summaryrefslogtreecommitdiffstats
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
...
| | * Add comments, remove unwanted whitespace changePetr Wolf2012-02-102-1/+11
| | |
| * | Fix tabs, typosPetrWolf2012-02-142-5/+5
| |/
| * Response filesunknown2012-02-0914-9/+301
| |
* | Merge pull request #218 from tfarina/fix-graph-todoEvan Martin2012-02-223-7/+6
|\ \ | | | | | | Fix the TODO in graph.h.
| * | fix depfile_ accessThiago Farina2012-02-161-1/+1
| | |
| * | Fix the TODO in graph.h.Thiago Farina2012-02-162-6/+5
| | | | | | | | | | | | Signed-off-by: Thiago Farina <tfarina@chromium.org>
* | | Merge pull request #223 from pcc/restat-clean-manifestEvan Martin2012-02-221-1/+6
|\ \ \ | | | | | | | | Do not reload the manifest if a restat cleans it while being rebuilt
| * | | Do not reload the manifest if a restat cleans it while being rebuiltPeter Collingbourne2012-02-211-1/+6
| |/ /
* | | Add a test for empty path in CanonicalizePath().Nicolas Despres2012-02-201-0/+3
|/ / | | | | | | | | | | | | This test covers bug fix introduced by 62e9139740. However, reverting this patch does not trigger a test failure. Maybe, I am not testing on the right platform (Linux). Anyway, in all cases I think this test deserves to be added.
* | Merge pull request #176 from pcc/exit-cleanupEvan Martin2012-02-149-111/+360
|\ \ | | | | | | Implement cleanup-on-interrupt
| * | If a command fails, wait for all running commands to terminate before we doPeter Collingbourne2012-02-044-21/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously, if a command fails, the fate of the other child processes running in parallel was inadequately controlled. On POSIX platforms, the processes were orphaned. Normally they would run to completion, but were liable to being killed by a SIGPIPE. On Windows, the child processes would terminate with the parent. The cleanup-on-interrupt patch caused the SubprocessSet and Builder destructors to clean up after themselves by killing any running child processes and deleting their output files, making the behaviour more predictable and consistent across platforms. If the build is interrupted by the user, this is correct behaviour. But in the case where the build is stopped by a failed command, this would be inconsistent with user expectations. In the latter case, we now let any remaining child processes run to completion before leaving the main loop in Builder::Build.
| * | Implement cleanup-on-interruptPeter Collingbourne2012-02-048-84/+318
| | | | | | | | | | | | | | | | | | | | | | | | This causes us to clean up by deleting any output files belonging to currently-running commands before we quit if we are interrupted (either by Ctrl-C or by a command failing). Fixes issue #110.
| * | Clean up how the build line is managed at the end of the buildPeter Collingbourne2012-02-041-6/+20
| | |
* | | graphviz: don't draw edges multiple timesEvan Martin2012-02-142-3/+9
| | | | | | | | | | | | From a patch from Ian Godin <iangodin@gmail.com>.
* | | Made graph horizontal (better for filenames)Ian Godin2012-02-141-0/+1
| |/ |/|
* | Merge pull request #214 from qhuo/onpipeready-fixEvan Martin2012-02-042-2/+4
|\ \ | | | | | | Avoid using undefined value in Subprocess::OnPipeReady()
| * | Avoid using undefined value in Subprocess::OnPipeReady()Qingning Huo2012-02-042-2/+4
| |/
* | Merge pull request #213 from syntheticpp/evaluate-only-onceEvan Martin2012-02-041-3/+2
|\ \ | | | | | | Evaluate command only once
| * | compute command twice only in case of an errorPeter Kuemmel2012-02-041-3/+2
| |/
* | don't crash when CanonicalizePath removes all path componentsEvan Martin2012-02-042-0/+22
|/ | | | From a patch from Peter Kuemmel <syntheticpp@gmx.net>.
* Merge pull request #199 from qhuo/create-processEvan Martin2012-01-292-2/+38
|\ | | | | Mark CreateProcess "program not found" failure as non-fatal
| * Add a test, NoSuchCommand.Qingning Huo2012-01-251-0/+18
| |
| * Mark CreateProcess "program not found" failure as non-fatalQingning Huo2012-01-201-2/+20
| |
* | parse $:Peter Kuemmel2012-01-295-4/+78
| | | | | | | | | | '$:' is a valid string now, it expands to ':' update error messages and show a hint when something went wrong.
* | add colon escapingPeter Kuemmel2012-01-293-23/+37
| | | | | | | | | | | | | | | | Needed for Windows drive names. For instance configure with gtest: python configure.py --with-gtest=c$:\gtest-1.6.0
* | Fix cygwin build by including stdio.h hereokuoku2012-01-252-1/+1
| |
* | rearrange query/browse output to be more sensibleEvan Martin2012-01-242-71/+100
| |
* | Merge pull request #204 from syntheticpp/parser-crashEvan Martin2012-01-231-1/+4
|\ \ | | | | | | Fix parser crash
| * | check access to first element of string: string could be emptyPeter Kuemmel2012-01-231-1/+4
| | |
* | | fix search&replace error on struct namePeter Kuemmel2012-01-231-1/+1
| | |
* | | no error code about too long file names on WindowsPeter Kuemmel2012-01-231-0/+6
|/ /
* | only msvc needs these workaroundsPeter Kuemmel2012-01-222-2/+4
| |
* | windows: use _WIN32 define everywhereEvan Martin2012-01-229-21/+21
|/ | | | Rather than mixing use of WIN32 and _WIN32.
* Strip ansi escape sequences from subcommand output when not writing to a ↵Nico Weber2012-01-194-2/+68
| | | | smart terminal.
* Add a test that dry run shows all commands that could be run (noneFrances Buontempo2012-01-162-1/+37
| | | | cleaned) and a fix for this
* Stop warning about chdir etc which are treated as errors on MSVC with theFrances Buontempo2012-01-162-0/+2
| | | | current settings
* BuildLog: Use Log::insert(Log::value_type()) to avoid invalid strings.Evan Jones2012-01-131-2/+2
| | | | | | | The MSVC std::string implementation copies strings, so using make_pair resulted in a pointer to invalid memory. This ensures the insert uses a StringPiece without an intermediate std::string copy, so the correct pointer ends up in the hash_map.
* Merge branch 'stringpiece-hash'Evan Martin2012-01-127-49/+94
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This branch changed the fundamental key used in hashes in Ninja from char* to StringPiece. Using StringPiece allows us to eliminate some copies and null termination in various places. While I was at it, I switched the hash function to the Murmur hash. Seems to be ~100ms faster: evmar:/work/chrome/src$ ~/projects/ninja/misc/measure.py ~/projects/ninja/ninja-trunk -C out/Debug chrome sampling: 994ms 1010ms 1000ms 999ms 1019ms 994ms 1002ms 993ms 997ms 1040ms estimate: 993ms (mean err 11.6ms) evmar:/work/chrome/src$ ~/projects/ninja/misc/measure.py ~/projects/ninja/ninja -C out/Debug chrome sampling: 888ms 895ms 891ms 886ms 896ms 888ms 891ms 903ms 895ms 888ms estimate: 886ms (mean err 5.7ms)
| * decide on murmur hash, delete stl hashEvan Martin2012-01-121-18/+4
| | | | | | | | | | | | | | A nice thing about working on Google: a C++ expert who happened to be writing a proposal on hashing for the next C++ version wandered into my office. He seemed to think using Murmur here is fine.
| * make new hash work on windowsEvan Martin2012-01-111-4/+14
| |
| * back to murmur hashEvan Martin2012-01-091-7/+7
| |
| * remove string copy while loading depfilesEvan Martin2012-01-092-3/+3
| |
| * split canonicalizeEvan Martin2012-01-092-5/+14
| |
| * more stringpieceEvan Martin2012-01-093-9/+12
| |
| * murmur -> stl hashEvan Martin2012-01-091-1/+10
| |
| * switch node lookup to StringPieceEvan Martin2012-01-092-5/+5
| |
| * convert ExternalStringHash to use StringPieceEvan Martin2012-01-093-22/+50
| |
* | windows: hash_map bucket count has different getterFrances Buontempo2012-01-111-1/+6
| |
* | GetTempPath() needs windows.hFrances Buontempo2012-01-111-0/+4
|/
* include main hash load in -d stats outputEvan Martin2012-01-091-1/+8
|
generated by cgit v0.12 at 2024-08-23 16:07:53 (GMT)