On Mon, 15 Oct 2012 18:15:50 Antoine Martin wrote: > Would you mind trying r1941 to confirm that it is this changeset that is > causing the regression? (there's nothing else keyboard related after > this one - but better make sure)
If you think it's necessary... (I already tried r1944)... > > And if so, can you try each half separately (ignoring the win32 changes): > * the changes to xpra/xkbhelper.py (to do with zero keycodes) > * the changes to xpra/server_source.py ( In r1941 regression is manifested so it is positive that r1941 is broken. When I tried r1941 with xpra/server_source.py from r1940 I couldn't reproduce the issue. > Please be aware that due to some unfortunate asymmetry in the code X11 > keyboard API, it is possible for the keyboard mappings to degrade every > time you connect. This makes it much harder to pinpoint issues as it may > take 2 or more connections before the problems manifest themselves. (as > I found out at my expense when fixing Mac OS/win32 "AltGr" support) Interesting but I never experienced this. However you're right, something like this is there, for example (as I reported earlier) some keys like keypad '*' (and some others) may not work after switching from non-English keyboard layout and back, but start working again after switching layout one more time (to non-English and back again). It is a bit weird to notice keyboard problems after odd layout switch but not after even. :) All the best, Dmitry. _______________________________________________ shifter-users mailing list [email protected] http://lists.devloop.org.uk/mailman/listinfo/shifter-users
