[FFmpeg-devel] [PATCH] configure: check if NAN can be used as a constant initializer

[James Almer]

On 9/13/2017 3:08 PM, Reimar Döffinger wrote:
> On Wed, Sep 13, 2017 at 02:49:59PM -0300, James Almer wrote:
>> On 9/13/2017 2:23 PM, Carl Eugen Hoyos wrote:
>>> 2017-09-13 19:10 GMT+02:00 James Almer <jamrial at gmail.com>:
>>>> Some targets, like NetBSD and DJGPP, don't seem to support it.
>>>>
>>>> Signed-off-by: James Almer <jamrial at gmail.com>
>>>> ---
>>>>  configure | 6 ++++++
>>>>  1 file changed, 6 insertions(+)
>>>>
>>>> diff --git a/configure b/configure
>>>> index ab1550d39b..4855ca54c9 100755
>>>> --- a/configure
>>>> +++ b/configure
>>>> @@ -3222,6 +3222,7 @@ pixfmts_super2xsai_test_deps="super2xsai_filter"
>>>>  tinterlace_filter_deps="gpl"
>>>>  tinterlace_merge_test_deps="tinterlace_filter"
>>>>  tinterlace_pad_test_deps="tinterlace_filter"
>>>> +tonemap_filter_deps="const_nan"
>>>
>>> No objections but what was wrong with using another
>>> default to avoid the additional check?
>>
>> DBL_MIN and DBL_MAX are both valid values for the option. Turning one of
>> them into "auto" does not seem like a good idea to me.
>>
>> If someone else has a better solution than a simple check for broken
>> toolchains at configure time then I'll drop this patch.
> 
> Not having a better solution, but at least vf_zscale looks to me like it should
> have the same issue.
Ah, thanks for spotting that. No affected FATE client seems to link
libzimg so they wouldn't have failed in vf_zscale after this or any
other solution for vf_tonemap was committed.

> It would look at but less "silly" if it was a check used for more
> than just one filter.

Will add it and push. If someone has a better solution then this one can
be replaced.