On Wed, 19.12.07 12:12, Neil Thompson (abraxis(a)telkomsa.net) wrote:
> Also, I'd argue even if it worked before (which it
> behaviour exposed by PA+CK is the more correct one, from a security
> perspective. And hence speaking of a "regression" in this area is very
Something that used to work without issues, doesn't any more...looks like
a duck, walks like a duck...
Sorry, but it *didn't* work before.
By default the access mode of the dmix SHM is 0600. i.e. only a single
user ID may access the sound card at a time.
If you want to open up dmix to multiple users at the same time you
need to change it to 0666 or so, which is a security hole and needs
some non-trivial reconfiguration.
The reconfiguration necessary to open up PA to other users is a lot
simpler to do. Just copy a cookie file around.
And then the use case just gets dismissed. Paraphrasing..."the
who doesn't really use her machine will be OK, and the other folks will just have
to change the way they work".
I didn't dismiss your use case at all. All I told you is that the kind
of setup you envision required non-trivial reconfiguration before, and
it requires reconfiguration now. So, calling this is a regression is
Lennart Poettering Red Hat, Inc.
lennart [at] poettering [dot] net ICQ# 11060553