I found something. There's a function calling Display.setFullscreen() from LWJGL, which usually gets called after a check for F11:
if (Keyboard.getEventKey() == KEY_F11) { function(); }
On the occasions where this bug happens and I can't get fullscreen for more than a split second, the same function also gets called from somewhere else, doing a Display.setFullscreen(false) on top of the earlier call. That somewhere else is around a Display.isActive() check.
So, reading various topics, it looks like this might actually be a WM/driver/lwjgl issue even though it's age-old.
What bugs me is that it doesn't happen at the title menu so there is something in the game code that's triggering it.
Thanks, I will check MC-2067 out.
It did not come up come up in my searches before posting this bug (rather got the fullscreen crash ones).
I will investigate further.
While fullscreen usually crashes on my Macbook, I can also (unreliably) replicate the cursor/turning thing.
Linux cafebabe 3.8.0-31-generic #46-Ubuntu SMP Tue Sep 10 20:03:44 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
server glx vendor string: SGI
server glx version string: 1.4
Qt: 4.8.4
KDE: 4.10.5
Thanks, I'll update.
Though I hightly doubt it'll fix this age-old bug.
Crash report as per request
I'm a developer and am willing to further investigate this issue if needed.
It seems like the problem in the title is gone.
However, the "cursor confinement" issue is now 100% replicatable. (But can be worked around by freeing the cursor by opening a menu (ESC) before going fullscreen – also at 100%)
Feel free to resolve the issue.
I will open another one for the cursor or comment on an existing report as soon as I have the time.