Killing a stream with pavucontrol doesn't kill the client it belongs to

[jbglaw]

Just insalled a new version of pavucontrol that allows me to kill a client connection. Due to a misconfigured flash plugin, this looked quite helpful since the ~30 client connection limit is rapidly eaten up by it within some seconds... Since I almost always have pavucontrol opened, I tried to kill all those connections. They're removed from the stream list, but pulseaudio's file descriptors to the local socket aren't closed and (as before) further client connections are disallowed.

This is with SVN as of two days ago IIRC.

[lennart]

Killing a stream will only do just that -- kill a stream. The connection is kept around. That's intended behaviour.

It might make sense to add another option for killing a client. OTOH having both is a bit confusing. Hmm, I have to think about it.

[monreal]

Isn't "killing" a bad word to use in a GUI? At least the GNOME HIG says not to use violent words like that.

[mkbosmans]

OK, so I think this behaviour is by design. The stream is killed, not the pulse client.

If you want flash killed, just use top/killall/gnome-system-monitor or something similar. Killing a stream in pavucontrol certainly not a workaround for buggy clients (flash)

Can we close this?

[coling]

I don't really see a massive need for this. We even went to the extent of "hiding" the "Kill Stream" option behind a menu (it used to be a button) in pavucontrol to generally discourage this. I can't really think of any occasions when I've genuinely had to use the option, so I think we should just close this bug as wontfix. If people provide clean patches it will of course be considered and merged if no major issues with the concept.