GNOME Bugzilla – Bug 168516
Can't create guides with tablet on Windows
Last modified: 2008-01-15 12:48:51 UTC
Distribution/Version: XP using windows XP on a here-intern-have-a-laptop 500mhz ibm thinkpad have just downloaded 2.2.3 (latest windows binary) and the following persist when using my tablet (Wacom Graphire3 4x3) I don't know if the issues have been resolved in 2.2.4 but I have seen nothing even vaguely resembling them in the bugfix lists, so I will file this under 2.2.4 anyway. I am new to bugzilla, so if this is a naughty naughty thing, then I will refile the entry as 2.2.3 first, and most obvious, is an inability to use my tablet to create guides it selects the Move tool, as using the normal mouse does, when I click on the ruler, but does not let me drag guides out of the ruler second, more irritating but possibly more difficult to solve, is the way Gimp handles the multiple input devices that my tablet represents I do not use the tablet-mouse at all, for starters I also do not use the normal mouse at all, except for creating guides my major struggle is that when the tablet is inactive, Gimp automatically switches back to the mouse as its input, even though the mouse is not receiving any input this results in loss of curves, selected paths, and other annoyances any time I lift the stylus too far from the surface of the tablet while my cursor is over the image or tool settings my suggestion is that no input switching is done unless a device receives input, so that simply lifting the stylus briefly does not trigger a change this happens inconsistently, apparently only when lifting the stylus slowly enough for the change to be registered it would also be nice if I could disable the seperate settings for a particular input device, such that it simply uses whatever settings were in use by the previous device, so that similar problems would not be caused by slight bumps of a mouse or other device but that's bordering on feature request
when I have my new system up and running I will find out if the issues exist on a native linux build, assuming my tablet will work
I would like the same "feature". GIMP should keep the same input device unless it receives input from another. I don't even have a USB mouse installed on my system, and it always switches to the mouse cursor when I lift my tablet pen and loses what I was doing.
Brian, this is not some feature you are missing, it is simply a bug somewhere in the wintab support in gtk+/gdk. This bug report is not going to change anything about it. It should be reassigned to gtk+ and some Windows users with a tablet should help the gtk+ developers to sort out the issues with tablets on Windows.
Andy, can you try again with the latest gtk+ and gimp builds and report back if your problem still persists. We can then reassign the report to the gtk+ product.
Andy, any news?
(after finally getting logged in again) I'm just now using the tablet on *my* laptop, a fairly old latitude c600 with Gentoo Let me just load up Gimp (2.2.6 now) and see what we have... The guides work beautifully, much to my pleasure. I actually suspect now that part of the other problem may have been the hardware on the other laptop, if the touchpad were sending small signals to the system constantly (or the eraser stick, which I hate and suspect is the real culprit) it would have that effect.. so, there needs to be a way to disable core input devices from being unique inputs as well as disabling extended inputs, to prevent losing tool settings and the like when you lift the stylus too far off the tablet As for windows, I don't have a windows machine available for testing any more, except one which I wouldn't trust any results from tests run on it. That machine is used as a holding cell for downloads, mostly... just a side comment, I occasionally have my cursor lock up when I click immediately after a long paint operation (airbrush, ink, etc) irritating as it makes it difficult to work naturally.
What are we supposed to do with this report then? Can it be closed?
Comment #6 seems to include some enhancement requests, but they should be raised in new reports if desired. NOTGNOME seems to be a suitable reolution for this bug.
Reopening as I can reproduce the first issue mentioned in this bug report, see also http://gug.sunsite.dk/forum/?threadid=3322 As I also said on the forum thread: What basically happens is that when creating a guide by dragging from the ruler, the move tool receives a spurious button press event due to one or more limitations in the GTK+ win32 tablet code, and that makes GIMP stop creating the guide. I will create a bug report for GTK+ about this, but it looks nontrivial to fix completely in GTK+. There is however a rather simple workaround that can be implemented in GIMP: Enable extension events for the rulers. This will avoid the duplicate press events and creating rulers work nicely for me. I'll attach a patch for GIMP HEAD that does this. The second issue mentioned in this bug report, GIMP switching to the mouse device when lifting the pen, has been fixed in GTK+. See bug #167000.
Created attachment 51935 [details] [review] Workaround inability to create guides with tablet on Windows Here's the patch that implements the workaround. I've tested it on GIMP HEAD only, but the workaround will probably work for gimp-2-2 as well. Tested with a Wacom Graphire 3 tablet on Windows XP.
The patch should be tested on Linux (with mouse and tablet) and if it doesn't break anything, the patch should go into both branches.
I did a quick test on Linux with GIMP HEAD, X.Org 6.8.2 and the same hardware and saw no breakage with the patch. I can do some more testing later if you wish. Another option would be to wrap the workaround in an #ifdef G_OS_WIN32 or similar.
Thanks for the patch and the testing. Applied to both branches. Closing the bug as fixed, asking the bug reporter to open a new report if the other problems remain (and aren't reported yet). 2005-09-26 Sven Neumann <sven@gimp.org> * app/display/gimpdisplayshell.c (gimp_display_shell_new): applied patch from Robert Ögren that works around problem creating guides with a tablet on Windows by enabling extension events for the rulers. Fixes the first problem described in bug #168516.
*** Bug 304291 has been marked as a duplicate of this bug. ***