[FFmpeg-devel] fate: Add cuda_runtime.h to FATE source test ref file

Submitted by Andreas Rheinhardt on Aug. 4, 2019, 9:52 p.m.

Details

Message ID 20190804215243.57039-1-andreas.rheinhardt@gmail.com
State New
Headers show

Commit Message

Andreas Rheinhardt Aug. 4, 2019, 9:52 p.m.
It uses a nonstandard inclusion guard.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
---
 tests/ref/fate/source | 1 +
 1 file changed, 1 insertion(+)

Comments

Timo Rothenpieler Aug. 4, 2019, 11:52 p.m.
On 04.08.2019 23:52, Andreas Rheinhardt wrote:
> It uses a nonstandard inclusion guard.

Is there any problem with just fixing the inclusion guard instead?
I forgot that this exists, but the inclusion guard in it looked fairly 
standard to me.
Feel free to just fix it.
Andreas Rheinhardt Aug. 5, 2019, 12:34 a.m.
Timo Rothenpieler:
> On 04.08.2019 23:52, Andreas Rheinhardt wrote:
>> It uses a nonstandard inclusion guard.
> 
> Is there any problem with just fixing the inclusion guard instead?
> I forgot that this exists, but the inclusion guard in it looked fairly
> standard to me.
> Feel free to just fix it.
> 
The standard inclusion guard for this file should be
COMPAT_CUDA_CUDA_RUNTIME_H without AV_, as this file isn't in any of
the libav*/ paths (at least that is what the source test expects).
Given that this header is designed to fulfill FFmpeg's needs and given
that dynlink_loader.h also uses an AV_ prefix (and is therefore marked
as an exception in the source test ref file), I thought that it is
intended to use the AV_ prefix so that it needs to be marked as an
exception. But if this is not so, then said prefix can of course be
removed. I don't know which of the two ways should be choosen.

- Andreas
James Almer Aug. 5, 2019, 12:46 a.m.
On 8/4/2019 9:34 PM, Andreas Rheinhardt wrote:
> Timo Rothenpieler:
>> On 04.08.2019 23:52, Andreas Rheinhardt wrote:
>>> It uses a nonstandard inclusion guard.
>>
>> Is there any problem with just fixing the inclusion guard instead?
>> I forgot that this exists, but the inclusion guard in it looked fairly
>> standard to me.
>> Feel free to just fix it.
>>
> The standard inclusion guard for this file should be
> COMPAT_CUDA_CUDA_RUNTIME_H without AV_, as this file isn't in any of
> the libav*/ paths (at least that is what the source test expects).
> Given that this header is designed to fulfill FFmpeg's needs and given
> that dynlink_loader.h also uses an AV_ prefix (and is therefore marked
> as an exception in the source test ref file), I thought that it is
> intended to use the AV_ prefix so that it needs to be marked as an
> exception. But if this is not so, then said prefix can of course be
> removed. I don't know which of the two ways should be choosen.
> 
> - Andreas

Remove the AV_ prefix. It's unnecessary and was probably added by
mistake. And while at it, you can do the same with dynlink_loader.h

Patch hide | download patch | download mbox

diff --git a/tests/ref/fate/source b/tests/ref/fate/source
index 8e12582ce8..3695238e71 100644
--- a/tests/ref/fate/source
+++ b/tests/ref/fate/source
@@ -25,6 +25,7 @@  compat/avisynth/avs/types.h
 compat/avisynth/avxsynth_c.h
 compat/avisynth/windowsPorts/basicDataTypeConversions.h
 compat/avisynth/windowsPorts/windows2linux.h
+compat/cuda/cuda_runtime.h
 compat/cuda/dynlink_loader.h
 compat/djgpp/math.h
 compat/float/float.h