bugginess on x86-64 in focus-follows-mouse

Hi all, I just this morning installed xmonad on my computer at work, and I've run into some bugginess in focus-follows-mouse on this computer: it doesn't work! The only difference that seems notable is that this is a 64-bit machine, while my home computer is 32-bit. In any case, the bugginess is just tha focus doesn't follow mouse. After some tracing, I discovered that the problem was that the ev_mode is weird. The following debug code in Main (right before the trace at the bottom) reveals that handle e@(CrossingEvent {ev_window = w, ev_event_type = t}) | t == enterNotify && ev_detail e /= notifyInferior = trace ("mode must be abnormal "++show (ev_mode e) ++ " vs " ++ show (ev_mode e `mod` 16)) mode must be abnormal 17179869184 vs 0 mode must be abnormal 8589934593 vs 1 Now according to the docs, ev_mode should be a small integer. I suspect that we're seeing here an Int vs CInt issue. :( I've glanced at the code, but really have no idea how to read a .hsc file. In the meantime, we can work around the bug by using the above `mod` in our crossing check. I'll send a patch shortly (although you're welcome to replace it with a real fix, but since the x11 package is included in ghc's distribution, I think some sort of workaround is in order). -- David Roundy Department of Physics Oregon State University

On Tue, 15 May 2007 19:02:50 -0700
David Roundy
Hi all,
I just this morning installed xmonad on my computer at work, and I've run into some bugginess in focus-follows-mouse on this computer: it doesn't work! The only difference that seems notable is that this is a 64-bit machine, while my home computer is 32-bit. In any case, the bugginess is just tha focus doesn't follow mouse.
X11 1.2 is almost totally broken on 64-bit. I suggest you upgrade to X11 1.2.1 (just released with GHC 6.6.1) or darcs X11. The problem should disappear with a new version of X11. Cheers, Spencer Janssen

On Tue, May 15, 2007 at 10:41:25PM -0500, Spencer Janssen wrote:
On Tue, 15 May 2007 19:02:50 -0700 David Roundy
wrote: Hi all,
I just this morning installed xmonad on my computer at work, and I've run into some bugginess in focus-follows-mouse on this computer: it doesn't work! The only difference that seems notable is that this is a 64-bit machine, while my home computer is 32-bit. In any case, the bugginess is just tha focus doesn't follow mouse.
X11 1.2 is almost totally broken on 64-bit. I suggest you upgrade to X11 1.2.1 (just released with GHC 6.6.1) or darcs X11. The problem should disappear with a new version of X11.
Argh. It's rather annoying that a broken version of X11 was included in the just-released debian stable... :( I'll see about compiling a newer X11. Thanks for the info! -- David Roundy http://www.darcs.net

droundy:
On Tue, May 15, 2007 at 10:41:25PM -0500, Spencer Janssen wrote:
On Tue, 15 May 2007 19:02:50 -0700 David Roundy
wrote: Hi all,
I just this morning installed xmonad on my computer at work, and I've run into some bugginess in focus-follows-mouse on this computer: it doesn't work! The only difference that seems notable is that this is a 64-bit machine, while my home computer is 32-bit. In any case, the bugginess is just tha focus doesn't follow mouse.
X11 1.2 is almost totally broken on 64-bit. I suggest you upgrade to X11 1.2.1 (just released with GHC 6.6.1) or darcs X11. The problem should disappear with a new version of X11.
Argh. It's rather annoying that a broken version of X11 was included in the just-released debian stable... :( I'll see about compiling a newer X11.
Let us know if that fixes the bug (it should, I think). -- Don

On Wed, May 16, 2007 at 11:26:40PM +1000, Donald Bruce Stewart wrote:
Argh. It's rather annoying that a broken version of X11 was included in the just-released debian stable... :( I'll see about compiling a newer X11.
Let us know if that fixes the bug (it should, I think).
X11$ runhaskell Setup.hs configure --user --prefix=/home/droundy Setup.hs:17:30: Couldn't match expected type `UserHooks' against inferred type `Maybe UserHooks' Expected type: Hook a Inferred type: PackageDescription -> LocalBuildInfo -> Maybe UserHooks -> BuildFlags -> IO () In the second argument of `($)', namely `buildHook defaultUserHooks' In the `buildHook' field of a record I guess presumably darcs version of X11 itself depends on a cabal more recent than that included in ghc 6.6? -- David Roundy Department of Physics Oregon State University

Hi David,
Setup.hs:17:30: Couldn't match expected type `UserHooks' against inferred type `Maybe UserHooks' Expected type: Hook a Inferred type: PackageDescription -> LocalBuildInfo -> Maybe UserHooks -> BuildFlags -> IO () In the second argument of `($)', namely `buildHook defaultUserHooks' In the `buildHook' field of a record
I guess presumably darcs version of X11 itself depends on a cabal more recent than that included in ghc 6.6?
Yes, but this error is usually caused by a type signature - not an actual problem. If there is a type signature in that Setup.hs, removing it may make it work. Otherwise upgrade to the latest Cabal. Thanks Neil
participants (4)
-
David Roundy
-
dons@cse.unsw.edu.au
-
Neil Mitchell
-
Spencer Janssen