5466b8d4 user: jan.nijtmans tags: tkt-change-hook) | |
| 14:46 | add back the original "ticket-change" hook, as it was in trunk ... (check-in: abef3f1957 user: jan.nijtmans tags: tkt-change-hook) | |
| 10:33 | manifest_crosslink() should always be called, whether run_common_script() fails or not. revert experimental change in timeline ... (check-in: b6b656a0ec user: jan.nijtmans tags: tkt-change-hook) | |
| 10:06 | Fix redirect after ticket submission, as suggested by Justin Forest. Rename "ticket-change" to "xfer-ticket-script" everywhere ... (check-in: 26d6877ff6 user: jan.nijtmans tags: tkt-change-hook) | |
| 09:53 | merge trunk ... (check-in: a4327ba0b6 user: jan.nijtmans tags: tkt-change-hook) | |
|
2013-10-11
| ||
| 12:40 | Fix Win64 build with MinGW. Problem reported by Martin Gagnon. ... (check-in: 0d70a87234 user: jan.nijtmans tags: trunk) | |
| 11:44 | merge trunk ... (Closed-Leaf check-in: 1a310615b6 user: jan.nijtmans tags: fix-win64-mingw) | |
| 11:31 | Revert makefile changes in my previous commit: It broke the normal MinGW build! Sorry! ... (check-in: 6e55cfe61c user: jan.nijtmans tags: trunk) | |
| 11:20 | With this additional change, it works. But I don't understand why, and it means patching SQLite. ... (check-in: 57cf68c421 user: jan.nijtmans tags: fix-win64-mingw) | |
| 11:13 | Attempt to fix Win64 MinGW build. Doesn't work yet. ... (check-in: e25410039d user: jan.nijtmans tags: fix-win64-mingw) | |
| 08:36 | Make sure that EVERY source file includes "config.h" before any other system includes. This is the first step in making fossil compile for Win64 using MinGW. ... (check-in: c30cd93b91 user: jan.nijtmans tags: trunk) | |
|
2013-10-10
| ||
| 13:18 | Make sure that in Win64 builds _USE_32BIT_TIME_T is never defined. ... (check-in: d66cfb164f user: jan.nijtmans tags: trunk) | |
|
2013-10-09
| ||
| 09:20 | Make sure that the UUID of a generated P-card is the UUID of a valid check-in in this repository. If not, bail out early. ... (check-in: d30abe150a user: jan.nijtmans tags: trunk) | |
| 08:24 | Be carefull comparing a char < ' ': If the platform 'char' type is signed, this will be true for all UTF-8 characters >0x7f as well. ... (check-in: c426475eeb user: jan.nijtmans tags: trunk) | |
| 07:56 | Fix the abbreviated project name escape mechanism for two less obvious dangers: - spaces at the end of the filename should be stripped. - control charaters are not allowed in filenames either. ... (check-in: e161527c3a user: jan.nijtmans tags: trunk) | |
|
2013-10-08
| ||
| 15:21 | double-quote and question mark are a problem as well. ... (check-in: 760f80cd8f user: jan.nijtmans tags: trunk) | |
| 07:23 | Even better error message in case of "manifest file (12892) is malformed". Detection of this situation is not 100% failproof yet, but if zParentUuid=0 we know for sure that an illegal manifest will be generated later, so better catch this early. ... (check-in: 27dc7a5671 user: jan.nijtmans tags: trunk) | |
|
2013-10-07
| ||
| 08:35 | Hm. the db_finalize() and the newline append should be done within the "if"! ... (Closed-Leaf check-in: 6135c92e4b user: jan.nijtmans tags: pending-review) | |
| 08:10 | Prevent an invalid manifest to be generated if there is no parent manifest. ... (check-in: aa01de5d16 user: jan.nijtmans tags: pending-review) | |
| 07:41 | Better error message in case of "manifest file (12892) is malformed". It will now give an additional line: "line ???: wrong size UUID on P-card" (or whatever other parsing error happens), and using the "-n" option it will print out the complete manifest as well. This would have made it much easier to investigate Ron Aaron's commit problem (many thanks for reporting this!), without adding special debugging code to fossil. ... (check-in: ec81aee915 user: jan.nijtmans tags: trunk) | |
|
2013-10-04
| ||
| 09:59 | Cygwin64 has a 64-bit "long" while Win64 has a 32-bit "long". Correct for that. ... (check-in: 4ff370c832 user: jan.nijtmans tags: trunk) | |
| 09:30 | Better error-message in case of "manifest file is malformed" ... (Closed-Leaf check-in: 1eb438d61a user: jan.nijtmans tags: pending-review) | |
| 09:10 | Fix compiler warning on Cygwin ... (check-in: c05cc27300 user: jan.nijtmans tags: trunk) | |
|
2013-10-03
| ||
| 14:59 | Revert [b7625d8d8c], instead make sure the correct header file is included for Cygwin. ... (check-in: 19b2b2403c user: jan.nijtmans tags: trunk) | |
|
2013-10-02
| ||
| 14:45 | Take over some improvements/fixes from the CleanX2 branch. Still not very well tested, so please don't look ;-) ... (check-in: b2a2ce5312 user: jan.nijtmans tags: cleanX-no-clean-glob) | |
|
2013-10-01
| ||
| 09:21 | merge trunk ... (check-in: 5724c62cf6 user: jan.nijtmans tags: cleanX-no-clean-glob) | |
|
2013-09-27
| ||
| 07:49 | unloadTcl is only needed on Win32, not for Win64 or any other platform. ... (Closed-Leaf check-in: 0013d9ccb6 user: jan.nijtmans tags: win64-unloadTcl) | |
|
2013-09-26
| ||
| 07:17 | Fossil compiled with MinGW version 4.0 didn't work on Windows XP any more, because it uses the 64-bit time API. This can be fixed by specifying -D_USE_32BIT_TIME_T. (It's the same in MSVC, but Microsoft does not support Windows XP in its development tools any more. Adding the same in Makefile.msc doesn't help anything on that, unfortunately) ... (check-in: a8214df372 user: jan.nijtmans tags: trunk) | |
| 06:58 | Workaround for [http://sourceforge.net/p/mingw/bugs/2062/|MinGW bug #2062] ... (check-in: f2ce2f80f4 user: jan.nijtmans tags: trunk) | |
|
2013-09-25
| ||
| 08:36 | Remove unnecessary code which does not give any advantage. This fixes dynamical loading of Tcl 8.4 as well. ... (Closed-Leaf check-in: 132c1b4cc2 user: jan.nijtmans tags: pending-review) | |
| 08:29 | FOSSIL_ENABLE_TCL_PRIVATE_STUBS only makes sense when USE_TCL_STUBS is defined as well. Use that consistantly in the code. Easier testing whether Tcl integration works fine, just by "fossil version -v". ... (check-in: 0038f4c999 user: jan.nijtmans tags: pending-review) | |
|
2013-09-23
| ||
| 12:09 | Loading Tcl 8.4 only works when USE_TCL_EVALOBJV=1. Reason: the function Tcl_GetCommandFromObj is introduced in Tcl 8.5 (TIP #139). Actually, the source code history for Tcl indicates that this is not correct. Both Tcl_GetCommandFromObj and Tcl_GetCommandInfoFromToken are present for Tcl 8.4. The Tcl_GetCommandInfoFromToken function was added in TIP #32 and the Tcl_GetCommandFromObj function exists since the initial check-in in the Tcl repository, circa 1998. ... (Closed-Leaf check-in: 2234fabe76 user: jan.nijtmans tags: support-tcl84-stubs) | |
| 10:17 | Prevent a crash in fossil during exit, when a mingw-compiled (with dw2) Tcl version is still loaded. This is clearly a dw2 bug (see: [http://comments.gmane.org/gmane.comp.gnu.mingw.user/41724]), but the suggested workaround works and is managable. ... (check-in: da96f916cb user: jan.nijtmans tags: trunk) | |
|
2013-09-19
| ||
| 15:10 | Some fossil_panic() -> fossil_fatal(). Revert output value of manifest_crosslink() to what it was: 0=error. ... (check-in: b6cb91ca5b user: jan.nijtmans tags: tkt-change-hook) | |
| 14:42 | merge trunk ... (check-in: 06f08ed02b user: jan.nijtmans tags: tkt-change-hook) | |
| 11:48 | Fix MSVC build error and warnings: c:\localdata\workspace\fossil\src\util.c(249) : warning C4715: 'fossil_timer_fetch' : not all control paths return a value c:\localdata\workspace\fossil\src\util.c(270) : warning C4715: 'fossil_timer_reset' : not all control paths return a value ... (check-in: 6e1e4bb83d user: jan.nijtmans tags: trunk) | |
| 11:29 | Notepad and Cygwin interpret absolute paths differently. Solution: make the path relative. This is generally better anyway because relative paths are shorter and less likely to overflow platform path limitations. ... (check-in: 3cadf76c3d user: jan.nijtmans tags: trunk) | |
| 11:18 | Bug-fix in file_relative_name(): If filename is a super-directory of the current directory then the final slash is missing. Test-case: "fossil test-relative-name /home/ --slash". Result "../../.." should be "../../../". This bugfix is needed for my next commit. ... (check-in: add7524533 user: jan.nijtmans tags: trunk) | |
| 08:30 | Add --slash option to test-canonical-name and test-relative-name ... (check-in: e6ced76797 user: jan.nijtmans tags: trunk) | |
|
2013-09-17
| ||
| 15:05 | When using private stubs, really use private stubs ... (check-in: 07c8b73072 user: jan.nijtmans tags: trunk) | |
| 11:33 | Don't set USE_TCL_EVALOBJV=1 for Tcl 8.6b3 and higher by default. ... (check-in: 8c8c6bce21 user: jan.nijtmans tags: trunk) | |
| 09:00 | Use Tcl_EvalObjv in any Tcl version, not only when compiled with Tcl 8.6 headers. ... (Closed-Leaf check-in: 92fd8ddc13 user: jan.nijtmans tags: pending-review) | |
| 08:43 | new clean-zlib, clean-openssl and clean-tcl targets in the mingw makefiles. ... (check-in: 9112867c2d user: jan.nijtmans tags: pending-review) | |
|
2013-09-16
| ||
| 19:41 | If compat/tcl-8.6 contains an already built Tcl environment, allow it to be used by Fossil. ... (check-in: 0ad8af4bf6 user: jan.nijtmans tags: trunk) | |
| 12:49 | For some reason, fossil.rc could not be compiled any more. Anyway with stubs, the patchlevel doesn't say anything. ... (Closed-Leaf check-in: 7c232dc0c3 user: jan.nijtmans tags: tcl-stubs-v2) | |
| 12:16 | Add "clean" target as well. Remove noncence part. ... (check-in: 4966e15d04 user: jan.nijtmans tags: tcl-stubs-v2) | |