It was impossible to specify null arguments in the exclude array,
while they were passed properly in the code, they were stringified
while the entries in the options array were just completely removed.
To fix this, do not early stringify the argument name and option
together as that would make it impossible to properly match it to
None without introducing a new special case for the string "None".
So instead they are now kept as tuples and only stringified when
its actually needed.
Since commit 32b14b1bb5, hotdoc is run
during configure as an external program. It turns out though, that in
some cases we passed NoneType in the cmd array, previously to
hotdoc.run_hotdoc.run() and now via subprocesses. The former "worked"
due to ignoring unknown arguments (?) but the latter was broken because
command line interfaces don't accept python NoneType objects, naturally.
End result: when for example building Meson's own documentation, this
fails with a python traceback.
The reason this happens to begin with turns out to be, once again,
because of the legacy debt of homebrewed kwargs parsing. We have a
function for `process_known_args` that handles args and ignores them if
they are NoneType, and then include_paths is handled via a custom
processor that internally adds them, then returns a *list* of NoneType
which then gets appended to the global cmd, because the logic ends up as
`[None, None] is None` which is a failed check, so we go ahead and add
it.
It's odd that we ever attempted to process it twice to begin with, so
let's simply not do that.
We are just using this dictionary to get keys, and we could also just
set it ourselves but with None values. But we have a code abstraction
for this already; use it.
We use a dummy project with a vague name and try to find flags for it
based on the installable pkg-config files. This sort of works,
generally, because it attempts to match `-lct` which doesn't exist
because the test case isn't installed, and that link argument is passed
directly through.
Except, sometimes that library does exist. It is provided by the
"freetds" project, which may be a dependency other tools. In that case,
Meson finds a library, and the `dependency()` resolves to
`/usr/lib/libct.so` and the test fails.
Fortunately, we do have an API to say that we really want to get back
the same flags pkg-config returned. Use this.
We test a couple ways of running the meson command, and double check
that the debug logs record the meson command used (including python).
But when running the meson command as an executable instead of as a
python script, we see whichever version `env` in the shebang sees.
Fix this by mocking the python command as well.
*.qrc files converted to C++ sources could make use of Qt headers, and
in practice for qt4 they seem to. Since this is Qt code to begin with,
it makes sense to depend on the Qt being targeted regardless of version.
We need to run it as an external command at build time anyway, and we
detect it by looking it up as an ExternalProgram. It seems odd to then
import it into Meson's python interpreter and run the main function.
Moreover, this errors out when you are running two different pythons,
one for Meson and one for hotdoc. For example, when hotdoc is installed
normally, but you're testing Meson against a nondefault newer version of
python.
We have divergent implementations of loading a pickled *.dat file. The
Build class loader has a better error message. But the generic loader
handles TypeError and ModuleNotFoundError. Merge the implementations,
and use it for Build as well.
Fixes#11051
Since commit 1420d0dace we use coredata's
cmd_line.txt handler to get the right setup arguments. But there's a bug
in that -- it mishandles cross/native files, producing invalid
descriptions of the command line. The only other place this was used,
though, is when generating meson-log.txt.
Fix it to produce correctly formatted arguments.
Fixes#10980
The coverage report was always the final section of the main test run.
This made it hard to scroll around and find exactly what went wrong --
particularly as not everyone realizes that coverage isn't part of the
test run, but also because the output from coverage is... excessively
long.
This mirrors what we do in our other workflows.
The system linkers don't support this particular GNU extension on
these OS'es, so don't build them there.
Based on an OpenIndiana patch created by @alarcher. Closes#9882.
In a couple of python module tests, we try to test things that rely on
the default python being the same one we look up in the python module.
This is unsolvable for the deprecated python3 module, as it actually
uses the in-process version of python for everything. For the python
module, we could actually look up the default system python instead of
the one we are running with, but then we wouldn't be testing the
functionality of that alternative python... and also the install
manifest tests would see files installed for the wrong version of
python, and report a combination of missing+extra files...
Solve both tests by just skipping the parts we cannot check.
- qt5 -> qt6
- remove version information from when the Qt6 module was not a thing
- linked to dependency function
- highlight version information with *...* and placing it at the front of options or on new lines in text
- reformatted for shorter lines
Fixes:
- Incorrect, redundant, or overabundant usage of "just"
- Missing punctuation
- Missing "the"
- Parenthesized text far from what it describes
There are some subjective changes, I hope those aren't controversial.
We consistently use the "dependencies" kwarg to refer to C-like
CFLAGS/LDFLAGS interfaces. And for hotdoc, we actually accept libraries
for this as well, as we may want to document their (generated?) sources,
so we want their CFLAGS too.
But we also accepted custom targets and just added a build order
dependency on these, which was odd and typically we call that "depends".
Let's deprecate this in favor of the depends kwarg.
Internally we pass this as extra_depends to the CustomTarget
initializer, so it makes sense to call it that rather than confusing the
topic by referring to "dependencies", a term that indicates
CFLAGS/LDFLAGS interfaces.
We would like to use the default value (the one python urged us not to
use), but without getting a warning for it. Luckily, we have the correct
value already, so we can pass it manually to avoid the complaint.
python 3.11 adds a warning that in 3.15, UTF-8 mode will be default.
This is fantastic news, we'd love that. Less fantastic: this warning is
silly, we *want* these checks to be affected. Plus, the recommended
alternative API would (in addition to warning people when UTF-8 mode
removed the problem) also require using a minimum python version of 3.11
(in which the warning was added) or add verbose if/else soup.
The simple, and obvious, approach is to add a warnings filter to hide
it.
It is, after all, "safe". ;) That's why it exists. There's no reason to
think listing all pkg-config entries cannot print unicode descriptions,
it's absolutely possible, and we should handle it properly if we
encounter it.
Marvin Scholz
Jussi Pakkanen
Eli Schwartz
Dylan Baker
Zihua Wu
Xeonacid
Xavier Claessens
Tristan Partin
Alan Coopersmith
tastytea
Hagen Möbius
Ralf Gommers
Kleis Auke Wolthuizen
Guilherme Janczak
Will Thompson
Denis Fortin
Ole André Vadla Ravnås