[PATCH v3] glibcextract.py: Add compile_c_snippet

[Adhemerval Zanella Netto]

On 11/08/22 03:23, Florian Weimer wrote:
> * Adhemerval Zanella:
> 
>> It might be used on tests to check if a snippet build with the provided
>> compiler and flags.
>> ---
>>  scripts/glibcextract.py | 20 ++++++++++++++++++++
>>  1 file changed, 20 insertions(+)
>>
>> diff --git a/scripts/glibcextract.py b/scripts/glibcextract.py
>> index 43ab58ffe2..78735fad43 100644
>> --- a/scripts/glibcextract.py
>> +++ b/scripts/glibcextract.py
>> @@ -17,6 +17,7 @@
>>  # License along with the GNU C Library; if not, see
>>  # <https://www.gnu.org/licenses/>.
>>  
>> +import collections
>>  import os.path
>>  import re
>>  import subprocess
>> @@ -173,3 +174,22 @@ def compare_macro_consts(source_1, source_2, cc, macro_re, exclude_re=None,
>>              if not allow_extra_2:
>>                  ret = 1
>>      return ret
>> +
>> +CompileResult = collections.namedtuple("CompileResult", "returncode output")
>> +
>> +def compile_c_snippet(snippet, cc, extra_cc_args=''):
>> +    """Compile and return whether the SNIPPET can be build with CC along
>> +       EXTRA_CC_ARGS compiler flags.  Return a CompileResult with RETURNCODE
>> +       being 0 for success, or the failure value and the compiler output.
>> +    """
>> +    with tempfile.TemporaryDirectory() as temp_dir:
>> +        c_file_name = os.path.join(temp_dir, 'test.c')
>> +        obj_file_name = os.path.join(temp_dir, 'test.o')
>> +        with open(c_file_name, 'w') as c_file:
>> +            c_file.write(snippet + '\n')
>> +        cmd = cc.split()
>> +        if len(extra_cc_args) > 0:
>> +            cmd += [extra_cc_args]
>> +        cmd += ['-c', '-o', obj_file_name, c_file_name]
>> +        r = subprocess.run(cmd, capture_output=True)
>> +        return CompileResult(r.returncode, r.stderr)
> 
> Sadly that doesn't work because extra_cc_args isn't split.

It even simplify the code in fact, I will send an update.

> 
> Maybe we should just keep using the shell (without the redirect).
> 
> Thanks,
> Florian
>