On 03. 05. 23 13:24, Oswald Buddenhagen wrote:
On Tue, May 02, 2023 at 01:50:08PM +0200, Jaroslav Kysela wrote:
# Add silence in drain (-1 = auto /default/, 0 = off, > 0 silenced frames) [drain_silence INT]
i find this wholly inadequate as a description. specifically, it's missing a motivation.
and how would one use this in a meaningful way, given that the actual silence size is dependent on the period size and preferably the pointer alignment?
what i could imagine _hypothetically_ making sense is making the 1/10th sec "overshoot" configurable, as it's hardware-dependent. but in practice, i don't see how that would be actually useful, as the cost of doing too much is negligible, and the default you chose seems more than safe enough.
The positive value is a bit bonus. I just picked an easy understandable way. But looking to this issue for the second time, I changed the meaning for the positive value to milliseconds. In this way, it's time/rate related.
The main purpose for this option is to turn the code off. The other silence size calculation schemes may be described by another configuration field in future on demand.
Thanks for the review of all patches - I picked some proposals and pushed changes to the alsa-lib repository:
https://github.com/alsa-project/alsa-lib/commits/master
Jaroslav