[vlc-devel] [PATCH v2 18/18] video_output: "stop" the vout on close even if there's no display

[Steve Lhomme]

On 2020-11-25 16:59, Rémi Denis-Courmont wrote:
> Le mercredi 25 novembre 2020, 09:07:53 EET Steve Lhomme a écrit :
>> On 2020-11-24 16:28, Rémi Denis-Courmont wrote:
>>> Le mardi 24 novembre 2020, 12:46:41 EET Steve Lhomme a écrit :
>>>> We should always disable the window so it's not hanging with no display
>>>> module.
>>>
>>> The code is that way because disabling an already disabled window is (or
>>> at
>>> least was) not defined. The window is not supposed to have been left
>>> enabled without a display.
>>>
>>> This does not look right.
>>
>> There might be cases where restarting the display failed. In that case
>> we're stuck with a window and no display module attached to it. This is
>> transparent to the vout user who will never know about that and will
>> just release the vout normally when it's done.
> 
> Maybe the assumption is already broken, or maybe it's broken by the patchset.
> In that later case, the fix must come before the break in patch order.

OK

> And either way, we can't disable already disabled windows. If you want to

There is already a check for that in vout_DisableWindow() which is the 
only call to vout_window_Disable() done in video_output.c. So it's safe 
to call that function even if the window is already disabled.

> change those semantics, you need to audit existing modules and update the
> Doxygen first.
> 
> -- 
> Rémi Denis-Courmont
> 
> 
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel
>