| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The code for this exists and works, but the type annotations don't allow
it. This fixes the annotations
|
|
There is no reason for these inititializers to exist, all they do is
defer to the parent initializer. Worse, since they are not type
annotated thy prevent the parent type annotations from being used
|
|
|
|
|
|
As we now require python 3.6, we can declare their types without
initializing them.
|
|
This reverts commit c89aa2094170b2ffd7151187c1c092db2a178f44.
We no longer support 3.5 so this can go.
|
|
Two tests are failing on Cygwin because the argument is passed as
a long-path and the Path is ending up as a short-path:
AllPlatformTests.test_run_target_files_path
Traceback (most recent call last):
File "/cygdrive/d/a/meson/meson/test cases/common/51 run target/check-env.py", line 22, in <module>
assert build_root == env_build_root
AssertionError
SubprojectsCommandTests.test_purge
> self.assertEqual(deleting(out), sorted([
str(self.subprojects_dir / 'redirect.wrap'),
str(self.subprojects_dir / 'sub_file'),
str(self.subprojects_dir / 'sub_git'),
]))
E AssertionError: Lists differ: ['/cygdrive/c/Users/runneradmin/AppData/Local/Temp/tmpeaa2a49[205 chars]git'] != ['/cygdrive/c/Users/RUNNER~1/AppData/Local/Temp/tmpeaa2a49z/s[196 chars]git']
[...]
['/cygdrive/c/Users/runneradmin/AppData/Local/Temp/tmpeaa2a49z/src/subprojects/redirect.wrap',
^^^^^^^^^^^
['/cygdrive/c/Users/RUNNER~1/AppData/Local/Temp/tmpeaa2a49z/src/subprojects/redirect.wrap',
^^^^^^^^
The fix is to not use the tempdir for all tests, but only for tests
that check the mode.
|
|
fixes #8721
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Typing for the Windows module
|
|
|
|
|
|
setup_vsenv() was not propagating the return value, so build.need_vsenv
was always False.
|
|
install: false was only available since 0.50, so we should not warn
people who support older versions to use something they cannot.
Fortunately, we can do FeatureDeprecated for this -- and then it even
gets summarized. Unfortunately, it's not well used, and certainly isn't
here.
|
|
|
|
It turns out this could be missing in GResource*Target as well, due
mostly to the same problem, side effects of mutating a shared
dictionary; though it could also happen with a specific set of keywords
given and other omitted.
Fixes #9350
|
|
Unless parsing TAP output, there is no strict requirement for
"meson test" to process test output one line at a time; it simply
looks nicer to not print a partial line if it can be avoided.
However, in the case of extremely long lines StreamReader.readline
can fail with a ValueError. Use readuntil('\n') instead and
just process whatever pieces of the line it returns.
Fixes: #8591
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
|
|
The `mkenums` functions can have this unset if, and only if, the
c file only variant is called. Due to side effects if the header file is
generated then `install_dir` is ensured to be set for the c file. I have
removed this side effect so that our tests actually cover this case.
Fixes #9472
|
|
Fixes: #9454
|
|
still find the correct linker and now linux will be able to use its
xc16-gcc as the linker.
|
|
Fixes: #9437
|
|
We could have an OptionOverrideProxy of an OptionOverrideProxy,
recursively. This fix is a minimal subset of the refactoring I did in
https://github.com/mesonbuild/meson/pull/9394. Instead of faking
UserOption we can just do a shallow copy of one and set a new value on
it.
Fixes: #9448
|
|
both of these do the same thing:
init -b <branchname>
-c init.defaultBranch=<branchname> init
The latter contributes to unreadably long lines of code, but has the
advantage of working on older versions of git.
Fixes #9449
|
|
The broken list was introduce by commit f72ee8e7 which fix too long lines.
|
|
|
|
We already have this magic string in mesonlib, and this should always
have used the cmake@ format which is identical to the meson format other
than the regex.
|
|
This reverts commit 6cc1b8441c0cf7428e52bdf1cd541ea830a4eb83.
The latest version of pyflakes learned to detect that correctly.
|
|
|
|
|
|
we return _log even though this entire family of functions returns None,
because a side effect of returning is that the other version of the
function is not run.
We can do that more obviously, using an else clause that doesn't attach
meaning to return values.
|
|
assertTrue and assertFalse are recommended against, if you can get a
more specific assertion. And sometimes it is considerably shorter, for
example we have a custom assertPathExists which we can take advantage
of.
|
|
Constructing a PackageDefinition is enough to assert that it raises an
error.
|
|
It never made sense here to save self.init() which returns a string
containing a log or stdout or something, and which was never actually
used.
Also we then overwrote the variable with a pathname...
|
|
In commit d932cd9fb48e8fecb167641c3d6bee8a68831c6e, we migrated to
meson's own static linker definition, and the old code that hardcoded
two of the possible exelists should have been removed in the process.
|
|
It's redefined on every loop iteration, and as the comment says, we just
want to make sure the next loop skips a value.
|
|
|
|
Building documentation fails with older hotdoc versions.
|
|
It has always been working even if not documented and there is no reason
to not accept it. However, change "True/False" to "true/false" to be
consistent with meson language.
Fixes: #9436
|
|
The dep.name() function schould always return the name of the
dependency as documented. No matter if it was found or not.
https://mesonbuild.com/Reference-manual_returned_dep.html#depfound
|
|
It's not a python file, so it will never end up in the installed package
unless we mark it as package_data. This causes problems for people using
non-git checkouts.
Fixes #9435
Closes #9443
|
|
|
|
Fixes: #9444
|
|
|
|
|
|
|
