Concider adding a wrapper script to call skype when apulse USE is set as media-sound/apulse no longer installs libraries in system location because of bug #529098. Wrapper should call skype using apulse.
I will add wrapper script and ebuild diff later.
Preferably use 'i686-pc-linux-gnu-apulse' since that will work correctly both on x86 and multilib amd64.
Created attachment 391022 [details, diff]
Created attachment 391024 [details]
@amynka: please, review and test these changes for x86 and amd64 (I have tested only for amd64, but should be ok for x86 as well) and commit them if you are ok with this.
Or comment here, so I can commit these changes.
(In reply to Jauhien Piatlicki from comment #4)
> @amynka: please, review and test these changes for x86 and amd64 (I have
> tested only for amd64, but should be ok for x86 as well) and commit them if
> you are ok with this.
> Or comment here, so I can commit these changes.
I can confirm the following to work:
I wasn't aware of the recent changes to apulse (or perhaps multilib?), and the default build of apulse on amd64 doesn't work for Skype since it's 32-bit.
It seems that your wrapper does the exact same thing as what I discovered, so consider me another vote in favor of the wrapper!
I tested it on amd64. However I don't have right now any machine for testing on x86.
So for amd64 I can confirm it seems to be working fine. I think you can try to commit it (or should I?). We will see if someone will test it on x86 (it should work).
They can still report bug I guess.
Please commit Jauhien's changes. With USE=apulse, sound is currently broken on both amd64 and x86.
Also ~arch _is_ called "testing" for a reason. :)
(In reply to Ulrich Müller from comment #7)
> Please commit Jauhien's changes. With USE=apulse, sound is currently broken
> on both amd64 and x86.
> Also ~arch _is_ called "testing" for a reason. :)
Yes Sir :)
Although I welcome the change, the implementation has a problem, see bug 531914
(I was not sure whether I should report this as a new bug).
As a second remark, it would be nice to pass arguments through the wrapper script
(see also bug 531914).