dfaure added a comment.
Proper fix submitted in https://phabricator.kde.org/D7829 -- now with
unittests.
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: broulik, martinkostolny, bcooksley, graesslin, #frameworks
graesslin added a comment.
In https://phabricator.kde.org/D6233#145981, @dfaure wrote:
> Qt (at shortcut recording time) sees this as Meta+Shift+Print while
KWindowSystem (xcbKeyPressEventToQt) sees it as Meta+Shift+SysReq.
>
> Qt does xkb_state_key_get_one_sym( state=51 , code=6b )
dfaure added a comment.
Qt (at shortcut recording time) sees this as Meta+Shift+Print while
KWindowSystem (xcbKeyPressEventToQt) sees it as Meta+Shift+SysReq.
Qt does xkb_state_key_get_one_sym( state=51 , code=6b ) = ff61 (XK_Print)
xcbKeyPressEventToQt does
dfaure added a comment.
Alt+Shift+right bug fixed in https://phabricator.kde.org/D7829
Meta+Shift+Print is another issue, I get
KGlobalAccelImpl::grabKey: keyQt (0x 100011e ) failed to resolve to x11
keycode
and
KGlobalAccelImpl::x11KeyPress: keyQt= "130a"
dfaure added a comment.
Thanks for the reports. These shortcuts get converted correctly
(https://commits.kde.org/kwindowsystem/af6d5ab12b2d99b39a73ed8df903368470ded55b),
so the problem must be in the shift handling in xcbKeyPressEventToQt().
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
broulik added a comment.
When I press Meta+Shift+PrintScreen previously I got
`QKeySequence(Meta+Shift+Print)` which is correct but now I get
`QKeySequence(Meta+Shift+SysReq)` which fails to map (on German layout SysReq
is the key and Shift+SysReq results in PrintScreen)
REPOSITORY
R278
graesslin added a comment.
This seems to be a big problem. Kglobalaccell got more new bugs reported this
week, than normally in a year.
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: broulik, martinkostolny, bcooksley,
broulik added a comment.
Confirmed. I just today experienced the "Rectangular screenshot" keyboard
shortcut (Meta+Shift+PrintScreen) having stopped working on two machines.
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc:
martinkostolny added a comment.
Sorry, to be a bit late, but after updating to kwindowsystem 5.38, global
shortcuts "meta+shift+" and "alt+shift+" stopped
working on my system (Arch Linux). They can be configured no-problem (they are
recognized and saved in configuration), but they do
dfaure added a comment.
Thanks for the investigation. Actually the deprecated method wasn't supposed
to change behaviour. This was actually a consequence of the rework of the
patch, which initially didn't have this problem :-)
Proper fix is up for review:
graesslin added a comment.
To me the question is rather: can we ensure that other users of KKeyServer
didn't get broken by the change. What KWin did could be considered a bug or an
application misuse. KWin used the Qt key and compared it directly to
Qt::Key_Enter. That one doesn't work
graesslin added a comment.
@dfaure please note that I already worked around the regression in KWin
master.
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: bcooksley, graesslin, #frameworks
bcooksley added a comment.
@dfaure according to our Docker image the SUSE package you need is
xorg-x11-server-wayland
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: bcooksley, graesslin, #frameworks
dfaure added a comment.
I'm trying to run that unittest, but it aborts without Xwayland, and even
after upgrading to OpenSuSE Leap 42.3, I still don't find a package containing
Xwayland. It seems to be part of recent Xorg?
I have xorg-x11-server-7.6_1.18.3-22.1.x86_64
REPOSITORY
R278
graesslin added a comment.
Sorry for having missed your pings. I have something like 1000 unread mails
in the frameworks folder :-(
Your change caused a regression in KWin and now a unit test is failing.
Please see:
This revision was automatically updated to reflect the committed changes.
Closed by commit R278:32526718eae9: KKeyServer: fix handling of KeypadModifier.
(authored by dfaure).
CHANGED PRIOR TO COMMIT
https://phabricator.kde.org/D6233?vs=15522=17994#toc
REPOSITORY
R278 KWindowSystem
CHANGES
dfaure marked 4 inline comments as done.
dfaure added a comment.
What if you write me down as maintainer of this code, can this then go in?
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: graesslin, #frameworks
dfaure added a comment.
ping?
REPOSITORY
R278 KWindowSystem
REVISION DETAIL
https://phabricator.kde.org/D6233
To: dfaure, graesslin
Cc: graesslin, #frameworks
dfaure added inline comments.
INLINE COMMENTS
> graesslin wrote in kkeyserver.cpp:976-980
> This is a special handling from KGlobalAccel. Are you sure it belongs into
> the generic implementation?
I would say yes, because e.g. Ctrl+& (on qwerty) is going to include Shift no
matter which
graesslin added inline comments.
INLINE COMMENTS
> kkeyserver.cpp:976-980
> +if (*keyQt != Qt::Key_Tab) { // KKeySequenceWidget does not map
> shift+tab to backtab
> +static const int FirstLevelShift = 1;
> +keySymX = xcb_key_symbols_get_keysym(symbols,
dfaure updated this revision to Diff 15522.
dfaure added a comment.
Adjustments suggested by Martin Graesslin
REPOSITORY
R278 KWindowSystem
CHANGES SINCE LAST UPDATE
https://phabricator.kde.org/D6233?vs=15487=15522
BRANCH
master
REVISION DETAIL
https://phabricator.kde.org/D6233
dfaure added inline comments.
INLINE COMMENTS
> graesslin wrote in kkeyserver.cpp:160-180
> This looks very unrelated to the described change. Maybe an own commit?
Well those are the XK_KP_* codes, i.e. Num Keypad keys, so it's related. But
yeah, it would probably work without this change, it
graesslin requested changes to this revision.
graesslin added inline comments.
This revision now requires changes to proceed.
INLINE COMMENTS
> kkeyserver.cpp:160-180
> +{ Qt::Key_Space, XK_KP_Space },
> +{ Qt::Key_Tab,XK_KP_Tab },
> +{ Qt::Key_Enter, XK_KP_Enter },
dfaure created this revision.
Restricted Application added a project: Frameworks.
Restricted Application added a subscriber: Frameworks.
REVISION SUMMARY
This required adding a new method symXModXToKeyQt since
symXToKeyQt (without modifier as input) has no way to find out
whether to use
24 matches
Mail list logo
