0.24.6 is the quiche version without the fix for proper handling fo
RESET streams. Require a verion higher than that to run test_05_02.
Follow-up to 14478429e7#19916Closes#19921
`LDAP_VENDOR_NAME` and `winber.h` are available in all supported
MS SDK and mingw-w64 versions. Stop checking for them.
Also drop redundant parenthesis in PP expression.
Closes#19918
Let nghttpx only use http/1.1 to backend. This reproduces the bug in
quiche with higher frequency. Allow test_14_05 to now return a 400 in
addition to the 431 we get from a h2 backend to nghttpx.
Skip test_05_02 in h3 on quiche not newer than version 0.24.4 in which
its bug is fixed: https://github.com/cloudflare/quiche/pull/2278
Ref: https://github.com/cloudflare/quiche/issues/2277Closes#19770 (original Issue)
Closes#19916
Do the whole realloc and the subsequent logging under mutex lock. This
fixed log entries that state allocation a memory location before realloc
logs it as being freed.
Closes#19900
Move new tests from test_12 to test_06 (eyeballing) where they better
fit. Increase reliability by check Alt-Svc redirects from h3 to a lower
version for a port where no h3 is available.
Closes#19903
Explicitly disable these warnings to allow using `-Weverything`.
There are around 600 of them across the codebase.
Silencing them has some drawbacks:
- enums (`CURLcode` mostly) would have to be cast to int to avoid
different signedness depending on C compiler.
(llvm/gcc: unsigned, MSVC/clang-cl: signed by default)
- hex masks need casts to unsigned to avoid the warning.
- fixing remaining warnings is annoying without fixing the above.
- without fixing all warnings the option cannot be enabled, to keep
the codebase warning free.
Ref: #18343 (silenced all warnings, but without the enum cast)
Follow-up to 92f215fea1#18477Closes#19907
CMake 3.18 was released on 2020-07-15.
It enables using (and/or dropping workarounds) for these features:
LTO support, better performance and pkg-config support, `OBJECT` target,
`-S`, `-B`, `--verbose`, `--install` on the command-line, lib directory
support in interface targets, target_link_options(), LINK_OPTIONS,
FetchContent, `list(PREPEND ...)`, unity, Ninja, fixed imported global
issues.
Ref: https://github.com/curl/curl/discussions/18704Closes#19902
All Windows platforms support it. It was permanently enabled with most
build methods. The exception is autotools where it is enabled by
default, with an option to disable it. It changed the build in a few
places for rarely tested code paths, but not bringing other advantages
(and used some 64-bit APIs anyway). This patch makes autotools'
`--disable-largefile` option a no-op for Windows.
Closes#19888
MSTATE_TUNNELING is no longer in use now that we have proxy connection
filters. Remove the state.
Remove the http handler `connect_it` method as it was merely a NOP.
Closes#19894
When the Alt-Svc points to the same host and port, add the destination
ALPN to the `wanted` versions and set it also as the `preferred` version
in negotiations.
This allows Alt-Svc for h3 to point to h2 and have it tried first. Also,
this allows Alt-Svc to say http/1.1 is preferred and changes the ALPN
protocol ordering for the TLS handshake.
Add tests in various combination to verify this works.
Reported-by: yushicheng7788 on github
Fixes#19740Closes#19874
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
Only the TLS 1.2 structure for now since it's simpler, and only has a
single label type. This has the bonus of also testing libressl that only
supports logging keys in TLS 1.2
Closes#19816
- TCP Fast Open support on Windows
TFO doesn't really work on the internet and isn't really used anywhere.
We use QUIC now.
- get rid of PATH_MAX
Not a priority and when using 3rd party libraries not really up to us.
- auto-detect proxy
A dream we can just stop having.
- config file parsing
Let's not do that in our library.
- add asynch getaddrinfo support
Let's not add a limited glibc specific backend with bad API
- FTP HOST
If we managed this far without name based vhost FTP, let's not.
Closes#19890
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
To make the test files XML-compliant, and the expected results
possibly easier to manage by keeping them in `.c` files.
Non-XML-compliant files are down to 36 after this patch.
Also:
- make all macro expansions apply to `%includetext` contents.
Closes#19799
- to avoid dupes.
- missing slashes.
- drop `.git` suffix from GitHub git repo URLs for a few outliers.
- use short YouTube URL like curl-www does.
- sync two RFC doc URLs with others.
Closes#19879
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
