Viktor Szakats [Thu, 29 Aug 2024 08:15:26 +0000 (10:15 +0200)]
build: drop unused feature-detection code for Apple `poll()`
Drop Apple-specific detection logic for `poll()`. This detection snippet
has been disabled for Apple in both configure and cmake, for `poll()`
being broken on Apple since 10.12 Sierra (2016).
Also replace `exit(1);` with `return 1;` in configure, to make the
snippets match.
Stefan Eissing [Thu, 29 Aug 2024 10:53:57 +0000 (12:53 +0200)]
gnutls: send all data
Turns out `gnutls_record_send()` does really what the name says: it
sends exactly one TLS record. If more than 16k are there to send, it
needs to be called again with new buffer offset and length.
Continue sending record until the input is all sent or a EAGAIN (or
fatal error) is returned by gnutls.
Stefan Eissing [Wed, 28 Aug 2024 10:54:29 +0000 (12:54 +0200)]
urldata: remove crlf_conversions counter
Since the introduction of client writers, we check the body length in
the PROTOCOL phase and do FTP lineend conversions laster in the
CONTENT_DECODING phase. This means we no longer need to count the
conversions for length checks.
Viktor Szakats [Wed, 28 Aug 2024 09:01:26 +0000 (11:01 +0200)]
test1013.pl: require case match for features, order match for protos, fix issue
Update the script for test 1013 and 1014 to require:
- case-sensitive match for the curl feature list.
(Continue to allow case-difference for protocols. They've always been
in uppercase within curl config.)
- matching order for the protocol list.
(Continue to allow any order for features. autotools builds on
platforms without `sort -f` need it. E.g. Old Linux CI)
Also:
- fix casing of the `gsasl` feature in `configure`, to match `curl -V`
and cmake.
Viktor Szakats [Wed, 28 Aug 2024 12:07:03 +0000 (14:07 +0200)]
GHA/windows: vcpkg GnuTLS started breaking CI, temp drop it
Starting today vcpkg wants to rebuild GnuTLS but fails:
```
error: building shiftmedia-libgnutls:x64-windows failed with: BUILD_FAILED
```
Ref: https://github.com/curl/curl/actions/runs/10594890318/job/29359499149#step:5:144
Temporary solution:
- drop it from the MultiSSL job.
- replace with mbedTLS job. This job still tests libssh and I could
not find a better place for it right away.
GnuTLS to be restored once it builds again. Possibly when this hash
reaches the GHA `windows-latest` runner:
https://github.com/microsoft/vcpkg/commit/f5ec6f30ff70f04f841436a0f36600bdbabfcfbf
Also:
- switch to Debug for the mbedTLS job. Should also work now with
GnuTLS, once it's back:
Ref: https://github.com/microsoft/vcpkg/pull/40473
Stefan Eissing [Tue, 27 Aug 2024 10:12:37 +0000 (12:12 +0200)]
connect: limit update IP info
Update IP related information at the connection and the transfer in two
places only: once the filter chain connects and when a transfer is added
to a connection. The latter only updates on reuse when the filters
already are connected.
The only user of that information before a full connect is the HAProxy
filter. Add cfilter CF_QUERY_IP_INFO query to let it find the
information from the filters "below".
This solves two issues with the previous version:
- updates where often done twice with the same info
- happy eyeballing filter "forks" could overwrite each others
updates before the full winner was determined.
Stefan Eissing [Mon, 26 Aug 2024 08:56:28 +0000 (10:56 +0200)]
transfer: skip EOS read when download done
When we downloaded all we wanted, and we did not want a response body,
and no Trailer: has been announced, and the receive gives EAGAIN, do not
hang around unnecessarily.
Some servers are buggy in HEAD processing and fail to send the HTTP/2
EOS. Since we do not need any more data, end the request right there.
This will cause us to send a RST_STREAM to the server.
Stefan Eissing [Fri, 23 Aug 2024 11:58:41 +0000 (13:58 +0200)]
cpool: rename "connection cache/conncache" to "Connection Pools/cpool"
This is a better match for what they do and the general "cpool"
var/function prefix works well.
The pool now handles very long hostnames correctly.
The following changes have been made:
* 'struct connectdata', e.g. connections, keep new members
named `destination` and ' destination_len' that fully specifies
interface+port+hostname of where the connection is going to.
This is used in the pool for "bundling" of connections with
the same destination. There is no limit on the length any more.
* Locking: all locks are done inside conncache.c when calling
into the pool and released on return. This eliminates hazards
of the callers keeping track.
* 'struct connectbundle' is now internal to the pool. It is no
longer referenced by a connection.
* 'bundle->multiuse' no longer exists. HTTP/2 and 3 and TLS filters
no longer need to set it. Instead, the multi checks on leaving
MSTATE_CONNECT or MSTATE_CONNECTING if the connection is now
multiplexed and new, e.g. not conn->bits.reuse. In that case
the processing of pending handles is triggered.
* The pool's init is provided with a callback to invoke on all
connections being discarded. This allows the cleanups in
`Curl_disconnect` to run, wherever it is decided to retire
a connection.
* Several pool operations can now be fully done with one call.
Pruning dead connections, upkeep and checks on pool limits
can now directly discard connections and need no longer return
those to the caller for doing that (as we have now the callback
described above).
* Finding a connection for reuse is now done via `Curl_cpool_find()`
and the caller provides callbacks to evaluate the connection
candidates.
* The 'Curl_cpool_check_limits()' now directly uses the max values
that may be set in the transfer's multi. No need to pass them
around. Curl_multi_max_host_connections() and
Curl_multi_max_total_connections() are gone.
* Add method 'Curl_node_llist()' to get the llist a node is in.
Used in cpool to verify connection are indeed in the list (or
not in any list) as they need to.
I left the conncache.[ch] as is for now and also did not touch the
documentation. If we update that outside the feature window, we can
do this in a separate PR.
Multi-thread safety is not achieved by this PR, but since more details
on how pools operate are now "internal" it is a better starting
point to go for this in the future.
Viktor Szakats [Tue, 27 Aug 2024 10:32:56 +0000 (12:32 +0200)]
CI: add test timeouts, more cmake build tests, fix VS2010 C warning
- GHA/macos: set timeout for test runs.
Double the value for autotools to fit torture tests.
- GHA/cygwin: reduce test run timeout for autotools.
- GHA/cygwin: enable building tests with cmake.
- GHA/windows: enable building tests with MSYS cmake.
- GHA/windows: enable building tests with MSVC UWP.
- appveyor: enable building tests with VS2008 x86 and VS2010 x64.
- tests: add workaround compiler warnings when building with VS2010:
```
tests\server\util.c(482): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(486): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(490): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(720): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(726): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(732): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(781): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(785): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
tests\server\util.c(789): warning C4306: 'type cast' : conversion from 'int' to 'void (__cdecl *)(int)' of greater size
```
Ref: https://ci.appveyor.com/project/curlorg/curl/builds/50485633/job/4ujlwxod3cexmn2q#L1535
With this, tests are built in all GHA cygwin/windows jobs.
Timeouts avoid long runs with runaway issues, example:
https://github.com/curl/curl/actions/runs/10575522173
Viktor Szakats [Sun, 25 Aug 2024 17:32:10 +0000 (19:32 +0200)]
configure: fix WinIDN builds targeting old Windows
1. GHA/windows: enable WinIDN in Linux cross-builds.
(to reveal the issue in CI.)
2. fix compiler warning when building with mingw-w64 supporting
WinIDN, while targeting pre-Vista Windows, with a `WINVER` set to
target Vista or newer. (Such was Ubuntu's mingw-w64 with the
classic-mingw-specific trick in point 3 of this PR.)
```
../../lib/idn.c:154:23: error: redundant redeclaration of ‘IdnToAscii’ [-Werror=redundant-decls]
154 | WINBASEAPI int WINAPI IdnToAscii(DWORD dwFlags,
| ^~~~~~~~~~
In file included from /usr/share/mingw-w64/include/windows.h:73,
from /usr/share/mingw-w64/include/winsock2.h:23,
from ../../lib/setup-win32.h:91,
from ../../lib/curl_setup.h:308,
from ../../lib/idn.c:29:
/usr/share/mingw-w64/include/winnls.h:1075:30: note: previous declaration of ‘IdnToAscii’ was here
1075 | WINNORMALIZEAPI int WINAPI IdnToAscii (DWORD dwFlags, LPCWSTR lpUnicodeCharStr, int cchUnicodeChar, LPWSTR lpASCIICharStr, int cchASCIIChar);
| ^~~~~~~~~~
[...same for IdnToUnicode...]
```
Ref: https://github.com/curl/curl/actions/runs/10542832783/job/29210098553#step:7:89
Viktor Szakats [Mon, 26 Aug 2024 12:26:29 +0000 (14:26 +0200)]
tests: delete `libhostname.so` and `chkhostname`
Before this patch, `libhostname.so` and `chkhostname` were a test
facility for overriding `gethostname()` in non-debug builds on
Linux and other Unix platforms supporting `LD_PRELOAD`.
`gethostname()` has a single use with SMTP.
The alternative way to override `gethostname()` is building in debug
mode, which allows to do this via the `CURL_GETHOSTNAME` env, on all
platforms.
Drop the `LD_PRELOAD` solution in favour of the above.
Also:
- delete inactive NTLM code with a `gethostname()` call made from it.
- streamline NTLM code by dropping a `printf()` and a macro.
- tests: stop setting `CURL_GETHOSTNAME` where unnecessary.
Eric Norris [Mon, 26 Aug 2024 19:13:46 +0000 (15:13 -0400)]
share: don't reinitialize conncache
Before this change, calling curl_share_setopt w/ CURL_LOCK_DATA_CONNECT
a second time would re-initialize the connection cache, rather than use
the existing one.
After this change, calling curl_share_setopt w/ CURL_LOCK_DATA_CONNECT
multiple times will have no effect after the first call.
Viktor Szakats [Fri, 23 Aug 2024 15:42:08 +0000 (17:42 +0200)]
cmake: `Libs.private` improvements
- skip adding pkg-config libdirs if they are system locations.
- replace custom regexes with `get_filename_component()`.
- collect `-L` and `-framework` separately.
It means these will appear before libs in the `Libs.private` entry,
syncing it with `./configure`.
- collect in a list variable (was: string).
- use `list(REMOVE_DUPLICATES)` to deduplicate libdirs.
- rename internal variable that is now solely used for system libdirs.
Daniel Stenberg [Mon, 26 Aug 2024 09:56:32 +0000 (11:56 +0200)]
transfer: remove comments, add asserts
Curl_xfer_send and Curl_xfer_recv had commented FIXMEs about protocol
setting up the transfers badly, but in reality these functions are too
low-level to be able to depend on the protocol transfer setups having
been done yet. Removed.
The functions had checks for data and data->conn that I convered to
asserts since they SHOULD always be valid in this function. The same
goes for the runtime check for buffer_size > 0 that I also converted to
an assert since that should never be set to an invalid value.
Viktor Szakats [Sun, 25 Aug 2024 11:16:49 +0000 (13:16 +0200)]
GHA/configure-vs-cmake: add Windows build, fix issues
- configure: disable pthreads by default on Windows.
- configure: disable detecting `fseeko()` on Windows.
(It exists in mingw-w64 2.0.0 and newer, but it's permanently ignored
in CMake, as this function is never necessary on Windows.)
- extend existing exceptions with their Windows variants.
- `lib/formdata.c`: prioritize `_fseeki64()` over `fseeko()`.
To reduce the difference between Windows builds, which now all use
`_fseeki64()`.
- cmake: perm-enable `HAVE_DIRENT_H` and `HAVE_OPENDIR` for mingw-w64,
to match configure.
Follow-up to bfe54b0e88239da542493321e795cd71c14af9cc #13137
This in theory could make the dir listing feature work in mingw-w64
build, but in my tests (on WINE) it failed at the preceding `open()`
call.
- cmake: perm-enable `HAVE_STRINGS_H` and `HAVE_UTIME_H` for mingw-w64,
to match configure. (They are wrappers and make no difference in the build.)
Also:
- configure: sync `USE_MANUAL` macro with cmake, by only setting it for
`src`. Drop checker exception.
- CI: use `--disable-dependency-tracking` in existing jobs.
- CI: install packages before git checkout, in existing jobs.
Viktor Szakats [Sat, 24 Aug 2024 19:14:02 +0000 (21:14 +0200)]
build: check OS-native IDN first, then libidn2
If AppleIDN or WinIDN is selected, don't look for libidn2. Do this by
moving libidn2 detection after AppleIDN/WinIDN and skipping it if any
of them was selected.
Also:
- disable AppleIDN by default with autotools to sync behaviour with
CMake.
- limit WinIDN checks to native Windows with autotools, as with CMake.
Before this patch libidn2 was detected by default even if AppleIDN or
WinIDN was explicitly selected. libidn2 wasn't used in the build, but
it was left enabled as a dependency and appeared in `libcurl.pc` and
`curl-config`.
Viktor Szakats [Mon, 19 Aug 2024 19:05:25 +0000 (21:05 +0200)]
cmake: minor tidy-ups
- show `OpenSSL v3+` when detected (as in `./configure`).
(this string also makes its way to `curl-config`.)
- prefer `unset(VAR)` over `set(VAR)`.
Same effect, but `unset()` tells the intent unambiguously.
https://cmake.org/cmake/help/latest/command/set.html
- drop "implementation" from an `option()` description.
- FindGSS: replace legacy keyword alias with modern alternative.
https://cmake.org/cmake/help/latest/command/get_filename_component.html
- move `CURL_STATIC_CRT` logic next to its `option()`.
- improve order of `libcurl.pc`/`curl-config` variable init lines.
- tests: drop/shorten custom target names.
They inflated generated make files by 550KB.
Keep target name logic for sync between code snippets.
Follow-up to a2ef5d36b3cdec8120a4a9b782d18d58a70d4236 #14660
- clear a variable after use.
- restore `STATUS` for `Features:`/`Protocols:` `message()`s:
Without it the output goes to stderr, and appears in red in CMake GUI.
It doesn't seem possible to show a line on stdout without leading
underscores to match `curl -V` and `./configure` output.
Partial revert of acbc6b703f6b0ee568d053f6f2565fbc107b5fd3 #14197
- WindowsCache: move `HAVE_LINUX_TCP_H` into the header group.
- move strings to the same line as their `STRING` keyword.
- formatting in generated code.
- delete bogus comment.
- unfold lines for readability.
- fix a too long line. (for cmakelint)
- missing quotes, whitespace, comments.
Daniel Stenberg [Fri, 23 Aug 2024 12:31:33 +0000 (14:31 +0200)]
README: refresh
- call them "manpages", not "man pages" and drop the .1 and .3
- replace link to "manual" with a link to everything curl
- remove the link to the FAQ
- add a new "open source "subtitle for the license link, and rephrase
- mention and link GitHub services in the contact section
- rename the "git" subtitle to "source code"
- shorten the source code section somewhat
Dan Fandrich [Fri, 23 Aug 2024 23:09:17 +0000 (16:09 -0700)]
tests: tweak use of impacket in smbserver
The missing message only uses a single line now and has a consistent
prefix making it easier for log parsers to skip. Remove a precheck test
looking for impacket in test1451 because it's incomplete, unnecessary
and in the wrong layer; the test harness will automatically discover
that the smbserver won't start and the test was only checking a single
dependency. Also update the tests README to no longer mention Python 2.
The servers were switched to Python 3 several years ago and no attempt
is being made to maintain Python 2 compatibility
Viktor Szakats [Thu, 22 Aug 2024 07:10:07 +0000 (09:10 +0200)]
cmake: sync code between test/example targets
- reuse local variable names.
- sync `PROJECT_LABEL`, add where missing.
- namespace all target names.
- bind header directories to each target.
- tests/server: limit `CURL_STATICLIB` to Windows (as in autotools.)
- drop functions with a single caller.
Dan Fandrich [Fri, 23 Aug 2024 04:24:03 +0000 (21:24 -0700)]
CI: consolidate workflows for source and docs check
A bunch of tiny jobs that run various source or documentation checks are
consolidated into two workflow files: checksrc.yml and checkdocs.yml.
This reduces the proliferation of new files containing one-line checks
and brings those that operate similarly together for better reasoning
about them. The man-examples check is also now running again for the
first time in 7 months.
Various calls to find, xargs and git ls-files are changed where possible
to use NUL line terminators in pipes to avoid issues with oddly-named
files that might find their way into the repo.
Daniel Stenberg [Fri, 23 Aug 2024 08:41:26 +0000 (10:41 +0200)]
urlapi: verify URL *decoded* hostname when set
It was previously wrongly verifying the input in its URL encoded format
when setting the hostname component with curl_url_set(), so it wrongly
rejected '%'.
Now it URL decodes the name appropriately before the check.
Added tests to lib1560 to verify that a fine %-code is okay and that a
bad %-code (that decodes to '%') is rejected.
Viktor Szakats [Thu, 22 Aug 2024 15:05:54 +0000 (17:05 +0200)]
cmake: respect cflags/libdirs of native pkg-config detections
In Find modules with native pkg-config detection (libgsasl, libidn2,
libssh, libuv, nettle) use the C compiler flags returned by pkg-config.
Also use the library paths, and return the pathless library names.
Also:
- add these library paths to `libcurl.pc`/`curl-config`.
- fix libgsasl detection to use the detected header directory.
Viktor Szakats [Thu, 22 Aug 2024 14:14:44 +0000 (16:14 +0200)]
cmake/FindGSS: bring closer to other Find modules
- add pkg-config-specific 'Found' message to the Find module.
- update non-pkg-config 'Found' message to show the flavour instead of
the library name. (= the first value listed after `REQUIRED_VARS`)
- delete extra 'Found' message from `CMakeLists.txt`.
- rename internal result variables to match with default pkg-config
names, in preparation of introducing them in other Find modules:
- `GSS_LINK_DIRECTORIES` -> `GSS_LIBRARY_DIRS`
- `GSS_LINKER_FLAGS` -> `GSS_LDFLAGS`
- `GSS_COMPILER_FLAGS` -> `GSS_CFLAGS`
Daniel Stenberg [Wed, 21 Aug 2024 12:58:57 +0000 (14:58 +0200)]
test1521: verify setting options to NULL better
Previously this test allowed several error values when setting options.
This made this test miss #14629.
Now, errors are generally not accepted for setopts:
- numerical setopts accept CURLE_BAD_FUNCTION_ARGUMENT for funny input
- the first setopt to an option accepts CURLE_NOT_BUILT_IN or
CURLE_UNKNOWN_OPTION for when they are disabled/not built-in
- there is an allowlist concept for some return code for some variables,
managed at the top of the mk-lib1521.pl script
In curl.h: remove the OBSOLETE named values from the setopt list.
Chris Swan [Wed, 21 Aug 2024 13:26:59 +0000 (14:26 +0100)]
docs: Clarify OpenSSF Best Practices vs Scorecard
SECURITY.md has a recently added section titled OpenSSF Scorecard
that actually documents OpenSSF Best Practices. Scorecard [0] is a
different OpenSSF project, that incorporates Best Practices, but is
distinct in its objectives and how it achieves them.
This change clarifies the terminology, and also removes any
implication that Gold Best Practices is an award rather than a self
certification programme.
As curl was a leader in implementing Best Practices some folk may be
more familiar with the earlier Core Infrastructure Initiative (CII)
naming, so a reference to that has been added.
[0] https://scorecard.dev/
Signed-off-by: Chris Swan <478926+cpswan@users.noreply.github.com>
Ref: #14319
Closes #14635
Jan Venekamp [Wed, 21 Aug 2024 17:54:19 +0000 (19:54 +0200)]
tests: constrain http pytest to tests/http directory
Running the http pytest had to be done from tests directory or above,
because the repeat argument fixture was defined in tests/conftest.py.
However, the repeat argument is not needed because its functionality
can be provided by pytest-repeat as documented in the test's
README.md. So, removed the pytest_addoption function for the repeat
argument and the pytest_report_header function is moved to
tests/http/conftest.py.
TODO: Remove repeat argument from all tests. As a stopgap, a
one-element list is defined for it for now.
Viktor Szakats [Wed, 21 Aug 2024 22:21:51 +0000 (00:21 +0200)]
build: make `CURL_FORMAT_CURL_OFF_T[U]` work with mingw-w64 <=7.0.0
Add tweak for mingw-w64 when building tests/http/client programs to
avoid a bogus `-Wformat` warning when using mingw-w64 v7.0.0 or older.
The warning is bogus because these programs use curl's `printf()`
implementation that is guaranteed to support that format spec.
Add this for both CMake and autotools. (But only CMake is CI tested with
an old toolchain.)
Apply the workaround to `docs/examples`, and fix an example to use
curl's `printf()` with `CURL_FORMAT_CURL_OFF_T`.
Reintroduce curl `printf()` calls into `tests/http/client`, via #14625.
Also restore large number masks to a printf, changed earlier in #14382.
Viktor Szakats [Wed, 21 Aug 2024 11:36:29 +0000 (13:36 +0200)]
src: fix potential macro confusion in cmake unity builds
Sources used `lib/curlx.h` with both `ENABLE_CURLX_PRINTF` set and unset
before including it.
In a cmake "unity" batch where the first included source had it unset,
the next sources did not get the macros requested with
`ENABLE_CURLX_PRINTF` because `lib/curl.x` had already been included
without them.
Fix it by by making the macros enabled permanently and globally for
internal sources, and dropping `ENABLE_CURLX_PRINTF`.
This came up while testing unity builds with smaller batches. The full,
default unity build where all `src` is bundled up in a single unit, was
not affected.
Fixes:
```
$ cmake -B build -DCMAKE_UNITY_BUILD=ON -DCMAKE_UNITY_BUILD_BATCH_SIZE=15
$ make -C build
...
curl/src/tool_getparam.c: In function ‘getparameter’:
curl/src/tool_getparam.c:2409:11: error: implicit declaration of function ‘msnprintf’; did you mean ‘vsnprintf’? [-Wimplicit-function-declaration]
2409 | msnprintf(buffer, sizeof(buffer), "%" CURL_FORMAT_CURL_OFF_T "-",
| ^~~~~~~~~
| vsnprintf
curl/src/tool_getparam.c:2409:11: warning: nested extern declaration of ‘msnprintf’ [-Wnested-externs]
[...]
```
Reported-by: Daniel Stenberg
Bug: https://github.com/curl/curl/pull/14626#issuecomment-2301663491
Daniel Stenberg [Wed, 21 Aug 2024 21:35:31 +0000 (23:35 +0200)]
CURLOPT_XFERINFOFUNCTION: clarify the callback return codes
also done in CURLOPT_PROGRESSFUNCTION.md
This changes the wording to say that 1 should be returned instad of
non-zero to return error from the callback. I did this to simplify, even
if other non-zero values still do the same thing I figure we might just
as well just leave out the others from the documentation.
Fixes #14627 Reported-by: Benjamin Riefenstahl Mecom
Closes #14637
This value tells how many sources files to bundle in a single "unity"
compilation unit.
The CMake default is 8 sources, curl's CMake set this to 0, meaning
to bundle all sources into a single unit.
This patch makes it possible to override the 0 value, and potentially
optimize the build process further by better utilizing multiple cores
in conjunction with `make -jN`.
The number of sources in lib is 172 at the time of writing this. For
a 12-core CPU, this can give a job for them all:
`-DCMAKE_UNITY_BUILD_BATCH_SIZE=15`
(Compile time may be affected by a bunch of other factors.)
Viktor Szakats [Tue, 20 Aug 2024 13:44:28 +0000 (15:44 +0200)]
cmake: drop libssh CONFIG-style detection
Drop `find_package(libssh CONFIG)` detection method in favour of
the Find module that supports both `pkg-config`, and CMake-native
(since #14555) detection.
This aligns `libssh` detection with other dependencies. It makes the
build honor custom configuration via `LIBSSH_INCLUDE_DIR`,
`LIBSSH_LIBRARY`.
Also enable libssh in a GHA/macos cmake job for build coverage.
Fixing:
- curl-for-win requiring a hack to configure libssh:
https://github.com/curl/curl-for-win/blob/4f9acbed92fd4aac0e874c9a591bec7d621cd9f2/curl.sh#L255-L263
- after #14555, GHA/windows gnutls vcpkg job no longer auto-detected
libssh, due to a regression missing to enable libssh when
found via `find_package(libssh CONFIG)`.
Ref: https://github.com/curl/curl/actions/runs/10470138955/job/28994650338
Viktor Szakats [Tue, 20 Aug 2024 19:51:44 +0000 (21:51 +0200)]
version: fix shadowing a `libssh.h` symbol
```
/Users/runner/work/curl/curl/lib/version.c: In function 'curl_version_info':
/Users/runner/work/curl/curl/lib/version.c:584:15: error: declaration of 'ssh_buffer' shadows a global declaration [-Werror=shadow]
584 | static char ssh_buffer[80];
| ^~~~~~~~~~
In file included from /Users/runner/work/curl/curl/lib/vssh/ssh.h:35,
from /Users/runner/work/curl/curl/lib/urldata.h:185,
from /Users/runner/work/curl/curl/lib/altsvc.c:32,
from /Users/runner/work/curl/curl/bld/lib/CMakeFiles/libcurl_shared.dir/Unity/unity_0_c.c:4:
/opt/homebrew/include/libssh/libssh.h:99:35: note: shadowed declaration is here
99 | typedef struct ssh_buffer_struct* ssh_buffer;
| ^~~~~~~~~~
```
Ref: https://github.com/curl/curl/actions/runs/10477958747/job/29020250670#step:9:48
For libssh, it fixes a "unity" build issue where libssh deprecation
warnings were not suppressed before this patch, because the suppression
macro was only set before just one of the two `libssh.h` includes.
If the other was compiled first in unity mode, the warnings appeared.
Seen in local curl-for-win build (`CW_CONFIG=test-x64-libssh-quictls`)
with libssh 0.11.0. (Also in a GHA/macos cmake job upcoming in #14614)
Use this opportunity to drop duplicate SSH header includes from the SSH
modules. It's enough to include them via the common `ssh.h` header.
projects / thirdparty / curl.git / log
