James Courtier-Dutton wrote:
>
> I generally prefer the value returned from a function to be the error code
> and
> function parameters passed to a function call, to also act as the return
> method for any extra information which needs returning.
>
> It makes the source look neater.
>
> So I prefer: -
> int snd_pcm_mmap_commit(snd_pcm_t *pcm,
> snd_pcm_uframes_t offset,
> snd_pcm_uframes_t frames,
> snd_pcm_uframes_t *committed);
>
I'd like to make clear that
snd_pcm_sframes_t snd_pcm_mmap_commit(snd_pcm_t *pcm,
snd_pcm_uframes_t offset,
snd_pcm_uframes_t frames);
is perfectly analogue to snd_pcm_write/read and have the same
benefits/drawback.
I say this to precise that changing one will imply changing all.
Jaroslav wrote:
> I see the benefit of the extra parameter, that we can return the error
> code and transfer count in one pass. Otherwise the error code can be lost
> or returned later using another function like snd_pcm_avail_update() etc..
I see the benefits of that, but please don't consider asymmetric
changes.
However I'm wondering why Unix read/write does not have considered that
(yes, I'm serious).
Perhaps only because error will be returned on next call and taken for
granted that a partial transfer definitely imply a next call, they have
considered the early return of error a little benefit.
--
Abramo Bagnara mailto:[EMAIL PROTECTED]
Opera Unica Phone: +39.546.656023
Via Emilia Interna, 140
48014 Castel Bolognese (RA) - Italy
ALSA project http://www.alsa-project.org
It sounds good!
_______________________________________________
Alsa-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/alsa-devel
