Re: problems running gtk on a SGI




Moses DeJong <dejong@cs.umn.edu> writes:

> I ran into some problems with gtk 0.99.8 and gimp 0.99.22.
> The gtk errors and crashes happen on an SGI 6.2 IRIX box.
> The text of these problems is included below.
> 
> Mo DeJong
> dejong@cs.umn.edu
> gimme multimedia group
> 
> 
> 
> there are a lot of warnings about redefined constants (in every compiled
> file).
> 
[...]
> /usr/include/limits.h:227: warning: `DBL_MAX' redefined
> /opt/gnu/lib/gcc-lib/mips-sgi-irix5.3/2.7.2/include/float.h:63: warning: this is the location of the previous definition

This is, I would say, a problem with your gcc configuration. GLIB
includes both the float.h and limits.h header files, and apparently
one is coming from gcc and the other from your system header files
and they are conflicting.

> I get the error "** ERROR **: sigsegv caught" when I click on the
> "color selection" button in testgtk.

OK, I think I know which problem this is. It is a segfault that
is occuring in the XIM code. You can work around it for now
by compiling GTK with --disable-xim; I'm working on a fix now.

> The same things happens in "cursors"
> 
> when I open the dnd window I get these errors.
> 
> ** WARNING **: file gtkpixmap.c: line 83 (gtk_pixmap_new): "val != NULL"
> 
> ** WARNING **: file gtkfixed.c: line 126 (gtk_fixed_put): "widget != NULL"
> 
> ** WARNING **: file gtkwidget.c: line 1228 (gtk_widget_show): "widget != NULL"
> 
> ** WARNING **: file gtkpixmap.c: line 83 (gtk_pixmap_new): "val != NULL"
> 
> ** WARNING **: file gtkfixed.c: line 126 (gtk_fixed_put): "widget != NULL"
> 
> ** WARNING **: file gtkwidget.c: line 1228 (gtk_widget_show): "widget != NULL"
> 
> ** WARNING **: file gtkpixmap.c: line 83 (gtk_pixmap_new): "val != NULL"
> 
> ** WARNING **: file gtkfixed.c: line 126 (gtk_fixed_put): "widget != NULL"
> 
> ** WARNING **: file gtkwidget.c: line 1228 (gtk_widget_show): "widget != NULL"

Hmmm. Are you compiling with a separate builddir and srcdir? If
so, you should run testgtk with 'make test'. Otherwise, it won't
find the necessary pixmaps.

> then a small (about 100 x 100 pixels) empty square shows up on
> the screen. I can drag this box around but it does not seem to do anything.
> 
> I can grab another box off the "Drag me!" (it is the same empty square) button
> an drop it in the "To" button and the dialog comes up. But if I drag two
> of these boxes and bring up two dialogs I get this message when closing
> the dialog box.
> 
> ** WARNING **: file gtkwidget.c: line 1189 (gtk_widget_destroy): "widget != NULL"

This is something I see. (The problem is that DND bypasses the
GTK layer, so it ignores the modal grab, and another drag can
occur while not expected. It should be pretty easy to prevent
this in testgtk.c, though the problem really deeper.

> In the "Handle box test" the drag in and out of the window works
> but the Icons for the folder do not show up.

This is the same problem as the DND thing.
 
> When a selection is made in "test selection" mode and then the 
> "GetTargets" button is pressed this error message shows up.
> 
> Selection "TARGETS" was not returned as atoms!


> in the text test only the string "And even colored text" at the end
> of the test shows up. The entire gpl and the bg image is missing.

This is the same srcdir != builddir problem.

> the tree test exits with a sigsegv caught message.

Same XIM problem.
 
> I ran these same tests on linux and sun builds and they worked just fine.
> 
> I also noticed that the testgtk "reparent" example did not work
> properly on my linux box. On the sun box the window jumps from
> one window to the next but on the linux box the "hello world"
> message stays in the "Frame 1" window and a copy shows up in the
> "Frame 2" window. The linux is a 2.0.30 intel box.

Thanks for the report,
                                        Owen



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]