| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
| |
Fixes #1411.
|
|
|
|
|
|
|
|
| |
Borrow macro implementation from OpenSSL code.
Add the macro after each fallthrough switch case to indicate our
intention to the compiler. This silences GCC -Wimplicit-fallthrough
warnings, which is implied by GCC 7.x -Wextra.
|
| |
|
|
|
|
| |
This prevents overflow on Windows where 'long' is not 64-bit.
|
|
|
|
| |
This prepares it for higher-resolution timestamps.
|
|
|
|
|
|
|
|
|
|
| |
https://groups.google.com/forum/#!msg/ninja-build/YQuGNrECI-4/ti-lAs9SPv8J
discusses a case where an rule updates its output file and then
fails. The next run of ninja considers the ouptut file clean
and doesn't rebuild it. Always stat output files after they are
built, and write the mtime into .ninja_log. Consider output files
dirty if the recorded mtime is older than the most recent input
file.
|
|
|
|
|
| |
On AIX, inttypes.h gets indirectly included by build_log.h.
It's easiest just to ask for the printf format macros right away.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Recompacting the build log used to be slow, so it made sense to print this
message. We then made recompaction much faster, but didn't remove this
message back then.
The deps log only has it because the build log had it.
Since both steps are effectively instant in practice, remove these log messages.
|
|
|
|
| |
Now tests don't print anything. Non-test behavior is unchanged.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
For .ninja_deps, remove objects that have no in-edges or whose in-edges
do not have a "deps" attribute. (This matches the behaviour of `-t
deps`).
BuildLog doesn't know about state, so let its recompact method take
delegate that decides is a path is life or not, and implement it in
NinjaMain.
|
| |
|
|
|
|
| |
Like f6f00aa40f0c541df06, but for the build log instead of the deps log.
|
|\
| |
| | |
Add a 'recompact' tool, which forces recompaction of the build and deps ...
|
| |
| |
| |
| | |
This is useful for performance comparisons between two build directories.
|
|/
|
|
| |
architectures like OpenBSD/mips64el and OpenBSD/hppa64.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
(Committing this on top of b56fe80 since they're related,
but I may end up reverting both.)
|
| |
|
|
|
|
|
|
| |
This reverts commit 904c9610fe66c4f4bd63a07d6f057c8603d24394.
The commit caused issue #380, this revert fixes it. The revert
also makes the test from the previous commit pass.
|
|\
| |
| | |
Make StringPiece data members private.
|
| |
| |
| |
| | |
Signed-off-by: Thiago Farina <tfarina@chromium.org>
|
| |
| |
| |
| | |
Not a real problem, mainly to make valgrind happy
|
|/ |
|
|
|
|
|
|
| |
Nico convinced me that it's safe to do this without revving the build
log format: the old decimal values will still parse as hex (just as
different numbers) and cause a superfluous rebuild.
|
|
|
|
|
| |
Log format 4 was introduced in January.
There's been plenty of time to migrate.
|
|
|
|
| |
Fixes a warning.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
.build_log load time 350ms -> 17ms, filesize 197MB -> 1.6MB on
Mac. On Windows, it's 500ms -> 20ms.
Makes the build log a lot less useful for scripts, but there could
be a tool for use cases that need log information. A prototype of
such a tool is in https://github.com/nico/ninja/commit/1b243d311
The hash function is 64bit murmurhash2. Assuming that that different
commands get the same hash only by chance, it's is very unlikely
for two different commands to hash to the same value with a 64bit
hash.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fgets() needs to search for a \n in the input, and LineReader's
strstr() did the same work again. By using fread() directly, this
duplicate work can be saved.
(I tried using readline() / fgetln() for the same saving, but those
functions are not standard, and were slower than the code in this CL.)
Results from running build_log_perftest, on OS X:
Before: min 369ms max 372ms avg 370.4ms
After: min 243ms max 247ms avg 244.8ms
On Windows:
Before: min 796ms max 904ms avg 858.0ms
After: min 359ms max 375ms avg 371.4ms
On Linux:
Before: min 161ms max 169ms avg 164.8ms
Before: min 130ms max 137ms avg 132.6ms
|
|
|
|
| |
change.
|
| |
|
|\
| |
| | |
Add .ninja_log load time to metrics.
|
| |
| |
| |
| | |
On my system, it takes 22% of the empty build time for chrome.
|
|/ |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
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.
|
| |
|
| |
|