On Sat, 14 Jun 2003, Tim Connors wrote: I didn't CC either of the email below to the bug, so included in full:
> On Sat, 14 Jun 2003, Martin Pitt wrote: > > > Am 2003-06-14 12:57 +1000 schrieb Tim Connors: > > > When in fullscreen mode, qiv incrrectly steals both keystrokes and > > > mouseclicks from the underlying window manager. This means one cannot > > > change virtual pages, nor lower the qiv window in the window stack, or > > > run other programs bound to keys (the former being the most painful > > > for me) > > > > > > > I habe never seen a fullscreen program allowing this up to now > > (mplayer, games, etc.). Do you? Fullscreen essentially means to turn > > off the window manager to unclutter the desktop from window frames, > > pagers staying on top and so on; but this is just my impression, not a > > fact from a book. I always considered this as a feature of fullscreen > > to allow a program to get full control. > > OK - I would have thought keyclicks were a different thing to displaying > your application over the top of everything else though. > > > After all, you can just press f to get back to window-managed mode, or > > do not use fullscreen mode and maximise your window instead. > > Yeah - my reason for using fullscreen though is because different aspect > ratio images keep wanting to resize the window frame. This is a *really* > slow operation on my computer, and several redraws seem to happen. > > > > If this is not a sideeffect of some kind of X-accalaration, would it > > > be possible to fix? > > > > I am not an expert of the internals of the gtk library but I will > > investigate. I will notify you if I hear something. > > Thankyou. > > > Would you object to downgrading this bug to "wishlist" or at most > > "minor"? > > OK - do you need to do that, or can I? (I can't seem to find an option on > the BTS) I believe I now have a reason to mark this as grave (dataloss). I just wish I knew how to after the fact... > On Sat, 14 Jun 2003, Martin Pitt wrote: > >> Hi Tim! >> >> Am 2003-06-14 12:57 +1000 schrieb Tim Connors: >> > When in fullscreen mode, qiv incrrectly steals both keystrokes and >> > mouseclicks from the underlying window manager. This means one cannot >> > change virtual pages, nor lower the qiv window in the window stack, or >> > run other programs bound to keys (the former being the most painful >> > for me) >> > >> >> I habe never seen a fullscreen program allowing this up to now >> (mplayer, games, etc.). Do you? Fullscreen essentially means to turn > > I forgot to mention that I actually do get my expected behaviour on the > small sample of apps that I have tried. > > These are mplayer in full screen, xawtv, acroread, etc. > > Thinking twice about it, perhaps not mplayer, but I cannot test this right > now (my grey-matter-memory seems to be suffering ECC errors, and my home > box is... well... at home). mplayer does indeed allow me to switch to another virtual page. As does xine, and absolutely every other fullscreen app I have ever used. And these apps have never caused problems with the window manager like upstream reported. > As for the others, they probably are using very different widget sets to > gtk2. The reason I want to mark this bug as grave, is it has essentially rendered my X session crashed. A few weeks ago, I had an qiv session that I quit, and fortunately, it wasn't fullscreen, when the process dissapeared, but the window remained behind. I just got the window manager to destroy the window (closing it merely wasn't enough). It has happened again just now, except that I was in fullscreen at the time. I can't do anything. The window manager can't intercept any keystrokes whatsoever (I haven't yet tried ctrl-alt-backspace; will leave that after I exhaust all other avenues). There are no qiv processes left behind. If I try to start xine, I get a message displayed over the old qiv window along involving "__net_wm_state" (that I can't reproduce), and if I try to start a fullscreen mplayer, nothing is displayed; but I can start up another fullscreen qiv application. Qiv really is doing something different to all other fullscreen apps, and the fact that the entire X session is unrecoverable when a qiv crashes is not acceptable really. (I am OK with qiv crashing -- I won't report a bug for that partly because it will be so hard to debug with it being such a rare and random occurence). Does upstream have a clue of how to recover this particular X session, or is it a lost hope? -- TimC "The application did not fail successfully because of an error" -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
