This fixes regression caused by
3162b901ca
that changes the order in which libraries are put on the link command.
In addition, that commit was wrong because libraries from dependencies
were processed before process_compiler() is called, which that commit
wanted to avoid.
We need a union of the compilers used by the target, and by those used
by all dependencies, not all the compilers in all projects. We do,
however, need to look compilers up from all possible compilers, as a
dependency that is a subproject could use a language not present in the
current project.
This saves on a 1500-line import at startup and may be skipped entirely
if no compiled languages are used. In exchange, we move the
implementation to a new file that is imported instead.
Followup to commit ab20eb5bbc.
To take good decisions we'll need to know if we are a Rust library which
is only know after processing source files and compilers.
Note that is it not the final list of compilers, some can be added in
process_compilers_late(), but those are compilers for which we don't
have source files any way.
Case 1:
- Prog links to static lib A
- A link_whole to static lib B
- B link to static lib C
- Prog dependencies should be A and C but not B which is already
included in A.
Case 2:
- Same as case 1, but with A being installed.
- To be useful, A must also include all objects from C that is not
installed.
- Prog only need to link on A.
This allows changing the crate name with which a library ends up being
available inside the Rust code, similar to cargo's dependency renaming
feature or `extern crate foo as bar` inside Rust code.
The library names are directly mapped to filenames by meson while the
crate name gets spaces/dashes replaced by underscores. This works fine
to a certain degree except that rustc expects a certain filename scheme
for rlibs that matches the crate name.
When using such a library as a dependency of a dependency compilation
will fail with a confusing error message.
See https://github.com/rust-lang/rust/issues/110460
We need to remember its value when reconfiguring, but the Build object
is not reused, only coredata is.
This also makes CLI more consistent by allowing `-Dvsenv=true` syntax.
Fixes: #11309
The current check results in *any* value to `export_dynamic` generating
vala import targets, even `false`. This is pretty clearly wrong, as it
really wants to treat an unset export_dynamic as false.
Generated objects can already be passed in the "objects" keyword argument
as long as you go through an extract_objects() indirection. Allow the
same even directly, since that is more intuitive than having to add them
to "sources".
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hook this up to installed dependency manifests. This is often needed
above and beyond just an SPDX string -- e.g. many licenses have custom
copyright lines.
At least, if you tried to use it when passing an install_dir. Because
T.Sequence is horrible and we should never use it, and the annotations
are a lie that produces bugs.
So, fix the annotations on CustomTarget to never allow this to happen
again, and also fix the function too. Move some definitions elsewhere
inline to satisfy the linter.
Fixes#11157
There are two problems here: a typing problem, and an algorithm problem.
We expect it to always be passed to CustomTarget() as a list, but we ran
list() on it, which became horribly mangled if you violated the types
and passed a string instead. This caused weird*er* errors and didn't
even do anything. We want to do all validation in the interpreter,
anyway, and make the build level dumb.
Meanwhile we type it as accepting a T.Sequence, which technically
permits... a string, actually. This isn't intentional; the point of
using T.Sequence is out of a misguided idea that APIs are supposed to be
"technically correct" by allowing "anything that fulfills an interface",
which is a flawed concept because we aren't using interfaces here, and
also because "technically string fulfills the same interface as a list,
if we're talking sequences".
Basically:
- mypy is broken by design, because it typechecks "python", not "what we
wish python to be"
- we do not actually need to graciously permit passing tuples instead of
lists
As far as historic implementations of this logic go, we have formerly:
- originally, typeslistified anything
- switched to accepting list from the interpreter, redundantly ran list()
on the list we got, and mishandling API violations passing a string
(commit 11f9638035)
- switched to accepting anything, stringlistifying it if it was not
`None`, mishandling `[None]`, and invoking list(x) on a brand new list
from stringlistify (commit 157d438835)
- stopped stringlistify, just accept T.List[str | None] and re-cast to
list, violates typing because we use/handle plain None too
(commit a8521fef70)
- break typing by declaring we accept a simple string, which still
results in mishandling by converting 'foo' -> ['f', 'o', 'o']
(commit ac576530c4)
All of this. ALL of it. Is because we tried to be fancy and say we
accept T.Tuple; the only version of this logic that has ever worked
correctly is the original untyped do-all-validation-in-the-build-phase
typeslistified version.
Let's just call it what it is. We want a list | None, and we handle it too.
When auto-generating e.g. a `clang-format` target, we first check to see
if the user has already defined one, and if so we don't bother creating
our own. We check for two things:
- if a ninja target already exists, skip
- if a run_target was defined, skip
The second check is *obviously* a duplicate of the first check. But the
first check never actually worked, because all_outputs was only
generated *after* generating all utility rules and actually writing out
the build.ninja file. The check itself compares against nothing, and
always evaluates to false no matter what.
Fix this by reordering the target creation logic so we track outputs
immediately, but only error about them later. Now, we no longer need to
special-case run_target at all, so we can drop that whole logic from
build.py and interpreter.py, and simplify the tracked state.
Fixes defining an `alias_target()` for a utility, which tried to
auto-generate another rule and errored out. Also fixes doing the same
thing with a `custom_target()` although I cannot imagine why anyone
would want to produce an output file named `clang-format` (unless clang
itself decided to migrate to Meson, which would be cool but feels
unlikely).
Which adds the `use-set-for-membership` check. It's generally faster in
python to use a set with the `in` keyword, because it's a hash check
instead of a linear walk, this is especially true with strings, where
it's actually O(n^2), one loop over the container, and an inner loop of
the strings (as string comparison works by checking that `a[n] == b[n]`,
in a loop).
Also, I'm tired of complaining about this in reviews, let the tools do
it for me :)
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
This reverts commit c94c492089.
This broke propagated deps as well as PCH in MSVC, and has not been
fixed in time for the final release of 0.64.0, so it needs to be
reverted and then brought back later.
Fixes#10745Fixes#10975
This introduce a new type of BuildTarget: CompileTarget. From ninja
backend POV it is the same thing as any other build target, except that
it skips the final link step. It could be used in the future for
transpilers too.
Those classes are used by wrapper scripts and we should not have to
import the rest of mesonlib, build.py, and all their dependencies for
that.
This renames mesonlib/ directory to utils/ and add a mesonlib.py module
that imports everything from utils/ to not have to change `import
mesonlib` everywhere. It allows to import utils.core without importing
the rest of mesonlib.
What happens is this:
- liba is a convenience static library
- libb is an installed static library
- libb links in liba with --link-whole
- libc links to libb
- we generate a link line with libb *and* liba, even though libb is a
strict superset of liba
This is a bug that has existed since the we stopped using link-whole to
combine convenience libraries, and to instead propagate their
dependencies up. For most linkers this is harmless, if inefficient.
However, for apple's ld64 with the addition calling `ranlib -c`, this
ends up causing multiple copies of symbols to clash (I think that other
linkers recognize that these symbols are the same and combine them), and
linking to fail.
The fix is to stop adding libraries to a target's `link_whole_targets`
when we take its objects instead. This is an all around win since it
fixes this bug, shortens linker command lines, and avoids opening
archives that no new symbols will be found in anyway.
There are two distinct cases here that need to be considered. The first
issue is https://github.com/mesonbuild/meson/issues/10723 and
https://github.com/mesonbuild/meson/issues/10724, which means that Meson
can't actually generate link-whole arguments with rust targets. The
second is that rlibs are never valid candidates for link-whole anyway.
The promotion happens to work because of another bug in the promotion
path (which is fixed in the next commit).
Generally plumb through the values of get_option() passed to
install_dir, and use this to establish the install plan name. Fixes
several odd cases, such as:
- {datadir} being prepended to "share" or "include"
- dissociating custom install directories and writing them out as
{prefix}/share/foo or {prefix}/lib/python3.10/site-packages
This is the second half of #9478Fixes#10601
Do to a bug in which a method, rather than an attribute, was checked
we would never detect that a target had no sources, even though we meant
to. As a result we would never actually error out when a target has no
sources. When fixing the check, I've changed the error to a warning
so that existing projects will continue to build without errors, only
warnings.
In order to reliably link to static libraries or individual object
files, we need to take their languages into account as well. For static
libraries this is easy: we just add the static library's list of
compilers to the build target. For extracted objects, we need to only
add the ones for the objects we use.
But we did this really inefficiently -- in fact, downright terribly. We
iterated over all source files from the extracted objects, then tried to
look up a new compiler for them. Even though the extracted objects
already had a list of compilers! This broke once compilers were made
per-subproject, because while the extracted objects have a reference to
all the compilers it needs (just like static archives do, actually) we
might not actually be able to look up that compiler from scratch inside
the current subproject.
Fix this by asking the extracted objects to categorize all its own
sources and return the compilers we want.
Fixes#10579
Regression in commit 7c757dff71.
SubprojectHolder is no longer an ObjectHolder and says so via a TODO:
this means that we have to fiddle with held_object. Yay.
The `add_deps` function did not behave correctly when a specified
dependency is not an instance of `dependencies.Dependency`.
Reorder the logic flow to perform this validation first.
Fixes#10468
Instead of asking the ExtractedObjects, but with a hook back into the backend,
use the existing function in the backend itself. This fixes using the
extract_objects(...) of a generated source file in a custom_target.
It should also fix recursive extract_all_objects with the Xcode backend.
Fixes: #10394
A single target could be picked for unity build, and in that case
extract_objects() should not be allowed.
Likewise for the opposite case, where extract_objects() should be allowed
if unity build is disabled for a single target. A test that covers that
case is added later.
CustomTarget.is_linkable_target has '.dylib'.
See also 93b1d31af9 that added '.dylib'
to CustomTarget.is_linkable_target but didn't add '.dylib' to
CustomTargetIndex.is_linkable_target.
There are a couple issues that combine to make the current handling a
bit confusing.
- we call it "install_dir_name" but it is only ever the class default
- CustomTarget always has it set to None, and then we check if it is
None then create a different variable with a safe fallback. The if is
useless -- it cannot fail, but if it did we'd get an undefined
variable error when we tried to use `dir_name`
Remove the special handling for CustomTarget. Instead, just always
accept None as a possible value of outdir_name when constructing install
data, and, if it is None, fall back to {prefix}/outdir regardless of
what type it used to be.
This reverts commit 5f02d0d9e1.
Which isn't correct, we have very strange behavior of "force on pie/pic
or let the toolchain do whatever it wants, but you can't turn it off."
Calling `isinstance(self, X)` is an anti-pattern, we should just be using
inheritance for this, letting the `StaticLibrary` override the method, and
having the base class always return `False`.
Since they are actually dependencies out the output not the Generator
itself.
This fixes dependency issues in the ninja backend, allowing Meson to
rebuild more accurately. It also does sometimes in the vs backend, but
there are problems in the vs backend I'm not sure how to solve. The
vsbackend is, itself, so fragile looking I don't want to get too
involved with it.
Currently, the code puts a placeholder in for the first output, then
replaces all of the outputs when it generates final value. Instead,
let's only replace the placeholder value.
override_options makes no sense for custom_target as we don't use it for
anything. Also, this was added in commit c3c30d4b06
despite not being allowed in permittedKwargsc3c30d4b0.
For inexplicable reasons, we had a known_kwargs for custom_target that
looped over kwargs and issued a warning, not an error, for unknown
kwargs. It was impossible to ever hit that check to begin with, though,
ever since commit e08d735105 which added
permittedKwargs and obsoleted those manual checks with real errors.
So at one point override_options was specially permitted to be used
without emitting a warning, and then for about half a decade it was an
error, and then based on some dead code it was allowed again for a bit.
But through all this it doesn't do anything and isn't documented.
In commit fb2cdd0fe2 the internal property
was renamed, but one use case of it in raising a MesonException was not
changed to go with it.
This meant that instead of erroring out with:
```
ERROR: GNU symbol visibility arg XXXX not one of: default, internal, hidden, protected, inlineshidden
```
we instead errored out with:
```
AttributeError: 'SharedLibrary' object has no attribute 'symbol_visibility'
```
Fixes#9659
The new get_env() method that returns an EnvironmentVariables object
will be needed in next commit that will pass it to CustomTarget.
This has the side effect to use the proper os specific path separator
instead of hardcoding `:`. It is the obvious right thing to do here, but
has caused issues in the past. Hopefully issues have been fixed in the
meantime. If not, better deal with fallouts than keep doing the wrong
thing forever.
Forcing serialization on when writing out the build rule makes very
little sense. It was always "forced" on because we mandated a couple of
environment variables due to legacy reasons.
Add an attribute to RunTarget to say that a given target doesn't *need*
those environment variables, and let ninja optimize them away and run
the command directly if set.
That method had nothing specific to the backend, it's purely a Target
method. This allows to cache the OptionOverrideProxy object on the
Target instance instead of creating a new one for each option lookup.
Xavier Claessens