dia release showstopper
Hello.
Thanks for fixing the beziergon bug. However, here is another
showstopper.
Open the demo 'UML-demo.dia' and press Ctrl-A -- dia crashed / core dump.
#0 0x40586efa in dependency_draw (dep=0x4, renderer=0x1) at dependency.c:284
284 if (strlen(dep-stereotype) != 0) {
(gdb) where
#0 0x40586efa in dependency_draw (dep=0x4, renderer=0x1) at dependency.c:284
#1 0x8184938 in ?? ()
#2 0x1 in ?? ()
#3 0x8073200 in end_render (renderer=0x768dc3) at render_gdk.c:242
#4 0x768dc3 in ?? ()
Open the demo 'ER-demo.dia' and press Ctrl-A -- dia hangs.
Maybe it's a gtk bug. Here is the stack trace:
#0 0x4018a51d in gtk_vruler_draw_ticks () from /usr/lib/libgtk-1.2.so.0
#1 0x401521bb in gtk_ruler_draw_ticks () from /usr/lib/libgtk-1.2.so.0
#2 0x40152754 in gtk_ruler_expose () from /usr/lib/libgtk-1.2.so.0
#3 0x401277fc in gtk_marshal_BOOL__POINTER () from /usr/lib/libgtk-1.2.so.0
#4 0x4015a3ad in gtk_signal_real_emit () from /usr/lib/libgtk-1.2.so.0
#5 0x40158125 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#6 0x40192b59 in gtk_widget_event () from /usr/lib/libgtk-1.2.so.0
#7 0x401974d3 in gtk_widget_real_draw () from /usr/lib/libgtk-1.2.so.0
#8 0x401279c8 in gtk_marshal_NONE__POINTER () from /usr/lib/libgtk-1.2.so.0
#9 0x4015a1f1 in gtk_signal_real_emit () from /usr/lib/libgtk-1.2.so.0
#10 0x40158125 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#11 0x40191419 in gtk_widget_draw () from /usr/lib/libgtk-1.2.so.0
#12 0x401910e1 in gtk_widget_idle_draw () from /usr/lib/libgtk-1.2.so.0
#13 0x80adeff in g_idle_dispatch ()
#14 0x80ad053 in g_main_dispatch ()
#15 0x80ad5c1 in g_main_iterate ()
#16 0x80ad73c in g_main_run ()
#17 0x40126043 in gtk_main () from /usr/lib/libgtk-1.2.so.0
#18 0x80842a7 in main (argc=-1073744676, argv=0x805a8f4) at main.c:40
Here is the configuration
dia (latest CVS version)
gtk+-1.2.9
glib-1.2.9
It did not work with gtk+ and glib 1.2.8 (redhat compiled version)
either.
Regards,
-Sebastian
horizontal lines and EPS export and stuff :)
mvh Kasper B. Graversen http://fermi.ruc.dk/~kbilsted/win
horizontal lines and EPS export and stuff :)
mvh Kasper B. Graversen http://fermi.ruc.dk/~kbilsted/win
ooh! dia -- sql
http://freshmeat.net/projects/dia2sql/ courtesy of squackers.. A.
Re: dia release showstopper
On 18 Mar 2001, Sebastian Fischmeister wrote:
Hello.
Thanks for fixing the beziergon bug. However, here is another
showstopper.
Open the demo 'UML-demo.dia' and press Ctrl-A -- dia crashed / core dump.
#0 0x40586efa in dependency_draw (dep=0x4, renderer=0x1) at dependency.c:284
284 if (strlen(dep-stereotype) != 0) {
(gdb) where
#0 0x40586efa in dependency_draw (dep=0x4, renderer=0x1) at dependency.c:284
#1 0x8184938 in ?? ()
#2 0x1 in ?? ()
#3 0x8073200 in end_render (renderer=0x768dc3) at render_gdk.c:242
#4 0x768dc3 in ?? ()
This one worked for me (using view - show all). It looks like there is
some stack corruption here. Especially frame #2, and the arguments to
dependency_draw (#0).
Open the demo 'ER-demo.dia' and press Ctrl-A -- dia hangs.
Maybe it's a gtk bug. Here is the stack trace:
#0 0x4018a51d in gtk_vruler_draw_ticks () from /usr/lib/libgtk-1.2.so.0
#1 0x401521bb in gtk_ruler_draw_ticks () from /usr/lib/libgtk-1.2.so.0
#2 0x40152754 in gtk_ruler_expose () from /usr/lib/libgtk-1.2.so.0
#3 0x401277fc in gtk_marshal_BOOL__POINTER () from /usr/lib/libgtk-1.2.so.0
#4 0x4015a3ad in gtk_signal_real_emit () from /usr/lib/libgtk-1.2.so.0
#5 0x40158125 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#6 0x40192b59 in gtk_widget_event () from /usr/lib/libgtk-1.2.so.0
#7 0x401974d3 in gtk_widget_real_draw () from /usr/lib/libgtk-1.2.so.0
#8 0x401279c8 in gtk_marshal_NONE__POINTER () from /usr/lib/libgtk-1.2.so.0
#9 0x4015a1f1 in gtk_signal_real_emit () from /usr/lib/libgtk-1.2.so.0
#10 0x40158125 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#11 0x40191419 in gtk_widget_draw () from /usr/lib/libgtk-1.2.so.0
#12 0x401910e1 in gtk_widget_idle_draw () from /usr/lib/libgtk-1.2.so.0
#13 0x80adeff in g_idle_dispatch ()
#14 0x80ad053 in g_main_dispatch ()
#15 0x80ad5c1 in g_main_iterate ()
#16 0x80ad73c in g_main_run ()
#17 0x40126043 in gtk_main () from /usr/lib/libgtk-1.2.so.0
#18 0x80842a7 in main (argc=-1073744676, argv=0x805a8f4) at main.c:40
I can reproduce this one. It looks like one of the "Standard - Line"
objects has a huge bounding box. I am looking into it.
Here is the configuration
dia (latest CVS version)
gtk+-1.2.9
glib-1.2.9
It did not work with gtk+ and glib 1.2.8 (redhat compiled version)
either.
The version number shouldn't make a difference to this particular problem.
James.
--
Email: [EMAIL PROTECTED]
WWW: http://www.daa.com.au/~james/
Re: dia release
On Mon, 19 Mar 2001, Hans Breuer wrote: Hi James, I like the idea of synchronizing the Dia release for the major platforms, but will probably be unable to test the required hacking until the next weekend. But please read on. I don't want to force a release schedule for the win32 port. I was just mentioning those changes I had made that I knew would be broken on win32. Release when you are ready. At 23:09 17.03.01 +0800, James Henstridge wrote: I would like to make a new release of dia soon. What we have in CVS has a lot of bug fixes over 0.86. So I have been tidying up a few things. These include: - linked up the new documentation to the menus for both gnome and non gnome builds. Hans will probably need to adjust things for win32. The relevant code is in app/commands.c(help_manual_callback), last 5 lines. Done. But there is a problem with all the png in Dia's cvs. To be readable on win32 they need to be marked as binary files in cvs. Otherwise the nice LF - CR,LF translation takes place ... I will tell Kevin to be a bit more careful when checking things in. I can't remember the CVS incantation to convert a text file to a binary file though. This is a bit strange though, as the cvswrappers file for GNOME CVS has the following in it: *.png -kb which should make sure all png files get added as binary files. - Add a --nosplash option for people who don't want a splash screen. - Fixed up the psprint print code so that it doesn't die on a SIGPIPE when you give an invalid printer command. It will now show a dialog if any SIGPIPEs were received during printing. This code is probably needs to be turned off for the win32 build. All your signal code isn't compileable with msvc, but it is not needed on windoze anymore anyway. I've changed the - never working - pipe execution to something more appropriate for windoze. Direct writing to the default printer. [ It is to introduce the next Win User FAQ: Q: When printing I get many pages of text starting with: %!PS-Adobe-2.0. What to do? A: Buy a postscript printer or implement GDI printing ... :-) ] The other option is to install ghostscript along with its print redirector so you can have a windows printer that filters things through gs. - fix up makefiles so that "make distcheck" completes. - fix up off by one error in beziershape_save which was adding an extra segment to beziergons during save. Are there any showstopper problems people can see that would hold up a release? (remember that there will always be another release). I beleive that what is in CVS fixes all the menu/locale issues, which was one of those annoying bugs (gtk+ and gnome based menu code should act as similarly as possible now). Not that I know of despite the never ending you-need-a-home-dir-faq ... Maybe it should be simply ignored and translated subdir of the installation dir? Is there some other appropriate place to put this sort of stuff on win32 if $HOME is not set? In the user's profile somewhere (that seems like the place where all the MS progs put things). There are only 4 calls to g_get_home_dir() in dia, judging by a quick grep. The call in preferences.c can probably be removed (I think everyone's preferences have been migrated from ~/.diarc to ~/.dia/diarc), and the code that uses it in app_procs.c could probably be moved to a function in dia_dirs.c, localising the use even further. It would be fairly easy to add a special case for win32 without $HOME in that file then. On the topic of GTK+ 2.0, it is not yet the right choice for the x11 port, as none of the gnome features will be able to be ported across for a while. At some point, we will branch dia so that we can port to gtk 2.0 while continuing to check in 1.2 bugfixes. Is this the right time to branch? Note that branching early means that we will need to merge changes between branches. IHMO we now could wait for the API freeze as well, but waiting much longer may lead to a frozen api, which lacks some highly needed capabilities. Additionally porting and using one of the mayor apps early could probably improve Gtk-2.0 release stability a lot. Well, I have been posting gtk bug reports when I find problems. And getting the new canvas up and running on gtk 2.0 helped find some problems. James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/
Re: dia release showstopper
On 18 Mar 2001, Sebastian Fischmeister wrote: Hello. Thanks for fixing the beziergon bug. However, here is another showstopper. [snip] Open the demo 'ER-demo.dia' and press Ctrl-A -- dia hangs. Tracked this bug down and fixed it. The fix is in CVS now: * objects/standard/line.c (line_update_data): line-extra.start_long and line-extra.end_long were not being initialised. This fixes potential problems with the bounding box of "Standard - Line" objects. Thanks for bringing this to my attention (bugs in the standard line would be very problematic). Still can't reproduce your first bug though. Remember that you can submit bugs to bugzilla.gnome.org any time you want. James. -- Email: [EMAIL PROTECTED] WWW: http://www.daa.com.au/~james/
