Carl Eugen Hoyos (2017-11-27):
> The patch that introduced the regression was not discussed and
> it used a guessed limit afair.

Two wrong don't make a right. You should not have pushed without

> The patch that fixed ticket #373 simply doubled the value to fix
> a particular use-case.

Well, it was wrong too, I did not notice at the time.

> After this patch that follows this logic, the possible allocation
> is still a magnitude smaller than the original commit claims.
> (I cannot test the failing case.)
> Do you want me to revert?

I want you to refrain from pushing this kind of patch.

> Any better suggestions?

When the problem is that the value returned by
snd_pcm_hw_params_get_buffer_size_max() is too large and the cap set by
ALSA_BUFFER_SIZE_MAX is too small, the obvious fix is to involve


  Nicolas George

Attachment: signature.asc
Description: Digital signature

ffmpeg-devel mailing list

Reply via email to