Proposed pulseaudio latency calculation fixes (fixes warning for unsigned int value)

[Gary Buhrmaster <gary.buhrmaster@…>]

In audiooutputpulse.cpp, clang reports a warning (comparison of unsigned expression < 0 is always false) for the following code fragment:

if (latency < 0)

latency is defined as:

pa_usec_t latency = (pa_usec_t) -1;

which, since pa_usec_t is an unsigned variable, actually sets latency to be the max value, and as the warning indicates, as an unsigned variable the comparison will always be false.

I am not at all familiar with the PulseAudio API, but it seems that one should check for the value being returned from get_latency being negative via the negative return int, and act accordingly if one wants to change a negative value to zero.

I think that it is possible with the current code for a latency value which should be interpreted as negative to be interpreted as positive, resulting in some confusion elsewhere.

I have provided a proposed patch, but the SME for this code needs to review it carefully to insure the patch would be an improvement. Since I do not currently use PulseAudio, I am not sure when I could test these changes.

Thanks.

Gary

[JYA]

The stream isn't a capture stream. As such, latency_negative will never be set to 1.

That whole test serves no purpose, it's just legacy code that wasn't properly adapted to pulse API.. Will simplify

[JYA]

Fixed in a6136f9fcbf9112cf5953277f7d2df641fec7824