Community
Participate
Working Groups
Run snippet 82 in eclipse.swt.snippets. Try switching around tabs; you will notice that there is a noticeable delay in between the switch.
Additional findings: switching between a hidden tab (selected through the chevron list) does not cause any sort of delay at all. There must be something that selection through the chevron list is not doing that selecting directly through the tab is.
New Gerrit change created: https://git.eclipse.org/r/83639
(In reply to Eclipse Genie from comment #2) > New Gerrit change created: https://git.eclipse.org/r/83639 This bug was discussed on the SWT call. Feedback: 1) We should take extra care with using context_iteration. It can change the order of events and potentially lead to crashes/NPEs. Ex: https://git.eclipse.org/r/#/c/83904/ Can you verify that event ordering has not changed? Ex, - In ControlExample, you can add mouse listeners. Ensure that things like double-click, defaultSelection, Mouse-up, motion events occur in the same order as before. Is there a way we can get this to work without using context_iteration btw? If we check mouse state, do we still need context iteration? For both X11 and Wayland: 2) Does Double click + Drag still work? 3) Does DnD still work in CStyled text? I.e, can you drag bits of code around? 4) Does tripple click work in Styled text? (select whole line?) 5) The comment in the bug didn't get updated in patch set 2?
(In reply to Leo Ufimtsev from comment #3) One also needs to test the DnD in the editors not just StyledText.
(In reply to Leo Ufimtsev from comment #3) > (In reply to Eclipse Genie from comment #2) > > New Gerrit change created: https://git.eclipse.org/r/83639 > > This bug was discussed on the SWT call. Feedback: > > 1) We should take extra care with using context_iteration. It can change the > order of events and potentially lead to crashes/NPEs. Ex: > https://git.eclipse.org/r/#/c/83904/ > > Can you verify that event ordering has not changed? > Ex, > - In ControlExample, you can add mouse listeners. Ensure that things like > double-click, defaultSelection, Mouse-up, motion events occur in the same > order as before. > > Is there a way we can get this to work without using context_iteration btw? > If we check mouse state, do we still need context iteration? > > > For both X11 and Wayland: > 2) Does Double click + Drag still work? > > 3) Does DnD still work in CStyled text? I.e, can you drag bits of code > around? > > 4) Does tripple click work in Styled text? (select whole line?) > > 5) The comment in the bug didn't get updated in patch set 2? Everything has been tested and been working with the patch that I have made, thankfully. The patch is working as intended.
Gerrit change https://git.eclipse.org/r/83639 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=04862c5f2af9148ae0d045ae927e8d64de4bd5b4
Also tested on Gtk2/Gtk3. jUnit tests and DnDExample as well as child Eclipse seem to function well. Thank you for fix. If no major issues are reported, we should probably backport in a few weeks time.
New Gerrit change created: https://git.eclipse.org/r/84825
Reverted patch for now due to report of DnD issue from Alex
New Gerrit change created: https://git.eclipse.org/r/84826
I've added a small patch that adds the additional gdk_threads_leave() call which seems to be used before every g_main_context_iteration() call. Tests seem to run fine on my machine, but I will leave it to Alex to see if it fixes his issue.
Gerrit change https://git.eclipse.org/r/84825 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=e83f9f1828a01db543458c9a2bdd403be4bb8421
Please retarget to M4 if there is a chance to complete in M4 time frame
Retarget for M5 as there is patch for review in M4.
New Gerrit change created: https://git.eclipse.org/r/86861
Gerrit change https://git.eclipse.org/r/86861 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=b48d644b780b20250a7212b8992d1d42bdd70d32
Ian, this broke Table/Tree MULTI selection DnD cause the mouse click on drag start is processed and selects single element only. Please let me know whether we should revert or you can come up with a patch for that fast.
So multi selection DnD is broken on both X11 and wayland. If you test on wayland though you must work with http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=d0fe4cf655fd77c7ae6184a4713c38c4ee88f04f or later commit as wayland uses cairo image surface not xlib.
(In reply to Alexander Kurtakov from comment #17) > Ian, this broke Table/Tree MULTI selection DnD cause the mouse click on drag > start is processed and selects single element only. Please let me know > whether we should revert or you can come up with a patch for that fast. Ill get to it now, Ill see if i can fix it ASAP without the need to revert
New Gerrit change created: https://git.eclipse.org/r/87142
Gerrit change https://git.eclipse.org/r/87142 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=2a32fc9def3cbbb5db971f5873c47b5c79127318
Created attachment 265954 [details] Patch to get tree multiselection Dnd (not working currently)
Alex, I have attached a patch that should give an outline to what I am trying to do. Currently, there is an issue with the compiler not seeing the selectionProc() method inside Display, even though I have put one there explicitly. The use of gtk_tree_selection_set_select_function() would be to set a function that just returns false, as we need it to decline any selections. Once we are on release event, it will check to see if it is in a condition to clear the other selections or not. THe logic seems sound, but the main thing is the implementation of GtkTreeSelectionFunc as one of the parameters. This is all seen within the release_event() and click_event() of Tree for a Proof of Concept.
Comment on attachment 265954 [details] Patch to get tree multiselection Dnd (not working currently) forgot to push the widget changes that make this semi-work
Created attachment 265955 [details] Patch to get tree multiselection Dnd (not working currently)
New Gerrit change created: https://git.eclipse.org/r/87440
New Gerrit change created: https://git.eclipse.org/r/87713
New Gerrit change created: https://git.eclipse.org/r/87967
Gerrit change https://git.eclipse.org/r/87967 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=a6ef223e37fdd4a902c9d9880134483238275ef7
(In reply to Eclipse Genie from comment #28) > New Gerrit change created: https://git.eclipse.org/r/87967 > Gerrit change https://git.eclipse.org/r/87967 was merged to [master]. > Commit: > http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/ > ?id=a6ef223e37fdd4a902c9d9880134483238275ef7 Accidentally linked to wrong bug. This was is related to another bug: Bug 505591 - Add test cases for Browser.evaluate() to browser test suite. (In reply to Eclipse Genie from comment #29)
New Gerrit change created: https://git.eclipse.org/r/88650
Moving to M7 as we are in M6 milestone week.
Gerrit change https://git.eclipse.org/r/88650 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=860641a84d915424de3c6c214008ed5fc87c3a6f
The recent patch is generating a build warning: 860641a84d915424de3c6c214008ed5fc87c3a6f https://git.eclipse.org/r/#/c/88650/ The warning is printed in Gtk2 and Gtk3: > os.c: In function ‘Java_org_eclipse_swt_internal_gtk_OS__1gtk_1tree_1selection_1set_1select_1function’: > os.c:14786:67: warning: passing argument 2 of ‘gtk_tree_selection_set_select_function’ from incompatible pointer type [-Wincompatible-pointer-types] > gtk_tree_selection_set_select_function((GtkTreeSelection *)arg0, (GtkTreeSelectionFunc *)arg1, (gpointer)arg2, (GDestroyNotify)arg3); > ^ > In file included from /usr/include/gtk-2.0/gtk/gtk.h:201:0, > from os.h:30, > from os_structs.h:19, > from os.c:20: > /usr/include/gtk-2.0/gtk/gtktreeselection.h:81:18: note: expected ‘GtkTreeSelectionFunc {aka int (*)(struct _GtkTreeSelection *, struct _GtkTreeModel *, struct _GtkTreePath *, int, void *)}’ but > argument is of type ‘gboolean (**)(GtkTreeSelection *, GtkTreeModel *, GtkTreePath *, gboolean, void *) {aka int (**)(struct _GtkTreeSelection *, struct _GtkTreeModel *, struct _GtkTreePath *, > int, void *)}’ > void gtk_tree_selection_set_select_function (GtkTreeSelection *selection, > ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The reason is that in OS.java:gtk_tree_selection_set_select_function The cast is wrong: * @param func cast=(GtkTreeSelectionFunc *) should be: * @param func cast=(GtkTreeSelectionFunc) # (no * at the end) As per gtk_tree_selection_set_select_function documentation. This removes the build warning. @Ian, can you test everything still works with the fixed cast and submit a relevant patch? Thank you.
This caused may test failures due to "SWTError: No more callbacks", see bug 513938.
(In reply to Markus Keller from comment #35) > This caused may test failures due to "SWTError: No more callbacks", see bug > 513938. Thank you for pointing this out. I've got a fix lined up for this, will submit today/tomorrow latest.
New Gerrit change created: https://git.eclipse.org/r/93465
Gerrit change https://git.eclipse.org/r/93465 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=81633b430a87caf2f0c020f10e108754d81a4415
New Gerrit change created: https://git.eclipse.org/r/93568
Gerrit change https://git.eclipse.org/r/93568 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=3f92612774ff0dda4e8e9c86127bb0f3a880c329
All known memory leaks fixed.
*** Bug 513829 has been marked as a duplicate of this bug. ***
New Gerrit change created: https://git.eclipse.org/r/94069
(In reply to Eclipse Genie from comment #43) > New Gerrit change created: https://git.eclipse.org/r/94069 There seems to be some delays/trouble with gerrit atm. Will followup/review tomorrow.
Gerrit change https://git.eclipse.org/r/94069 was merged to [master]. Commit: http://git.eclipse.org/c/platform/eclipse.platform.swt.git/commit/?id=820971df18851ed5497520d03bf061ad9990dadf
closing as patch has been merged to master, removing the warnings from comment 34
This guy: https://git.eclipse.org/r/#/c/88650/ Caused a regression: Bug 514531 – [GTK3][regression] Launch config is not opened after creation https://bugs.eclipse.org/bugs/show_bug.cgi?id=514531
!x11 guards added to code and known regressions addressed on wayland. Closing for now.
