With upgrade to 0.9.3 (from 0.9.2) I've noticed that mouse cursor on text selection "][" becomes quite big -- at least 3 times bigger than normal.
So far I've seen this effect on KDE in kmail and reportbug-ng applications.
This is also affects "hand" mouse cursor when hovering over links.
Hi - is this a regression from 0.9.2 to 0.9.3 as far as you know? Sounds like the cursor scaling code was useful after all. :(
Yes this is certainly a regression...
Also seeing this with a cursor in an xterm with 0.10 trunk r3467 windows client and 0.9.3 r3467 fedora server (not sure what caused it... was trying to paste from local windows app and my fingers missed, can't seem to reproduce yet).
This was caused by r3423 (r3429 for 0.9.x branch).
Please see ticket/189: until we can get Xdummy to export the correct screen size (in mm) and get the toolkits to use the correct cursor size, the only solution that can satisfy all applications is likely to be the option in that patch.
For 0.9.x, I am more likely to just revert the patch.
To help, please post the output of:
python -c "from gtk import gdk;print(gdk.display_get_default().get_default_cursor_size())"
From:
DISPLAY=:NNN thecommand)
Replying to antoine:
To help, please post the output of:
python -c "from gtk import gdk;print(gdk.display_get_default().get_default_cursor_size())"From:
- a regular terminal
21
- from inside xpra
21 (when attached, with DISPLAY=:NNN)
- against the xpra display before ever connecting (using:
DISPLAY=:NNN thecommand)
66
Without understanding much about why this is happening and how I'd vote for reversal in 0.9.x. There are many cursors and situations affected and having big mouse cursors jumping at you is more uncomfortable that I thought...
I will revert for 0.9.x
But we need to find a better solution generally - it seems to me that the cursor size changes depending on the reported screen size, so maybe the scaling also needs to be dynamic, reporting the current server default cursor size to the client and letting it decide how to handle it (and as per the patch in #189, if we can't satisfy all cases, adding a client option)
Replying to antoine:
- a regular terminal
18
- from inside xpra
21
- against the xpra display before ever connecting (using:
DISPLAY=:NNN thecommand)
66
Replying to antoine:
To help, please post the output of:
python -c "from gtk import gdk;print(gdk.display_get_default().get_default_cursor_size())"
This is the output from my notebook that I used to attach to Xpra session. Some cursors are shrinking. :-)
From:
- a regular terminal
16
- from inside xpra
21
The "fix" has been reverted in r3517
Will follow up in #189
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/343