[vlc-devel] [PATCH] Fix https://trac.videolan.org/vlc/ticket/10749 Video was getting stuck while using subtitle delay option when setting a negative subtitle delay.

[Pascal Thomet]

Hello,

Thanks for your answer.

You are right, this has to be handled differently depending on the source
of the subtitles.

This patch actually tries to introduce two different behaviors

- if the subtitles come from the subtitle demux, then the delay is removed
- if the subtitles come from another source, then the old behavior is
conserved (and a delay is still applied)


Details :

- a new variable ("sub-isfilesub") is added inside input/var.c. It is set
to *false* by default

- if this variable is set to false, the old behavior remains strictly
identical : the two functions that handle the delay
(namely input/input.c:UpdatePtsDelay and  input/var.c:EsDelayCallback) keep
the *exact* same behavior as before

- if this variable is set to true, than effectively, the delay will be
removed

- So, I agree, we would need to make sure that, inside we have a reliable
way to tell "file subtitles" apart from other subtitle types in order to
correctly set this variable ?
Is it impossible or too difficult ?

Here is what I can tell about the lifetime of this variable :
* it is set to false upon construction (inside input/var.c)
* it is also re-set to false at the very beginning of input.c:LoadSlaves
(which is called each time a new movie is loaded, I think)
* the only place where it is set to true is inside
input.c:input_SlaveSourceAdd (and only if "slave_type" is equal to
"SLAVE_TYPE_SPU". See input.c:3333)


Below is the callstack when entering demux/subtitle.c:Open():

1  Open                 subtitle.c 309
2  demux_Probe          demux.c    191
3  module_load          modules.c  183
4  vlc_module_load      modules.c  279
5  demux_NewAdvanced    demux.c    264
6  InputDemuxNew        input.c    2553
7  InputSourceNew       input.c    2663
8  input_SlaveSourceAdd input.c    3325    <-- sub-isfilesub is set to true
here if slave_type==SLAVE_TYPE_SPU
9  LoadSlaves           input.c    1161    <-- sub-filesub is set to false
here
10 Init                 input.c    1363
11 Run                  input.c    497


I assumed that testing for  slave_type==SLAVE_TYPE_SPU (inside
input.c:input_SlaveSourceAdd) would be a good test .
However, it could be done elsewhere in this callstack.

My initial intent was to set "sub-isfilesub" directly inside the Open()
function. However, it seems it is not possible to set an inherited var from
there. Did I miss something ?

Do you know of another place in the callstack where it would be more suited
to change it ?

Sorry for the long email, and thus many thanks for your attention,

Regards,

Pascal Thomet

On Tue, Oct 3, 2017 at 5:29 PM, Rémi Denis-Courmont <remi at remlab.net> wrote:

>         Hello,
>
> This seems to assume that all subtitles come from the subtitle demux,
> which is
> definitely not true.
>
> --
> 雷米‧德尼-库尔蒙
> https://www.remlab.net/
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20171003/0797d304/attachment.html>