On 11/03/2011 09:41 PM, Tanu Kaskinen wrote: > Some comments below. I don't have time right now to go through it all, > but I'll continue later. You might have to wait until Monday for the > rest, though, if Colin or Arun doesn't do the review before me. Thanks for the review. There are some cases where I think it's mostly your taste against mine and I would like to know (from Colin or Arun) on how to proceed. I don't want to lose momentum here (altough it took a few days for me to answer; because I was busy with the Ubuntu Developer's conference last week). > So pa_alsa_path_probe() should always set p->probed to TRUE. Instead > of setting it three times, I think it would be better to set it > immediately after the "if (p->probed)" check in the beginning. Good point. Fixing. > I have not thought this through, but would it make sense to use the > path name as the key to ps->paths, and if there are collisions, do > the path renaming here instead of using the > path_set_make_paths_unique() function later? In general, it would make the most sense to do the paths_unique after all possible removals, to avoid unnecessary indices: e g, if you have paths "Foo", "Bar" and one more "Foo", and the first "Foo" is a subset of "Bar", then we would not like the second "Foo" to be renamed to "Foo 2". > What if some path is removed from all path sets? Does it still exist > in the profile set, and if so, will a client-visible port be created > for it? It will remain in the profile set, but no port will be created - when adding ports (see module-alsa-card.c:add_profiles) we'll walk through all profiles to see what ports to add. -- David Henningsson, Canonical Ltd. http://launchpad.net/~diwic
