[MPlayer-dev-eng] [PATCH] ALSA: get rid of potentially endless loop

[Uoti Urpala]

On Tue, 2009-09-29 at 15:38 +0200, Reimar Döffinger wrote:
> Any why does the code have to rely on "ALSA will eventually return an
> error"? Why should MPlayer code assume that ALSA is not buggy when it is
> no more effort to code in a robust way (obviously I could be missing
> 1000s of reasons why my proposed patch is a bad idea, but arguments
> along the line "as long as ALSA is working perfectly the current code is
> not wrong" don't really convince me.)

I don't consider exiting the loop to be any more robust. If MPlayer
hangs in the loop because of an ALSA bug that is an obvious, clean and
easily debuggable problem. Trying to continue in cases that are clearly
assumed to be bugs will obfuscate the problem by creating potentially
much more vague and harder to diagnose symptoms.