To make CI turnaround time shorter, by cutting the longest running jobs.
After this patch all jobs should finish around 10-11 minutes. Down from
15-16 minutes before this patch.
Suggested-by: Stefan Eissing
The fuzzing workflow is now the slowest (with a 7-minute startup time
needed to build deps from source on each run), followed by macOS
and Windows torture tests (both split in two now). Without fuzzing, it's
under 10 minutes.
Notes:
- an extra cost with job-pairs is installing prereqs,
configuring/building curl and tests twice. GitHub doesn't support
making a matrix job a prereq for another workflow that may fix this:
https://github.com/orgs/community/discussions/42335
This overhead is significant on Windows: 11m20 -> 9m20 + 8m40
- job-pairs are annoying to maintain and keep in sync.
- splitting tests into halves is a manual process and needs to be
revisited from time to time. Possibly something to automate with
a runtests option, e.g. with `1 of 50%` and `2 of 50%`?
Also:
- split torture tests in two equal pieces, replacing the `FTP` + `!FTP`
split used earlier.
Related perf improvements from today:
- make scan-build 2x fast:
ff958fc4b2#20159
- drop build-only Testi386 fuzz workflow (from curl), saving 9m per run:
0106023c1fhttps://github.com/curl/curl-fuzzer/pull/236Closes#20153
To make the longest running FreeBSD job finish 1.5 minutes faster
(9.5m -> 8m).
Examples are still built with both autotools and cmake, one on Intel and
one on ARM.
Closes#20146
- update `github/codeql-action` from 4.31.8 to 4.31.9
- update `cross-platform-actions/action` from 0.30.0 to 0.32.0
- support for OmniOS and FreeBSD 15.0
- releases are now immutable
Closes#20140
Tailored for each job with a relatively tight limits. Also with no
tolerance in valgrind tests: 4 of the 4 has to be run.
Based on Test Clutch feature matrix which displays the minimum and
actual number of tests:
https://testclutch.curl.se/static/reports/feature-matrix.html
Also:
- runtests.pl: include total number of tests in the error message shown
when the limit was not met.
Assisted-by: Dan Fandrich
Follow-up to 3f1cd809ee#19942Closes#20050
Also:
- vquic-tls.h: do not include unused headers for non-H3 builds.
- autotools: stop looking for `openssl/x509.h` header.
- cmp-config.pl: delete exception for `openssl/x509.h`.
- examples: format/comment sync between the two touched files.
- openssl: drop unused `curlx/wait.h` include.
Closes#20049
- update `filelock` from 3.20.0 to 3.20.1 (CVE-2025-68146) (used in pytests)
- update `pytest` from 9.0.1 to 9.0.2
- update `ruff` from 0.14.8 to 0.14.9
Closes#20004
- lib: delete two unused `<sys/un.h>` includes.
- lib: drop interim macro `WIN32_SOCKADDR_UN`.
Follow-up to 0fe9018e1a#7737
Also fixing a potential issue of leaving unix socket support disabled
if any header would include Windows' `afunix.h`, and define
`UNIX_PATH_MAX` on its own.
- connect: honor unix socket disable option.
- connect: simplify unix socket PP condition.
`USE_UNIX_SOCKETS` already means the necessary header/type are
available, guaranteed by configure. `AF_UNIX` is already used
elsewhere in the code without explicit checks.
- curl_setup.h: document availability of `afunix.h` on Windows more.
It requires mingw-w64 10+ or MS SDK 10.17763.0 VS2017 15.8+.
- curl_setup.h: use `afunix.h` with mingw-w64 v10+ to start avoiding
the local workaround if possible.
- GHA/windows: test disable unix socket option on Windows.
Ref: https://devblogs.microsoft.com/commandline/af_unix-comes-to-windows/Closes#19989
To run more pytest sshd tests, and for more static analysis.
Also:
- drop redundant option from `openssl libssh2 ...` config.
- GHA/linux: enable pytest in the LTO job (to test libssh2).
- avoid both with local builds of OpenSSL-forks, due to crypto lib
mixups causing a mixture of build error, crashes, test failures.
Follow-up to eb39fee40b#19934Closes#19943
To formalize they are now XML-compliant (with some asterisks.)
Also to help syntax highlighters work on them to make their content more
readable.
Also:
- Delete empty comment decorations.
- GHA/checksrc: simplify XML check.
- runtests: fail to load test data with XML prolog missing.
Follow-up to bfe6eb1c06#19927
Follow-up to 87ba80a6dfCloses#19946
- cmake: automatically disable typecheck when running clang-tidy,
to avoid possible interference, and to improve performance.
- INSTALL-CMAKE: document both this, and unity=off for clang-tidy.
- GHA/linux: disable for some static analyzers CI jobs to avoid possible
interference.
- GHA/linux: disable in Fil-C job to improve build performance.
Follow-up to 9e6f1c5efb#19637
Follow-up to fd2ca2399e#17955Closes#19941
To detect mistakes made in the runtests framework that reduce
the number of test runs. Before this patch it could go undetected with
a green CI.
The minimum thresholds will need light maintenance going forward (either
bumping them periodically, or adjust if some may fell below minimums for
justified reasons). We may also make minimums tighter or looser, or more
job-specific.
Latest number of test runs for each job can be seen at Test Clutch:
https://testclutch.curl.se/static/reports/feature-matrix.html
Also:
- GHA: set minimums.
Assisted-by: Dan Fandrich
Follow-up to f2a75a14dd
Follow-up to bb1391f943#19510Closes#19942
- drop stray duplicate empty lines in docs, scripts, test data, include,
examples, tests.
- drop duplicate PP parenthesis.
- curl-functions.m4: move literals to the right side in if expressions,
to match rest of the source code.
- FAQ.md: delete language designator from an URL.
- packages: apply clang-format (OS400, VMS).
- scripts/schemetable.c: apply clang-format.
- data320: delete duplicate empty line that doesn't change the outcome.
- spacecheck: extend to check for duplicate empty lines
(with exceptions.)
- fix whitespace nits
Closes#19936
Nothing conclusive for the last ~30 days when `taskkill` was made
a no-op. Jobs remained flaky with all known failure modes. Sometimes
they finish green on the first run, sometimes they fail. Hard to say
more without comparing detailed stats for this period and the
preceding (or upcoming) one.
In almost all runs, the PID to be killed did not exist at the time of
check.
Follow-up to 2701ac6a4d#19421Closes#19897
To make the test files XML-compliant, and the expected results
possibly easier to manage by keeping them in `.md`, `.html`, `.1`
and `.txt` files.
Non-XML-compliant files are down to 31 (1.6%) after this patch.
Closes#19882
With Schannel and Unicode, `-shallow=13`. It finishes in 12 minutes,
making it the slowest Windows job. It's still on par with torture jobs
on other platforms (though they manage to fit `-shallow=25`).
Also `-shallow=13` still caught leaks in multiple tests.
Also:
- test2300: exclude from CI Windows torture tests.
- experimental.
The downside of going with deeper torture tests, is that it requires
increasing the job timeout. This in turns means that a hung job takes
more minutes to be killed (due to GitHub bugs where a hung step does not
honor the per-step timeout on Windows, another bug where a hung job gets
killed +5 minutes above the workflow timeout, and another bug (or
feature?) where other failed/hung jobs in the the workflow cannot be
restarted till the last job finishes or gets killed. And all this
probably related to a Perl bug which makes it hang on fork errors, which
is turn related to Cygwin/MSYS2 runtime bugs which breaks fork in case
of curl's mixed MSYS2-Perl/native-curl-binaries environment.)
The end result in longer forced waits before being able to restart flaky
jobs, which slows down iterations and annoying.
Also tried:
- non-c-ares job: detected known issues much less often.
- replaced libidn2 with WinIDN: detected known issues much less often.
- runtests -j9-j20 values: did not make a difference.
- other `-shallow` values: 20 is the max feasible, but comes with the
downside described above.
Ref: #19675 (reboot of)
Follow-up to f08417c425#19863Closes#19865
- URLs specified outside of the markdown []() are now extracted and
checked
- also check TODO, FAQ and KNOWN_BUGS
- more aggressive avoiding to check github.com/curl/curl, all uses of
example domains and some more established URLs on the curl.se site
- list all errors in the end to make them easier to spot in CI logs
Closes#19848
To reduce to amount of Debian packages to install, which hopefully
removes some flakiness due to sometimes very slow Azure package
distro servers. Possible also making these jobs finish 20s faster.
Windows from Debian | llvm | gcc
:------------------ | :----------------: | :----------------:
build time | 2m41s -> 2m20s | 3m19s -> 2m57s
installed packages | 288 -> 142 | 247 -> 99
downloads | 403 MB -> 240 MB | 297 MB -> 134 MB
disk space | 2132 MB -> 1289 MB | 1582 MB -> 739 MB
Before: https://github.com/curl/curl/actions/runs/19765983026
After: https://github.com/curl/curl/actions/runs/19766373960?pr=19749
Ref: 02149b7e36Closes#19749
Install CMake from the Kitware GitHub release archive. To allow choosing
its version independently from the OS.
Switch to 3.7.0 (from 3.7.2) to test the earliest supported version.
Also tested OK with 3.18.4 and 3.7.2.
The download and install step takes 1-2 seconds.
Follow-up to c9e50e9e39#19737Closes#19738
