da-viper wrote:
> If we pass in an output file, _compiler_supports shouldn't muck with it, and
> trust that the caller is in charge.
I agree with this, also the way `_compiler_supports` currently works reuses the
same `OnDiskTempFile` for every call to it.
it only works because we only use the `__enter__` and `__exit__` functions.
Instead we default to `None` and create it if it is needed. and let python
handle the refcounting.
```py
class OnDiskTempFile:
...
def __exit__(self, exc_type, exc_val, exc_tb):
self.__del__() # if we don't need it past the with scope.
def __del__(self): # will clean up after the last reference
if self.path and os.path.exists(self.path):
os.remove(self.path)
```
```py
def _compiler_supports(compiler, flag, source="int main() {}",
output_file=None):
"""Test whether the compiler supports the given flag."""
if output_file is None:
output_file = OnDiskTempFile()
if platform.system() == "Darwin":
compiler = "xcrun " + compiler
try:
cmd = "echo '%s' | %s %s -x c -o %s -" % (
source,
compiler,
flag,
output_file.path if output_file else output_file.path,
)
subprocess.check_call(cmd, shell=True)
except subprocess.CalledProcessError:
return False
return True
```
https://github.com/llvm/llvm-project/pull/176463
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
