Aw: Re: New panel in trunk

Fabian Nowak timystery at arcor.de
Thu Oct 13 13:00:26 CEST 2005


Am Donnerstag, den 13.10.2005, 12:36 +0200 schrieb Jasper Huijsmans:
> Fabian Nowak schreef:
> > hi!
> > 
> > first of all, using svn from yesterday afternoon. rev 18213?
> > 
> 
> Hmm, lots of weirdness here :(
> 
> > removing separator while manager dialog was open after having added
> > several separators just deleted my bottom panel:
> > 
> > #0  0xb7884576 in poll () from /lib/tls/libc.so.6
> > #1  0xb792f57b in g_main_context_check () from /usr/lib/libglib-2.0.so.0
> > #2  0xb792fbd8 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
> > #3  0xb7c39989 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
> > #4  0x0804f44e in panel_app_run (argc=1, argv=0xbfc5fd64) at
> > panel-app.c:436
> > #5  0x0804ecf0 in main (argc=1, argv=0xbfc5fd64) at main.c:164
> > 
> 
> Actually, I think the panel wasn't actually removed, but hidden (or 1px 
> high). You would probably see error messages if a panel would be 
> unexpectedly deleted.
> 
> I just noticed that this happens when there are only separators on the 
> panel. I fixed the separator, but it may be better to simply force the 
> panel to the height (width) that was set in the dialog.

nope. panel was removed in file panel.xml, and there was no one pixel
line and it didn't show up when moving the mose over the bottom lines.

> 
> 
> > 
> > having added new bottom panel and inserted several items into the first
> > launcher made the bottom panel disappear. Ctrl-c and bt gave the same
> > results as previous experiment. kill, run made it reappear with all
> > settings just before disappearing. strange.
> 
> Well, it makes sense to not save settings when killed, because that 
> would mean something went wrong and it's safer to return to the old 
> situation.

nope. this time they were saved, most likely just a little time before
crashing. this leads to the conclusion, that the panel saves each added
item instantly, while the launchers do not save their setup when closing
the properties dialog of a launcher. actually, i think it might be best
to save any new item in the manager dialog right when adding the next
one or moving an item.
might be some overhead, but would not let the user enter the entries
from new all day long...

> 
> > 
> > 
> > inserting plugin after plugin, closing manager, reopening manager,
> > inserting login-in/lock buttons gave me a +-cursor and froze desktop,
> > inserting item was not possible, couldn't select other windows or
> > Alt-Tab through windows. had to kill -9 from console. thus no backtrace
> > available.
> 
> Hmm, a drag and drop error. Not sure there's much I can do about that. 
> Could be a problem in Gtk, or a problem with the way I handle dnd events.
> 
> > 
> > 
> > inserting four starters, one separator, editing first starter, second
> > starter worked. shift-tabbing when editing third starter into
> > image-combo-box (and pressing space?) made 
> > 
> > (xfce4-panel:3149): Gtk-CRITICAL **: gtk_toggle_button_set_active:
> > assertion `GTK_IS_TOGGLE_BUTTON (toggle_button)' failed
> > 
> > (xfce4-panel:3149): Gtk-CRITICAL **: gtk_toggle_button_set_active:
> > assertion `GTK_IS_TOGGLE_BUTTON (toggle_button)' failed
> > 
> > Program received signal SIGSEGV, Segmentation fault.
> > [Switching to Thread -1218070848 (LWP 3149)]
> > 0xb796ecf3 in g_type_check_instance_cast ()
> > from /usr/lib/libgobject-2.0.so.0
> > (gdb)
> > 
> 
> That's just weird.
> 
> > so, despite being able to mouse-select windows this time, xfce4-panel
> > had to be killed -9 from console to reenable giving keyboard input. the
> > already completely edited starters in this session have not been saved,
> > have to reedit them.
> > 
> > inserting one more starter gave the well-known +-cursor. kill -9...
> > 
> 
> If someone knowledgeable could have a look at the dnd handling in the 
> panel, that would be nice. It is rather complicated, but I believe I'm 
> doing things correctly.
> 
> > (gdb) run
> > Starting program: /usr/local/bin/xfce4-panel 
> > [Thread debugging using libthread_db enabled]
> > [New Thread -1217370432 (LWP 3306)]
> > 
> > Program received signal SIGSEGV, Segmentation fault.
> > [Switching to Thread -1217370432 (LWP 3306)]
> > 0xb7a13868 in g_signal_connect_data () from /usr/lib/libgobject-2.0.so.0
> > 
> 
> Weird. Memory corruption? I don't know.
> 
> > now the menus are weirdly hacked, in my graphics menu the panel-object
> > is missing (had name, no description, x-xfce-icon, no command), instead
> > the first one following is taken. Seems to be due to missing "Exec="
> > line in launcher.rc. fix it by saving this empty line or do not require
> > it when reading the rc file. in my inkscape entry, there is no exec-line
> > (i did not forget inserting it, but perhaps i did not move the input
> > cursor out of the icon entry field, thus the old info, which was empty,
> > was read and not saved), thus it is not shown in the menu. this graphics
> > menu was saved twice, the following starters were moved to the right by
> > one, the last (empty) starter was overwritten.
> > 
> 
> I'll have another look at the reading/saving of the config file.
i again propose to set entry="", command="" if xfce_rc_read doesn't
return anything. this should fix the problem. hmm, i could try mysql,
but i should better go on learning...
> 
> > so, i had lots of trouble, and very few backtraces this morning....
> > 
> 
> Needless to say, I can reproduce none of these problems :(
neither could i after a friend telling my software would crash when
performing certain actions. seems to be law of software writing.
> 
> Thanks,
> 	Jasper
> _______________________________________________
> Xfce4-dev mailing list
> Xfce4-dev at xfce.org
> http://foo-projects.org/mailman/listinfo/xfce4-dev




More information about the Xfce4-dev mailing list