Merge lp://qastaging/~vanvugt/compiz-core/fix-925979 into lp://qastaging/compiz-core/0.9.5

OK

I'm pretty sure this was added for a good reason, let me check with racarr on that one.

------------------------------------------------------------
revno: 2814.1.1
git commit: 6b8ecb998ce409462cfd72343696f4925aa33fdb
committer: Robert Carr <email address hidden>
timestamp: Thu 2011-09-15 11:28:55 -0400
message:
  event.cpp: Only ungrab the keyboard if the keystate is non zero.

Well it caused a regression. So whatever it was an attempt to fix should and probably can be fixed differently. In fact, I would be surprised if there wasn't a solution where we could remove XUngrabKeyboard and replace it with something else. But I don't know what else will work.

Whatever the alternative is, we should at least use the test case described in bug 925979.

Robert, could you please explain what you were fixing with this commit:
http://git.compiz.org/compiz/core/commit/?id=6b8ecb998ce409462cfd72343696f4925aa33fdb

Or rather: http://bazaar.launchpad.net/~compiz-team/compiz-core/0.9.5/revision/2815

On Sat, 4 Feb 2012, Daniel van Vugt wrote:

> Well it caused a regression. So whatever it was an attempt to fix should and probably can be fixed differently. In fact, I would be surprised if there wasn't a solution where we could remove XUngrabKeyboard and replace it with something else. But I don't know what else will work.
>
> Whatever the alternative is, we should at least use the test case described in bug 925979.

To be honest, I think this is the right approach, although I remember
racarr told me there was a good reason to only allow this on keys that
have a modifier (eg, not single keys). Perhaps that reason isn't relevant
anymore, I'll need to check.

I'll ask him when he gets in next week.

> --
> https://code.launchpad.net/~vanvugt/compiz-core/fix-925979/+merge/91412
> Your team Compiz Maintainers is subscribed to branch lp:compiz-core.
>

We could accommodate both. The issue with tapping a modifier is that it creates regular key events without any modifier state set. So maybe it needs:

 if (event->xkey.state || (keycode is a modifier))

But I'm reluctant to keep any "if" statement if Robert can't remember the justification for introducing it.

Needs resubmit for target branch lp:compiz-core (0.9.7)