mherger wrote:
> > I've recompiled the windows socket wrapper with a small change that
> only
> > activates the WDT mechanism when reading from the source socket (and
> > ignores long waits on writing to the LMS thread).
>
> Interesting... I was wondering whether this could fix issues we've seen
>
> with podcasts which would end pre-maturely. Unfortunately I'm not
> familiar with this code at all. Would be great if bpa or somebody else
> could give a second opinion.
>
> --
>
> Michael
I was originally thinking of increasing the WDT timeout duration, but
this would just push the problem around; how long do you make the
timeout, knowing that there are some very long tracks out there (think
progressive rock, classical music, podcasts, etc). I also thought about
using the original WDT increment (each packet increments the wdt
counter) coupled with decrement, where the main thread substract from
the counter on every timeout (as opposed to current logic that sets the
wdt counter to 0 after the first timeout), but I can always think of a
scenario that would make the wdt trigger prematurely.
So I am 100% confident that this would solve most of the issues related
to "not playing tracks fully" that, when searching for solutions, seem
to have started appearing a long while ago. We'd need a second opinion
to understand potential issues I can think of like
- Similar to what happened to the sockerwrapper, source streams might
"die" if the source side doesn't read often enough
- Processes could possibly become orphans if the LMS reader ends
prematurely.
I'm happy to take the "risk" in my own environment, though, and spotty
playback is now solid as a rock (no pun intended).
LMS 7.9 - 1xRadio, 1xBoom, 5xDuet,3xTouch, 1 SB2. Sony PlayStation,
Emby, Chromecast v1 and v2 and...
*SQUEEZEAMP*!
------------------------------------------------------------------------
sle118's Profile: http://forums.slimdevices.com/member.php?userid=35009
View this thread: http://forums.slimdevices.com/showthread.php?t=110455
_______________________________________________
plugins mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/plugins
