Re: [PATCH] fix build failure with Python 3.7

[Phil Muldoon <pmuldoon@redhat.com>]

On 31/05/18 22:20, Sergio Durigan Junior wrote:
> On Thursday, May 31 2018, Paul Koning wrote:
> 
>>> On May 31, 2018, at 4:10 PM, Sergio Durigan Junior <sergiodj@redhat.com> wrote:
>>>
>>>>
>>>> gdb/ChangeLog:
>>>>
>>>> 2018-05-31  Paul Koning  <paul_koning@dell.com>
>>>>
>>>> 	PR gdb/33470
>>>
>>> This bug number is actually from Python's bugzilla, not GDB's.  So it's
>>> not correct to mention it here in the ChangeLog/commit message.  AFAIK,
>>> there's no correspondent GDB bug filed for this issue.
>>
>> Ok, I removed that.
>>
>>>
>>>> 	* python/python.c (do_start_initialization):
>>>> 	Avoid call to internal Python API.
>>>> 	(PyInit__gdb): New function.
>>>>
>>>> diff --git a/gdb/python/python.c b/gdb/python/python.c
>>>> index c29e7d7a6b..89443aee25 100644
>>>> --- a/gdb/python/python.c
>>>> +++ b/gdb/python/python.c
>>>> @@ -1667,6 +1667,14 @@ finalize_python (void *ignore)
>>>>   restore_active_ext_lang (previous_active);
>>>> }
>>>>
>>>> +#ifdef IS_PY3K
>>>> +PyMODINIT_FUNC
>>>> +PyInit__gdb (void)
>>>> +{
>>>> +  return PyModule_Create (&python_GdbModuleDef);
>>>> +}
>>>> +#endif
>>>
>>> I think it's a good idea to add a comment to this function.
>>
>> I added this (after the #ifdef):
>>
>> /* This is called via the PyImport_AppendInittab mechanism called
>>    during initialization, to make the built-in _gdb module known to
>>    Python.  */
> 
> Thanks.
> 
> FWIW, I ran the full testsuite here and no regressions were found.
> 

Hey Sergio,

Thanks for doing this. Did you test it against a GDB built against 3.7
and a GDB build with a 2.x version of Python?

Cheers

Phil