- one patched to ignore failure... but it doesn't fail?
- one upstream patch removing an apparently useless test
Closes: https://bugs.gentoo.org/933450
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
There is more than just an ebuild :) let's get rid of remaining Manifest
/ metadata.xml as well.
Fixes: 88be015150
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
When restoring old ebuild an extra old distfile came along for the ride
and I failed to regenerate it.
Fixes: 9735a21b66
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
Use the magic value of 1 which avoids invoking coredump handlers
like systemd-coredumpd which can be really slow when we have many
intentional or expected crashes.
This trick was suggested by Alexander Monakov.
Signed-off-by: Sam James <sam@gentoo.org>
No tests, but based on smoketesting it seems to work fine. Some
SyntaxWarnings exist when byte compiling but it works for current python
versions.
Closes: https://bugs.gentoo.org/929887
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
py3.13 is not ready, because deps aren't. But, the same number of tests
pass on either 3.11 or 3.12.
We could upgrade to 3.13 if dev-python/arsenic supported it, and locally
bumped, the wapiti tests pass. But, arsenic has no tests so I'm hesitant
to commit a bump for it.
Bug: https://bugs.gentoo.org/952156
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
Now that tests kinda sorta ***work***, we can see by testing, that we
have bad dependencies installed. This project is NOT compatible with
current httpx. The version cap results in around a hundred test failures
disappearing. Some of them start failing instead because of network
sandboxing, so add PROPERTIES for this.
We're down to two failures:
```
===================================== short test summary info =====================================
FAILED tests/attack/test_mod_log4shell.py::test_verify_dns - ValueError: nameserver is not a dns.nameserver.Nameserver instance or text form, IP address, nor a valid https URL
FAILED tests/cli/test_options.py::test_options - AssertionError: assert 11 == (11 + 1)
===================== 2 failed, 231 passed, 96 warnings in 120.44s (0:02:00) ======================
* ERROR: net-analyzer/wapiti-3.1.7-r1::gentoo failed (test phase):
* pytest failed with python3.11
```
Bug: https://bugs.gentoo.org/952156
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
> Many tests require execution of local test php server
is surely not true, it simply needs dev-lang/php installed. So, there's
no earthly reason to restrict it.
Once we try running tests, there is a cascade of failures. The biggest
one is the source tarball doesn't come with them -- perhaps this would
be noticed if we tried running them, eh?
There are a *bunch* of other missing test deps too.
And while we are here -- remove the need for pytest-cov via sed on
"addopts". There's never a good reason to depend on linting plugins just
because pytest.ini (or equivalent) enables it by default in CI.
Bug: https://bugs.gentoo.org/952156
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
Partially. And yeah I hate this. But:
restore 0.27 in an effort to get an absolutely trash package to pass
tests. :)
This reverts commit 1a1f0048a0.
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
sourceforge's project page is a meta-refresh for github pages, and the
README lists the latter. Both sites are used for release tarballs, so
list them both for redundancy.
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
gtkmm:2.4 is a C++ wrapper around gtk+:2, and installing the former
implies the latter. It's redundant to specify both, and also incorrect
since the latter is never actually used directly. qa-vdb even points out
it isn't used.
Same logic applies to cairo vs cairomm, but this package only listed the
(unused) C part. This is fine, because gtkmm noisily adds it itself, but
qa-vdb does point out cairomm is used.
No revbump because this makes no difference at either build or install
time. The deps are there regardless, but lint tools are a bit happier is
all.
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
The package performs crimes against technology: namely, it builds with
waf. This is the only reason it needs python at all.
This necessitates replacing bundled waf with a more modern vesion, since
the bundled version was already crudely patched to support py3.11
instead of updating waf, and it immediately broke again for 3.12. The
current replacement waf is used elsewhere in ::gentoo, and is tested to
install the same using any python from 3.11-3.14.
Closes: https://bugs.gentoo.org/929650
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
As we anyways use `domodule` there is no point permitting it to seek
site-package using legacy distutils, only to discard the info.
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
I didn't mean to push this as stable, though it's not a big deal as
the changes are really pretty small and the branch is winding down (EOL
shortly). Pointed out by Eli.
Signed-off-by: Sam James <sam@gentoo.org>
Switch to EPYTEST_PLUGINS just anyway but also because noseofyeti
broke tests for me as I happened to have it installed:
> file /var/tmp/portage/net-misc/streamlink-7.5.0/work/streamlink-7.5.0/tests/cli/test_argparser.py, line 355
> @pytest.fixture()
> def session(self, session: Streamlink, console: Mock, parser: ArgumentParser, plugin: type[Plugin]):
> E fixture 'self' not found
Signed-off-by: Sam James <sam@gentoo.org>