Sometimes, the machine file can include compiler command line options,
in order to pick the correct multilib. For example, Meson uses "$cc
--print-search-dirs" to find the library search path, where $cc is the
cc from the machine file. Because the outputs of "gcc -m32
--print-search-dirs" and "gcc --print-search-dirs" are different, this
only works if you have
[binaries]
cc = ['gcc', '-m32']
in the machine file. Right now, however, the cmake module assumes that
the compiler listed in the machine file is either a compiler, or a
"launcher" followed by the compiler. Check if the second argument
starts with a slash (for Microsoft-like compilers) or a dash (for
everyone else), and if so presume that the CMAKE_*_COMPILER_LAUNCHER
need not be defined.
install_dir needs to be set to something, because CustomTarget expects
that. genmarshal still relies on CustomTarget setting it internally
itself.
Fixes#9350
Theere is no 'output' key in the kwargs, so if this was it it would
result in an exception. Mypy spotted this in a branch I'm working on
full typing for gnome, but since that seems unlikely to be done before
the 0.60 branchpoint I'd like to get this in sooner
If either 'name' or 'name_formatted' are absolute paths,
then they are of the form:
C:\path/to/file.ext (example)
By only substituting slashes with the underscore, we get:
C:_path_to_file.ext
This is still not quite right, infact os.path.basename()
returns:
_path_to_file.ext
Also replace colons with underscores to overcome issues when
absolute paths are involved.
Yelp currently can take sources two different ways, the first is via
variadic arguments, the second is by a keyword argument. If the keyword
is passed then the variadic arguments are silently ignored, which is
obviously not ideal. Fortunately the variadic form was never documented,
and is likely not in wide use.
This patch fixes it by deprecating the variadic form, and warning if
both are passed. It does not change behavior as someone may be relying
on it.
This is the final refactoring for extracting the bultin object
logic out of Interpreterbase. I decided to do both arrays and
dicts in one go since splitting it would have been a lot more
confusing.
CMakes `target_link_libraries()` supports certain keywords to
only enable specific libraries for specific CMake configurations.
We now try our best to replicate this for Meson dependencies.
Fixes#9197
There are two possible issues, both of which emit very long messages
from git:
- when shallow cloning via depth + a pinned commit we locally init
rather than cloning; use an almost-certainly not conflicting dummy
branch name
- any time a checkout of a revision is performed, it might not be a
branch name, in which case it will be a detached HEAD. By default git
is very noisy about this -- it wants you to know what happened and how
not to mess up. But wraps aren't per default intended for user editing
and development, it's just part of the internal transport which meson
uses. So, temporarily squelch this advice. Do not permanently
configure the repo like this though, because the user *might* cd in
and start developing on the subproject; in this case, any additional
git advice they trigger is their responsibility (and they probably do
want it).
Fixes#9318
To be able to handle link_depends in backends that do not just operate on a
file basis like ninja, information about the targets, not just their output is
required.
Imported both inside and outside of T.TYPE_CHECKING, the runtime import
can be removed by quote-deferring one of the use sites.
Update:
In between then and now, this got removed from T.TYPE_CHECKING, move it
back there rather than preserving the runtime import.
(var,) is the correct way to pass values to a percent formatted string,
but not to .format so we would end up printing something like:
unexpected input at line (4,)
Upgrade to an f-string and insert the correct value correctly.
'{}'.format('foo') for any given value of 'foo' (in this case, a
function returning a string), can always just be 'foo' directly, which
is a lot more readable.
All changes were created by running
"pyupgrade --py3-only"
and committing the results. Although this has been performed in the
past, newer versions of pyupgrade can automatically catch more
opportunities, notably list comprehensions can use generators instead,
in the following cases:
- unpacking into function arguments as function(*generator)
- unpacking into assignments of the form x, y = generator
- as the argument to some builtin functions such as min/max/sorted
Also catch a few creeping cases of new code added using older styles.
We wrote:
ERROR: install_emptydir keyword argument "install_mode" permissions string must
be exactly 9 characters, got "4" in the form rwxr-xr-x
Let's change this around to be easier to read.
Also, 1-based numbering was used (for components) and 0-based for "bits".
And actually the "bits" are not bits, but octal digits. So say "permissions
character 1", "permissions character 2".
And finally change "must be … if provided" to "can only be". (If it isn't
provided, it "is not", so the sentence is still valid. The user will only
get this error if they provide something, so we don't need to be super precise
and say "if provided". And then we avoid confusing the reader whether
it's "if provided" attaches to the the "False" or to the whole sentence.)
The previous commit bd2fcb268b
accidentally used the wrong var so the param name was missing,
leading to an error of "unrecognized arguments" for the
datadirs parameter value.
We ended up passing a NoneType as the program binary, which is a very
awkward way to communicate an error when failing to write a basic
package version file.
There were two allowed kwargs that were unused, I've dropped them for
now, though I need to check if they were documented. If they are then we
need to deprecate them to remove them, otherwise we'll just leave them
removed.
Paolo Bonzini
Dylan Baker
Luca Bacci
Vladimír Čunát
Daniel Mensinger
Jesse Natalie
Eli Schwartz
Andres Freund
Zbigniew Jędrzejewski-Szmek
Dan Streetman
Jussi Pakkanen