The object traversal using reachability bitmap done by
"pack-object" has been tweaked to take advantage of the fact that
using "boundary" commits as representative of all the uninteresting
ones can save quite a lot of object enumeration.
* tb/pack-bitmap-traversal-with-boundary:
pack-bitmap.c: use commit boundary during bitmap traversal
pack-bitmap.c: extract `fill_in_bitmap()`
object: add object_array initializer helper function
Clang's sanitizer implementation seems to work better than GCC's.
* jk/ci-use-clang-for-sanitizer-jobs:
ci: drop linux-clang job
ci: run ASan/UBSan in a single job
ci: use clang for ASan/UBSan checks
When we started running sanitizers in CI via 1c0962c0c4 (ci: add address
and undefined sanitizer tasks, 2022-10-20), we ran them as two separate
CI jobs, since as that commit notes, the combination "seems to take
forever".
And indeed, it does with gcc. However, since the previous commit
switched to using clang, the situation is different, and we can save
some CPU by using a single job for both. Comparing before/after CI runs,
this saved about 14 minutes (the single combined job took 54m, versus
44m plus 24m for ASan and UBSan jobs, respectively). That's wall-clock
and not CPU, but since our jobs are mostly CPU-bound, the two should be
closely proportional.
This does increase the end-to-end time of a CI run, though, since before
this patch the two jobs could run in parallel, and the sanitizer job is
our longest single job. It also means that we won't get a separate
result for "this passed with UBSan but not with ASan" or vice versa).
But as 1c0962c0c4 noted, that is not a very useful signal in practice.
Below are some more detailed timings of gcc vs clang that I measured by
running the test suite on my local workstation. Each measurement counts
only the time to run the test suite with each compiler (not the compile
time itself). We'll focus on the wall-clock times for simplicity, though
the CPU times follow roughly similar trends.
Here's a run with CC=gcc as a baseline:
real 1m12.931s
user 9m30.566s
sys 8m9.538s
Running with SANITIZE=address increases the time by a factor of ~4.7x:
real 5m40.352s
user 49m37.044s
sys 36m42.950s
Running with SANITIZE=undefined increases the time by a factor of ~1.7x:
real 2m5.956s
user 12m42.847s
sys 19m27.067s
So let's call that 6.4 time units to run them separately (where a unit
is the time it takes to run the test suite with no sanitizers). As a
simplistic model, we might imagine that running them together would take
5.4 units (we save 1 unit because we are no longer running the test
suite twice, but just paying the sanitizer overhead on top of a single
run).
But that's not what happens. Running with SANITIZE=address,undefined
results in a factor of 9.3x:
real 11m9.817s
user 77m31.284s
sys 96m40.454s
So not only did we not get faster when doing them together, we actually
spent 1.5x as much CPU as doing them separately! And while those
wall-clock numbers might not look too terrible, keep in mind that this
is on an unloaded 8-core machine. In the CI environment, wall-clock
times will be much closer to CPU times. So not only are we wasting CPU,
but we risk hitting timeouts.
Now let's try the same thing with clang. Here's our no-sanitizer
baseline run, which is almost identical to the gcc one (which is quite
convenient, because we can keep using the same "time units" to get an
apples-to-apples comparison):
real 1m11.844s
user 9m28.313s
sys 8m8.240s
And now again with SANITIZE=address, we get a 5x factor (so slightly
worse than gcc's 4.7x, though I wouldn't read too much into it; there is
a fair bit of run-to-run noise):
real 6m7.662s
user 49m24.330s
sys 44m13.846s
And with SANITIZE=undefined, we are at 1.5x, slightly outperforming gcc
(though again, that's probably mostly noise):
real 1m50.028s
user 11m0.973s
sys 16m42.731s
So running them separately, our total cost is 6.5x. But if we combine
them in a single run (SANITIZE=address,undefined), we get:
real 6m51.804s
user 52m32.049s
sys 51m46.711s
which is a factor of 5.7x. That's along the lines we'd hoped for!
Running them together saves us almost a whole time unit. And that's not
counting any time spent outside the test suite itself (starting the job,
setting up the environment, compiling) that we're no longer duplicating
by having two jobs.
So clang behaves like we'd hope: the overhead to run the sanitizers is
additive as you add more sanitizers. Whereas gcc's numbers seem very
close to multiplicative, almost as if the sanitizers were enforcing
their overheads on each other (though that is purely a guess on what is
going on; ultimately what matters to us is the amount of time it takes).
And that roughly matches the CI improvement I saw. A "time unit" there
is more like 12 minutes, and the observed time savings was 14 minutes
(with the extra presumably coming from avoiding duplicated setup, etc).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When reachability bitmap coverage exists in a repository, Git will use a
different (and hopefully faster) traversal to compute revision walks.
Consider a set of positive and negative tips (which we'll refer to with
their standard bitmap parlance by "wants", and "haves"). In order to
figure out what objects exist between the tips, the existing traversal
in `prepare_bitmap_walk()` does something like:
1. Consider if we can even compute the set of objects with bitmaps,
and fall back to the usual traversal if we cannot. For example,
pathspec limiting traversals can't be computed using bitmaps (since
they don't know which objects are at which paths). The same is true
of certain kinds of non-trivial object filters.
2. If we can compute the traversal with bitmaps, partition the
(dereferenced) tips into two object lists, "haves", and "wants",
based on whether or not the objects have the UNINTERESTING flag,
respectively.
3. Fall back to the ordinary object traversal if either (a) there are
more than zero haves, none of which are in the bitmapped pack or
MIDX, or (b) there are no wants.
4. Construct a reachability bitmap for the "haves" side by walking
from the revision tips down to any existing bitmaps, OR-ing in any
bitmaps as they are found.
5. Then do the same for the "wants" side, stopping at any objects that
appear in the "haves" bitmap.
6. Filter the results if any object filter (that can be easily
computed with bitmaps alone) was given, and then return back to the
caller.
When there is good bitmap coverage relative to the traversal tips, this
walk is often significantly faster than an ordinary object traversal
because it can visit far fewer objects.
But in certain cases, it can be significantly *slower* than the usual
object traversal. Why? Because we need to compute complete bitmaps on
either side of the walk. If either one (or both) of the sides require
walking many (or all!) objects before they get to an existing bitmap,
the extra bitmap machinery is mostly or all overhead.
One of the benefits, however, is that even if the walk is slower, bitmap
traversals are guaranteed to provide an *exact* answer. Unlike the
traditional object traversal algorithm, which can over-count the results
by not opening trees for older commits, the bitmap walk builds an exact
reachability bitmap for either side, meaning the results are never
over-counted.
But producing non-exact results is OK for our traversal here (both in
the bitmap case and not), as long as the results are over-counted, not
under.
Relaxing the bitmap traversal to allow it to produce over-counted
results gives us the opportunity to make some significant improvements.
Instead of the above, the new algorithm only has to walk from the
*boundary* down to the nearest bitmap, instead of from each of the
UNINTERESTING tips.
The boundary-based approach still has degenerate cases, but we'll show
in a moment that it is often a significant improvement.
The new algorithm works as follows:
1. Build a (partial) bitmap of the haves side by first OR-ing any
bitmap(s) that already exist for UNINTERESTING commits between the
haves and the boundary.
2. For each commit along the boundary, add it as a fill-in traversal
tip (where the traversal terminates once an existing bitmap is
found), and perform fill-in traversal.
3. Build up a complete bitmap of the wants side as usual, stopping any
time we intersect the (partial) haves side.
4. Return the results.
And is more-or-less equivalent to using the *old* algorithm with this
invocation:
$ git rev-list --objects --use-bitmap-index $WANTS --not \
$(git rev-list --objects --boundary $WANTS --not $HAVES |
perl -lne 'print $1 if /^-(.*)/')
The new result performs significantly better in many cases, particularly
when the distance from the boundary commit(s) to an existing bitmap is
shorter than the distance from (all of) the have tips to the nearest
bitmapped commit.
Note that when using the old bitmap traversal algorithm, the results can
be *slower* than without bitmaps! Under the new algorithm, the result is
computed faster with bitmaps than without (at the cost of over-counting
the true number of objects in a similar fashion as the non-bitmap
traversal):
# (Computing the number of tagged objects not on any branches
# without bitmaps).
$ time git rev-list --count --objects --tags --not --branches
20
real 0m1.388s
user 0m1.092s
sys 0m0.296s
# (Computing the same query using the old bitmap traversal).
$ time git rev-list --count --objects --tags --not --branches --use-bitmap-index
19
real 0m22.709s
user 0m21.628s
sys 0m1.076s
# (this commit)
$ time git.compile rev-list --count --objects --tags --not --branches --use-bitmap-index
19
real 0m1.518s
user 0m1.234s
sys 0m0.284s
The new algorithm is still slower than not using bitmaps at all, but it
is nearly a 15-fold improvement over the existing traversal.
In a more realistic setting (using my local copy of git.git), I can
observe a similar (if more modest) speed-up:
$ argv="--count --objects --branches --not --tags"
hyperfine \
-n 'no bitmaps' "git.compile rev-list $argv" \
-n 'existing traversal' "git.compile rev-list --use-bitmap-index $argv" \
-n 'boundary traversal' "git.compile -c pack.useBitmapBoundaryTraversal=true rev-list --use-bitmap-index $argv"
Benchmark 1: no bitmaps
Time (mean ± σ): 124.6 ms ± 2.1 ms [User: 103.7 ms, System: 20.8 ms]
Range (min … max): 122.6 ms … 133.1 ms 22 runs
Benchmark 2: existing traversal
Time (mean ± σ): 368.6 ms ± 3.0 ms [User: 325.3 ms, System: 43.1 ms]
Range (min … max): 365.1 ms … 374.8 ms 10 runs
Benchmark 3: boundary traversal
Time (mean ± σ): 167.6 ms ± 0.9 ms [User: 139.5 ms, System: 27.9 ms]
Range (min … max): 166.1 ms … 169.2 ms 17 runs
Summary
'no bitmaps' ran
1.34 ± 0.02 times faster than 'boundary traversal'
2.96 ± 0.05 times faster than 'existing traversal'
Here, the new algorithm is also still slower than not using bitmaps, but
represents a more than 2-fold improvement over the existing traversal in
a more modest example.
Since this algorithm was originally written (nearly a year and a half
ago, at the time of writing), the bitmap lookup table shipped, making
the new algorithm's result more competitive. A few other future
directions for improving bitmap traversal times beyond not using bitmaps
at all:
- Decrease the cost to decompress and OR together many bitmaps
together (particularly when enumerating the uninteresting side of
the walk). Here we could explore more efficient bitmap storage
techniques, like Roaring+Run and/or use SIMD instructions to speed
up ORing them together.
- Store pseudo-merge bitmaps, which could allow us to OR together
fewer "summary" bitmaps (which would also help with the above).
Helped-by: Jeff King <peff@peff.net>
Helped-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Back in e8c58f894b (t: support GIT_TEST_WRITE_REV_INDEX, 2021-01-25), we
added a test knob to conditionally enable writing a ".rev" file when
indexing a pack. At the time, this was used to ensure that the test
suite worked even when ".rev" files were written, which served as a
stress-test for the on-disk reverse index implementation.
Now that reading from on-disk ".rev" files is enabled by default, the
test knob `GIT_TEST_WRITE_REV_INDEX` no longer has any meaning.
We could get rid of the option entirely, but there would be no
convenient way to test Git when ".rev" files *aren't* in place.
Instead of getting rid of the option, invert its meaning to instead
disable writing ".rev" files, thereby running the test suite in a mode
where the reverse index is generated from scratch.
This ensures that, when GIT_TEST_NO_WRITE_REV_INDEX is set to some
spelling of "true", we are still running and exercising Git's behavior
when forced to generate reverse indexes from scratch. Do so by setting
it in the linux-TEST-vars CI run to ensure that we are maintaining good
coverage of this now-legacy code.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since [1] first released with Git v2.37.0 the built-in version of "add
-i" has been the default. That built-in implementation was added in
[2], first released with Git v2.25.0.
At this point enough time has passed to allow for finding any
remaining bugs in this new implementation, so let's remove the
fallback code.
As with similar migrations for "stash"[3] and "rebase"[4] we're
keeping a mention of "add.interactive.useBuiltin" in the
documentation, but adding a warning() to notify any outstanding users
that the built-in is now the default. As with [5] and [6] we should
follow-up in the future and eventually remove that warning.
1. 0527ccb1b5 (add -i: default to the built-in implementation,
2021-11-30)
2. f83dff60a7 (Start to implement a built-in version of `git add
--interactive`, 2019-11-13)
3. 8a2cd3f512 (stash: remove the stash.useBuiltin setting,
2020-03-03)
4. d03ebd411c (rebase: remove the rebase.useBuiltin setting,
2019-03-18)
5. deeaf5ee07 (stash: remove documentation for `stash.useBuiltin`,
2022-01-27)
6. 9bcde4d531 (rebase: remove transitory rebase.useBuiltin setting &
env, 2021-03-23)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the sha1collisiondetection library was added and made the default
in [1] the interaction with APPLE_COMMON_CRYPTO added in [2] and [3]
seems to have been missed. On modern OSX and Darwin we are able to use
Apple's CommonCrypto both for SHA-1, and as a generic (but partial)
OpenSSL replacement.
This left OSX and Darwin without protection against the SHAttered
attack when building Git in its default configuration.
Let's also use sha1collisiondetection on OSX, to do so we'll need to
split up the "APPLE_COMMON_CRYPTO" flag into that flag and a new
"APPLE_COMMON_CRYPTO_SHA1".
Because of this we can stop conflating whether we want to use Apple's
CommonCrypto at all, and whether we want to use it for SHA-1. This
makes the CI recipe added in [4] simpler.
1. e6b07da278 (Makefile: make DC_SHA1 the default, 2017-03-17)
2. 4dcd7732db (Makefile: add support for Apple CommonCrypto facility, 2013-05-19)
3. 61067954ce (cache.h: eliminate SHA-1 deprecation warnings on Mac OS X, 2013-05-19)
4. 1ad5c3df35 (ci: use DC_SHA1=YesPlease on osx-clang job for CI,
2022-10-20)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Per [1] and the warnings our CI is emitting GitHub is phasing in
"macos-12" as their "macos-latest".
As with [2], let's pin our image to a specific version so that we're
not having it swept from under us, and our upgrade cycle can be more
predictable than whenever GitHub changes their images.
1. https://github.com/actions/runner-images/issues/6384
2. 0178420b9c (github-actions: run gcc-8 on ubuntu-20.04 image,
2022-11-25)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Adjust the GitHub CI to newer ubuntu release.
* jx/ci-ubuntu-fix:
ci: install python on ubuntu
ci: use the same version of p4 on both Linux and macOS
ci: remove the pipe after "p4 -V" to catch errors
github-actions: run gcc-8 on ubuntu-20.04 image
Python is missing from the default ubuntu-22.04 runner image, which
prevents git-p4 from working. To install python on ubuntu, we need
to provide the correct package names:
* On Ubuntu 18.04 (bionic), "/usr/bin/python2" is provided by the
"python" package, and "/usr/bin/python3" is provided by the "python3"
package.
* On Ubuntu 20.04 (focal) and above, "/usr/bin/python2" is provided by
the "python2" package which has a different name from bionic, and
"/usr/bin/python3" is provided by "python3".
Since the "ubuntu-latest" runner image has a higher version, its
safe to use "python2" or "python3" package name.
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There would be a segmentation fault when running p4 v16.2 on ubuntu
22.04 which is the latest version of ubuntu runner image for github
actions.
By checking each version from [1], p4d version 21.1 and above can work
properly on ubuntu 22.04. But version 22.x will break some p4 test
cases. So p4 version 21.x is exactly the version we can use.
With this update, the versions of p4 for Linux and macOS happen to be
the same. So we can add the version number directly into the "P4WHENCE"
variable, and reuse it in p4 installation for macOS.
By removing the "LINUX_P4_VERSION" variable from "ci/lib.sh", the
comment left above has nothing to do with p4, but still applies to
git-lfs. Since we have a fixed version of git-lfs installed on Linux,
we may have a different version on macOS.
[1]: https://cdist2.perforce.com/perforce/
Reviewed-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When installing p4 as a dependency, we used to pipe output of "p4 -V"
and "p4d -V" to validate the installation and output a condensed version
information. But this would hide potential errors of p4 and would stop
with an empty output. E.g.: p4d version 16.2 running on ubuntu 22.04
causes sigfaults, even before it produces any output.
By removing the pipe after "p4 -V" and "p4d -V", we may get a
verbose output, and stop immediately on errors because we have "set
-e" in "ci/lib.sh". Since we won't look at these trace logs unless
something fails, just including the raw output seems most sensible.
Reviewed-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
GitHub starts to upgrade its runner image "ubuntu-latest" from version
"ubuntu-20.04" to version "ubuntu-22.04". It will fail to find and
install "gcc-8" package on the new runner image.
Change some of the runner images from "ubuntu-latest" to "ubuntu-20.04"
in order to install "gcc-8" as a dependency.
The first revision of this patch tried to replace "$runs_on_pool" in
"ci/*.sh" with a new "$runs_on_os" environment variable based on the
"os" field in the matrix strategy. But these "os" fields in matrix
strategies are obsolete legacies from commit [1] and commit [2], and
are no longer useful. So remove these unused "os" fields.
[1]: c08bb26010 (CI: rename the "Linux32" job to lower-case "linux32",
2021-11-23)
[2]: 25715419bf (CI: don't run "make test" twice in one job, 2021-11-23)
Reviewed-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Address the root cause of technical debt we've been carrying since
sha1collisiondetection was made the default in [1]. In a preceding
commit we narrowly fixed a bug where the "DC_SHA1" variable would be
unset (in combination with "NO_APPLE_COMMON_CRYPTO=" on OSX), even
though we had the sha1collisiondetection library enabled.
But the only reason we needed to have such a user-exposed knob went
away with [1], and it's been doing nothing useful since then. We don't
care if you define DC_SHA1=*, we only care that you don't ask for any
other SHA-1 implementation. If it turns out that you didn't, we'll use
sha1collisiondetection, whether you had "DC_SHA1" set or not.
As a result of this being confusing we had e.g. [2] for cmake and the
recent [3] for ci/lib.sh setting "DC_SHA1" explicitly, even though
this was always a NOOP.
A much simpler way to do this is to stop having the Makefile and
CMakeLists.txt set "DC_SHA1" to be picked up by the test-lib.sh, let's
instead add a trivial "test-tool sha1-is-sha1dc". It returns zero if
we're using sha1collisiondetection, non-zero otherwise.
1. e6b07da278 (Makefile: make DC_SHA1 the default, 2017-03-17)
2. c4b2f41b5f (cmake: support for testing git with ctest, 2020-06-26)
3. 1ad5c3df35 (ci: use DC_SHA1=YesPlease on osx-clang job for CI,
2022-10-20)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
7b8cfe34 (Merge branch 'ed/fsmonitor-on-networked-macos',
2022-10-17) broke the build on macOS with sha1dc by bypassing our
hash abstraction (git_SHA_CTX etc.), but it wasn't caught before the
problematic topic was merged down to the 'master' branch. Nobody
was even compile testing with DC_SHA1 set, although it is the
recommended choice in these days for folks when they use SHA-1.
This was because the default for macOS uses Apple Common Crypto, and
both of the two CI jobs did not override the default. Tweak one of
them to use DC_SHA1 to improve the coverage.
We may want to give similar diversity for Linux jobs so that some of
them build with other implementations of SHA-1; they currently all
build and test with DC_SHA1 as that is the default on everywhere
other than macOS.
But let's start small to fill only the immediate need.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The current code is clean with these two sanitizers, and we would
like to keep it that way by running the checks for any new code.
The signal of "passed with asan, but not ubsan" (or vice versa) is
not that useful in practice, so it is tempting to run both santizers
in a single task, but it seems to take forever, so tentatively let's
try having two separate ones.
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As noted in a preceding commit the leak checking done by
"GIT_TEST_PASSING_SANITIZE_LEAK=true" (added in [1]) is incomplete
without combining it with "GIT_TEST_SANITIZE_LEAK_LOG=true".
Let's run our CI with that, to ensure that we catch cases where our
tests are missing the abort() exit code resulting from a leak for
whatever reason. The reasons for that are discussed in detail in a
preceding commit.
1. 956d2e4639 (tests: add a test mode for SANITIZE=leak, run it in
CI, 2021-09-23)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Recent CI update hides certain failures in test jobs, which has
been corrected.
* js/ci-github-workflow-markup:
ci(github): also mark up compile errors
ci(github): use grouping also in the `win-build` job
ci(github): bring back the 'print test failures' step
When GCC produces those helpful errors, we will want to present them in
the GitHub workflow runs in the most helpful manner. To that end, we
want to use workflow commands to render errors and warnings:
https://docs.github.com/en/actions/using-workflows/workflow-commands-for-github-actions
In the previous commit, we ensured that grouping is used for the build
in all jobs, and this allows us to piggy-back onto the `group` function
to transmogrify the output.
Note: If `set -o pipefail` was available, we could do this in a little
more elegant way. But since some of the steps are run using `dash`, we
have to do a little `{ ...; echo $? >exit.status; } | ...` dance.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We already do the same when building Git in all the other jobs.
This will allow us to piggy-back on top of grouping to mark up compiler
errors in the next commit.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Git now shows better information in the GitHub workflow runs when a test
case failed. However, when a test case was implemented incorrectly and
therefore does not even run, nothing is shown.
Let's bring back the step that prints the full logs of the failed tests,
and to improve the user experience, print out an informational message
for readers so that they do not have to know/remember where to see the
full logs.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update the GitHub workflow support to make it quicker to get to the
failing test.
* js/ci-github-workflow-markup:
ci: call `finalize_test_case_output` a little later
ci(github): mention where the full logs can be found
ci: use `--github-workflow-markup` in the GitHub workflow
ci(github): avoid printing test case preamble twice
ci(github): skip the logs of the successful test cases
ci: optionally mark up output in the GitHub workflow
ci/run-build-and-tests: add some structure to the GitHub workflow output
ci: make it easier to find failed tests' logs in the GitHub workflow
ci/run-build-and-tests: take a more high-level view
test(junit): avoid line feeds in XML attributes
tests: refactor --write-junit-xml code
ci: fix code style
A workflow change for translators are being proposed.
* jx/l10n-workflow-change:
l10n: Document the new l10n workflow
Makefile: add "po-init" rule to initialize po/XX.po
Makefile: add "po-update" rule to update po/XX.po
po/git.pot: don't check in result of "make pot"
po/git.pot: this is now a generated file
Makefile: remove duplicate and unwanted files in FOUND_SOURCE_FILES
i18n CI: stop allowing non-ASCII source messages in po/git.pot
Makefile: have "make pot" not "reset --hard"
Makefile: generate "po/git.pot" from stable LOCALIZED_C
Makefile: sort source files before feeding to xgettext
"git add -i" was rewritten in C some time ago and has been in
testing; the reimplementation is now exposed to general public by
default.
* js/use-builtin-add-i:
add -i: default to the built-in implementation
t2016: require the PERL prereq only when necessary
In the preceding commit we moved away from using xgettext(1) to both
generate the po/git.pot, and to merge the incrementally generated
po/git.pot+ file as we sourced translations from C, shell and Perl.
Doing it this way, which dates back to my initial
implementation[1][2][3] was conflating two things: With xgettext(1)
the --from-code both controls what encoding is specified in the
po/git.pot's header, and what encoding we allow in source messages.
We don't ever want to allow non-ASCII in *source messages*, and doing
so has hid e.g. a buggy message introduced in
a6226fd772 (submodule--helper: convert the bulk of cmd_add() to C,
2021-08-10) from us, we'd warn about it before, but only when running
"make pot", but the operation would still succeed. Now we'll error out
on it when running "make pot".
Since the preceding Makefile changes made this easy: let's add a "make
check-pot" target with the same prerequisites as the "po/git.pot"
target, but without changing the file "po/git.pot". Running it as part
of the "static-analysis" CI target will ensure that we catch any such
issues in the future. E.g.:
$ make check-pot
XGETTEXT .build/pot/po/builtin/submodule--helper.c.po
xgettext: Non-ASCII string at builtin/submodule--helper.c:3381.
Please specify the source encoding through --from-code.
make: *** [.build/pot/po/builtin/submodule--helper.c.po] Error 1
1. cd5513a716 (i18n: Makefile: "pot" target to extract messages
marked for translation, 2011-02-22)
2. adc3b2b276 (Makefile: add xgettext target for *.sh files,
2011-05-14)
3. 5e9637c629 (i18n: add infrastructure for translating Git with
gettext, 2011-11-18)
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The full logs are contained in the `failed-tests-*.zip` artifacts that
are attached to the failed CI run. Since this is not immediately
obvious to the well-disposed reader, let's mention it explicitly.
Suggested-by: Victoria Dye <vdye@github.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes the output easier to digest.
Note: since workflow output currently cannot contain any nested groups
(see https://github.com/actions/runner/issues/802 for details), we need
to remove the explicit grouping that would span the entirety of each
failed test script.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The current output of Git's GitHub workflow can be quite confusing,
especially for contributors new to the project.
To make it more helpful, let's introduce some collapsible grouping.
Initially, readers will see the high-level view of what actually
happened (did the build fail, or the test suite?). To drill down, the
respective group can be expanded.
Note: sadly, workflow output currently cannot contain any nested groups
(see https://github.com/actions/runner/issues/802 for details),
therefore we take pains to ensure to end any previous group before
starting a new one.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When investigating a test failure, the time that matters most is the
time it takes from getting aware of the failure to displaying the output
of the failing test case.
You currently have to know a lot of implementation details when
investigating test failures in the CI runs. The first step is easy: the
failed job is marked quite clearly, but when opening it, the failed step
is expanded, which in our case is the one running
`ci/run-build-and-tests.sh`. This step, most notably, only offers a
high-level view of what went wrong: it prints the output of `prove`
which merely tells the reader which test script failed.
The actually interesting part is in the detailed log of said failed
test script. But that log is shown in the CI run's step that runs
`ci/print-test-failures.sh`. And that step is _not_ expanded in the web
UI by default. It is even marked as "successful", which makes it very
easy to miss that there is useful information hidden in there.
Let's help the reader by showing the failed tests' detailed logs in the
step that is expanded automatically, i.e. directly after the test suite
failed.
This also helps the situation where the _build_ failed and the
`print-test-failures` step was executed under the assumption that the
_test suite_ failed, and consequently failed to find any failed tests.
An alternative way to implement this patch would be to source
`ci/print-test-failures.sh` in the `handle_test_failures` function to
show these logs. However, over the course of the next few commits, we
want to introduce some grouping which would be harder to achieve that
way (for example, we do want a leaner, and colored, preamble for each
failed test script, and it would be trickier to accommodate the lack of
nested groupings in GitHub workflows' output).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the web UI of GitHub workflows, failed runs are presented with the
job step that failed auto-expanded. In the current setup, this is not
helpful at all because that shows only the output of `prove`, which says
which test failed, but not in what way.
What would help understand the reader what went wrong is the verbose
test output of the failed test.
The logs of the failed runs do contain that verbose test output, but it
is shown in the _next_ step (which is marked as succeeding, and is
therefore _not_ auto-expanded). Anyone not intimately familiar with this
would completely miss the verbose test output, being left mostly
puzzled with the test failures.
We are about to show the failed test cases' output in the _same_ step,
so that the user has a much easier time to figure out what was going
wrong.
But first, we must partially revert the change that tried to improve the
CI runs by combining the `Makefile` targets to build into a single
`make` invocation. That might have sounded like a good idea at the time,
but it does make it rather impossible for the CI script to determine
whether the _build_ failed, or the _tests_. If the tests were run at
all, that is.
So let's go back to calling `make` for the build, and call `make test`
separately so that we can easily detect that _that_ invocation failed,
and react appropriately.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In b92cb86ea1 (travis-ci: check that all build artifacts are
.gitignore-d, 2017-12-31), a function was introduced with a code style
that is different from the surrounding code: it added the opening curly
brace on its own line, when all the existing functions in the same file
cuddle that brace on the same line as the function name.
Let's make the code style consistent again.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
macOS CI jobs have been occasionally flaky due to tentative version
skew between perforce and the homebrew packager. Instead of
failing the whole CI job, just let it skip the p4 tests when this
happens.
* cb/ci-make-p4-optional:
ci: use https, not http to download binaries from perforce.com
ci: reintroduce prevention from perforce being quarantined in macOS
ci: avoid brew for installing perforce
ci: make failure to find perforce more user friendly
Since 522354d70f (Add Travis CI support, 2015-11-27) the CI has used
http://filehost.perforce.com/perforce/ to download binaries from
filehost.perforce.com, they were then moved to this script in
657343a602 (travis-ci: move Travis CI code into dedicated scripts,
2017-09-10).
Let's use https instead for good measure. I don't think we need to
worry about the DNS or network between the GitHub CI and perforce.com
being MitM'd, but using https gives us extra validation of the payload
at least, and is one less thing to worry about when checking where
else we rely on non-TLS'd http connections.
Also, use the same download site at perforce.com for Linux and macOS
tarballs for consistency.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
5ed9fc3fc8 (ci: prevent `perforce` from being quarantined, 2020-02-27)
introduces this prevention for brew, but brew has been removed in a
previous commit, so reintroduce an equivalent option to avoid a possible
regression.
This doesn't affect github actions (as configure now) and is therefore
done silently to avoid any possible scary irrelevant messages.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Perfoce's cask in brew is meant[1] to be used only by humans, so replace
its use from the CI with a scripted binary download which is less likely
to fail, as it is done in Linux.
Kept the logic together so it will be less likely to break when moved
around as on the fly code changes in this area are settled, at which
point it will also feasable to ammend it to avoid some of the hardcoded
values by using similar variables to the ones Linux does.
In that same line, a POSIX sh syntax is used instead of the similar one
used in Linux in preparation for an unrelated future change that might
change the shell currently configured for it.
This change reintroduces the risk that the installed binaries might not
work because of being quarantined that was fixed with 5ed9fc3fc8 (ci:
prevent `perforce` from being quarantined, 2020-02-27) but fixing that
now was also punted for simplicity and since the affected cloud provider
is scheduled to be retired with an on the fly change, but should be
addressed if that other change is not integrated further.
The discussion on the need to keep 2 radically different versions of
the binaries to be tested with Linux vs macOS or how to upgrade to
newer versions now that brew won't do that automatically for us has
been punted for now as well. On that line the now obsolete comment
about it in lib.sh was originally being updated by this change but
created conflicts as it is moved around by other on the fly changes,
so will be addressed independently as well.
[1] https://github.com/Homebrew/homebrew-cask/pull/122347#discussion_r856026584
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for a future change that will make perforce installation
optional in macOS, make sure that the check for it is done without
triggering scary looking errors and add a user friendly message instead.
All other existing uses of 'type <cmd>' in our shell scripts that
check the availability of a command <cmd> send both standard output
and error stream to /dev/null to squelch "<cmd> not found" diagnostic
output, but this script left the standard error stream shown.
Redirect it just like everybody else to squelch this error message that
we fully expect to see.
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a regression in 707d2f2fe8 (CI: use "$runs_on_pool", not
"$jobname" to select packages & config, 2021-11-23).
In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in
$PATH points to clang, we need to use gcc-9 instead. Likewise for the
linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would
select GCC 9.4.0 instead of GCC 8.4.0.
Furthermore in 25715419bf (CI: don't run "make test" twice in one
job, 2021-11-23) when the "linux-TEST-vars" job was split off from
"linux-gcc" the "cc_package: gcc-8" line was copied along with
it, so its "cc_package" line wasn't working as intended either.
As a table, this is what's changed by this commit, i.e. it only
affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:
|-------------------+-----------+-------------------+-------+-------|
| jobname | vector.cc | vector.cc_package | old | new |
|-------------------+-----------+-------------------+-------+-------|
| linux-clang | clang | - | clang | clang |
| linux-sha256 | clang | - | clang | clang |
| linux-gcc | gcc | gcc-8 | gcc | gcc-8 |
| osx-clang | clang | - | clang | clang |
| osx-gcc | gcc | gcc-9 | clang | gcc-9 |
| linux-gcc-default | gcc | - | gcc | gcc |
| linux-TEST-vars | gcc | gcc-8 | gcc | gcc-8 |
|-------------------+-----------+-------------------+-------+-------|
Reported-by: Carlo Arenas <carenas@gmail.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have a copy of uncompress2() implementation in compat/ so that we
can build with an older version of zlib that lack the function, and
the build procedure selects if it is used via the NO_UNCOMPRESS2
$(MAKE) variable. This is yet another "annoying" knob the porters
need to tweak on platforms that are not common enough to have the
default set in the config.mak.uname file.
Attempt to instead ask the system header <zlib.h> to decide if we
need the compatibility implementation. This is a deviation from the
way we have been handling the "compatiblity" features so far, and if
it can be done cleanly enough, it could work as a model for features
that need compatibility definition we discover in the future. With
that goal in mind, avoid expedient but ugly hacks, like shoving the
code that is conditionally compiled into an unrelated .c file, which
may not work in future cases---instead, take an approach that uses a
file that is independently compiled and stands on its own.
Compile and link compat/zlib-uncompress2.c file unconditionally, but
conditionally hide the implementation behind #if/#endif when zlib
version is 1.2.9 or newer, and unconditionally archive the resulting
object file in the libgit.a to be picked up by the linker.
There are a few things to note in the shape of the code base after
this change:
- We no longer use NO_UNCOMPRESS2 knob; if the system header
<zlib.h> claims a version that is more cent than the library
actually is, this would break, but it is easy to add it back when
we find such a system.
- The object file compat/zlib-uncompress2.o is always compiled and
archived in libgit.a, just like a few other compat/ object files
already are.
- The inclusion of <zlib.h> is done in <git-compat-util.h>; we used
to do so from <cache.h> which includes <git-compat-util.h> as the
first thing it does, so from the *.c codes, there is no practical
change.
- Until objects in libgit.a that is already used gains a reference
to the function, the reftable code will be the only one that
wants it, so libgit.a on the linker command line needs to appear
once more at the end to satisify the mutual dependency.
- Beat found a trick used by OpenSSL to avoid making the
conditionally-compiled object truly empty (apparently because
they had to deal with compilers that do not want to see an
effectively empty input file). Our compat/zlib-uncompress2.c
file borrows the same trick for portabilty.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Helped-by: Beat Bolli <dev+git@drbeat.li>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Drop support for TravisCI and update test workflows at GitHub.
* ab/ci-updates:
CI: don't run "make test" twice in one job
CI: use "$runs_on_pool", not "$jobname" to select packages & config
CI: rename the "Linux32" job to lower-case "linux32"
CI: use shorter names that fit in UX tooltips
CI: remove Travis CI support
The "reftable" backend for the refs API, without integrating into
the refs subsystem, has been added.
* hn/reftable:
Add "test-tool dump-reftable" command.
reftable: add dump utility
reftable: implement stack, a mutable database of reftable files.
reftable: implement refname validation
reftable: add merged table view
reftable: add a heap-based priority queue for reftable records
reftable: reftable file level tests
reftable: read reftable files
reftable: generic interface to tables
reftable: write reftable files
reftable: a generic binary tree implementation
reftable: reading/writing blocks
Provide zlib's uncompress2 from compat/zlib-compat.c
reftable: (de)serialization for the polymorphic record type.
reftable: add blocksource, an abstraction for random access reads
reftable: utility functions
reftable: add error related functionality
reftable: add LICENSE
hash.h: provide constants for the hash IDs
CI has been taught to catch some Unicode directional formatting
sequence that can be used in certain mischief.
* js/ci-no-directional-formatting:
ci: disallow directional formatting
In 9a5315edfd (Merge branch 'js/patch-mode-in-others-in-c',
2020-02-05), Git acquired a built-in implementation of `git add`'s
interactive mode that could be turned on via the config option
`add.interactive.useBuiltin`.
The first official Git version to support this knob was v2.26.0.
In 2df2d81ddd (add -i: use the built-in version when
feature.experimental is set, 2020-09-08), this built-in implementation
was also enabled via `feature.experimental`. The first version with this
change was v2.29.0.
More than a year (and very few bug reports) later, it is time to declare
the built-in implementation mature and to turn it on by default.
We specifically leave the `add.interactive.useBuiltin` configuration in
place, to give users an "escape hatch" in the unexpected case should
they encounter a previously undetected bug in that implementation.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "linux-clang" and "linux-gcc" jobs both run "make test" twice, but
with different environment variables. Running these in sequence seems
to have been done to work around some constraint on Travis, see
ae59a4e44f (travis: run tests with GIT_TEST_SPLIT_INDEX, 2018-01-07).
By having these run in parallel we'll get jobs that finish much sooner
than they otherwise would have.
We can also simplify the control flow in "ci/run-build-and-tests.sh"
as a result, since we won't run "make test" twice we don't need to run
"make" twice at all, let's default to "make all test" after setting
the variables, and then override it to just "all" for the compile-only
tests.
Add a comment to clarify that new "test" targets should adjust
$MAKE_TARGETS rather than being added after the "case/esac". This
should avoid future confusion where e.g. the compilation-only
"pedantic" target will unexpectedly start running tests. See [1] and
[2].
1. https://lore.kernel.org/git/211122.86ee78yxts.gmgdl@evledraar.gmail.com/
2. https://lore.kernel.org/git/211123.86ilwjujmd.gmgdl@evledraar.gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change the setup hooks for the CI to use "$runs_on_pool" for the
"$regular" job. Now we won't need as much boilerplate when adding new
jobs to the "regular" matrix, see 956d2e4639 (tests: add a test mode
for SANITIZE=leak, run it in CI, 2021-09-23) for the last such commit.
I.e. now instead of needing to enumerate each jobname when we select
packages we can install things depending on the pool we're running
in.
That we didn't do this dates back to the now gone dependency on Travis
CI, but even if we add a new CI target in the future this'll be easier
to port over, since we can probably treat "ubuntu-latest" as a
stand-in for some recent Linux that can run "apt" commands.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As a follow-up to the preceding commit's shortening of CI job names,
rename the only job that starts with an upper-case letter to be
consistent with the rest. It was added in 88dedd5e72 (Travis: also
test on 32-bit Linux, 2017-03-05).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove support for running the CI in travis. The last builds in it are
from 5 months ago[1] (as of 2021-11-19), and our documentation has
referred to GitHub CI instead since f003a91f5c (SubmittingPatches:
replace discussion of Travis with GitHub Actions, 2021-07-22).
We'll now run the "t9810 t9816" and tests on OSX. We didn't before, as
we'd carried the Travis exclusion of them forward from
522354d70f (Add Travis CI support, 2015-11-27). Let's hope whatever
issue there was with them was either Travis specific, or fixed since
then (I'm not sure).
The "apt-add-repository" invocation (which we were doing in GitHub CI)
isn't needed, it was another Travis-only case that was carried forward
into more general code. See 0f0c51181d (travis-ci: install packages
in 'ci/install-dependencies.sh', 2018-11-01).
Remove the "linux-gcc-4.8" job added in fb9d7431cf (travis-ci: build
with GCC 4.8 as well, 2019-07-18), it only ran in Travis CI.
1. https://travis-ci.org/github/git/git/builds
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As described in https://trojansource.codes/trojan-source.pdf, it is
possible to abuse directional formatting (a feature of Unicode) to
deceive human readers into interpreting code differently from compilers.
For example, an "if ()" expression could be enclosed in a comment, but
rendered as if it was outside of that comment. In effect, this could
fool a reviewer into misinterpreting the code flow as benign when it is
not.
It is highly unlikely that Git's source code wants to contain such
directional formatting in the first place, so let's just disallow it.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Junio C Hamano
Jeff King
Taylor Blau
Ævar Arnfjörð Bjarmason
Jiang Xin
Johannes Schindelin
Carlo Marcelo Arenas Belón