Issue #15763 is again mostly false positive crashes of the unit tests
because the code is only partly GIMP code, arranged differently as what
a full GIMP process would actually do.
The new tests just runs on the real GIMP code. Now the difference is
that we don't test the UI by doing this, but this can be debated whether
the previous tests were actually running on the UI themselves (they were
mostly running some core code directly and sometimes activating some
actions or raising dialogs (gimp_test_utils_create_image_from_dialog()),
but again not by actually clicking or hitting keys as a human would do.
It's not proper UI testing (cf. #9339 of #13376 for further discussions
on this topic).
So in the meantime, let's go with this intermediate step. At least now,
such tests would run actual GIMP code and would catch issues which could
really happen in a GIMP process.
This is a followup of previous commit. We must set the win_subsystem
option on executable() so that the result binary is compiled as a GUI
application (and doesn't output a console every time).
The previous commit is still needed and is what allows us to control
when to actually display a console.
On Windows, it will basically be a copy of the main binaries. On other
OSes, it will be symlinks.
This will help for Windows installers where we want to use major version
only in registry and we need a binary named appropriately for this
apparently.
As for the "major" version, I will in fact use the next major when minor
is 99 (also called API major version), not that it will matter much very
soon!
Note that the executables are not built yet at this point, but we just need to
pass the name for configuration. The gimp.ico generation step is run manually
anyway and requires a fully functional and installed GIMP.
See commit gimp-data@40d4822.
Since we are already building a static (build-only) libapp, to be used with
tests or tools, let's also use it for the main (GUI) executable.
For the CLI executable though (gimp-console-2.99), we must re-compile
libapp_sources from scratch anyway, because we define the additional macro
'GIMP_CONSOLE_COMPILATION'.
Reviewer's message (Jehan): This was a work-in-progress by Niels, which
we only keep in this state because Lukas worked over it. I have rebased
and fix-amended many broken part of this commit, because various things
had been changed in these areas of code since this commit was initially
written.
Now that we bumped our meson requirement, meson is complaining about
several features now deprecated even in the minimum required meson
version:
s/meson.source_root/meson.project_source_root/ to fix:
> WARNING: Project targets '>=0.56.0' but uses feature deprecated since '0.56.0': meson.source_root. use meson.project_source_root() or meson.global_source_root() instead.
s/meson.build_root/meson.project_build_root/ to fix:
> WARNING: Project targets '>=0.56.0' but uses feature deprecated since '0.56.0': meson.build_root. use meson.project_build_root() or meson.global_build_root() instead.
Fixing using path() on xdg_email and python ExternalProgram variables:
> WARNING: Project targets '>=0.56.0' but uses feature deprecated since '0.55.0': ExternalProgram.path. use ExternalProgram.full_path() instead
s/get_pkgconfig_variable *(\([^)]*\))/get_variable(pkgconfig: \1)/ to
fix:
> WARNING: Project targets '>=0.56.0' but uses feature deprecated since '0.56.0': dependency.get_pkgconfig_variable. use dependency.get_variable(pkgconfig : ...) instead
- Setting an exec_dir variable is an error. As meson docs says, if
relative, it is installed relatively to prefix anyway: "If this is a
relative path, it is assumed to be relative to the prefix."
On the other hand, it would make problems if someone tried to set an
absolute bindir.
Moreover it is a lot clearer now. When we want to install in the
binary directory unconditionally, then get_option('bindir') is the
meson way, hence the way to go.
- On the other hand, the `gimp-debug-tool` is installed either in bindir
for Windows or macOS and libexecdir for all other platforms, at least
that's how it's set in the autotools build. So let's keep both builds
consistent.
- Make a hopefully clearer description for enable-default-bin option.
Let's clarify this is just about creating unversionning links pointing
to versionned files.
- Adding an item in the "Optional Features" part of the summary listing
during meson configure, for better discovery.
For the ".exe" extension on Windows, I wished we had an $(EXEEXT)
equivalent on meson rather than trying to set it ourselves (are there
other platforms where we must set a different extension?). But I could
not find any.
Our meson build system was not properly building the enums.c file,
because they are versionned.
I did a similar trick as what I did for the pdbgen, which is that I used
a wrapper script around the existing perl script, which sets proper
options and generate a stamp file in the end (which is considered by
meson as the actual custom target, not the C file since it is generated
in the source dir).
The most important part is that the stamp file is a generated header
source (not just a random text file) which is **included** by the
generated C file. This is what will force meson to regenerate the C file
if the header is updated, **then** build using this new version, not use
an outdated versionned version (which would make for hard to diagnose
bugs), through the indirection of the intermediate stamp header.
See #4201.
See also: https://github.com/mesonbuild/meson/issues/10196#issuecomment-1080742592
These changes make it possible to build against homebrew for mac
libraries. Homebrew stores it's libraries in seperate folders and so
blanket includes were not working.
PYTHONPATH is not needed and interferes with homebrew build.
This is untested on my side, because the bug only happens on native
builds with meson (our CI has cross-builds with meson and native builds
with autotools and I only do cross-builds locally) but I think/hope it
will work.
Basically we were using .full_path() because these rc files were also
used as input of some configure_file() calls which doesn't like custom
target objects as input (it wants strings or file objects). Yet a bug
in meson didn't like the colon used in native Windows full paths ('C:'
and such) when used in windows.compile_resources(). This has been fixed
by Luca Bacci in: https://github.com/mesonbuild/meson/pull/9368
Yet we just cannot depend on very early meson (or worse dev meson code).
On the other hand, if the input is a custom_tgt object, it uses the
object ID which we give as first parameter of custom_target() so we know
it's appropriately named without colons (such as 'gimp_plugins_rc').
Thus we should not bump into this issue again.
For the few usage in configure_file(), I just add a .full_path() only
when needed at call time.
Last but not least, I replace the bogus `meson --version` call by a
`python3 -c 'exit()'` as advised by Eli Schwartz:
2afa019c70 (note_1284951)
The reason is that it is apparently possible (or will be when some
reimplementation of meson will be done) that the `meson` executable
itself does not exist. On the other hand, `python3` should always be
there, as a mandatory dependency of the build tool.
In order to use an appropriate `python3`, I made the
pythonmod.find_installation() check required in our build (which should
not be a problem since it's a meson requirement as well), even when the
-Dpython option is false (this one depends on other requirements too
anyway, such as version and pygobject). This way I can call this meson
variable of discovered python in my bogus call, instead of calling a
(potentially different) python from PATH environment.
Meson has a very useful tool for combining a set of linked libraries and
compiler arguments (like an internal pkg-config):
`declare_dependency()`. Use this command to ensure that we can easily
reuse libapp* over multiple libraries, executables and tests.
That way, if we also add a new dependency later on in one of the libapp
libraries, this won't need to be added multiple times.
The tests weren't building because they weren't linked to the static
libraries.
Note that some tests still need fixing, but at least they can be
built/run now.
More of the files were wrong, or at least not absolutely identical to
the files generated by the autotools. I am not doing any code change
other than trying to make both build systems produce identical files
(except for slight differences on 2 files not worth the effort) even
though maybe some things can be improved (especially on the include
list). Maybe to be improved later.
Also fixing 2 of the previously autotools-generated files because of
space typos which should have been committed earlier.
Finally it is to be noted that there is no logics to copy the generated
files back to the source directory in the meson rules. I am not sure
anyway this is really worth it and maybe we should just stop tracking
these generated files eventually.
