Hi all,
I have a real pickle in trying to debug Pd. Namely, when I grep-ed every single instance of pdtk_canvas_getscroll and commented it out, it still gets invoked somehow and I am at a loss how. I know it occurs as many times as there are objects selected that have been moved (so it is definitely per-object)
The only thing I managed to isolate so far is the following code: g_editor.c
void gobj_displace(t_gobj *x, t_glist *glist, int dx, int dy) { if (x->g_pd->c_wb && x->g_pd->c_wb->w_displacefn) (*x->g_pd->c_wb->w_displacefn)(x, glist, dx, dy); }
I am however unable to decipher what function is exactly being called by (*x->g_pd->c_wb->w_displacefn). I only found in g_canvas.h definition of w_displacefn as part of _widgetbehavior struct but nowhere (except in iemgui objects) is it linked to anything I can actually understand. Where does this struct get initialized?
Any help on this one would be most appreciated as that will put a stop on me pulling my hairs out :-)
Best wishes,
Ico
Never mind... My brain finally decided to get out of bed and join me in my office... Apologies for spam...
On Thu, 2010-11-18 at 11:22 -0500, Ivica Ico Bukvic wrote:
Hi all,
I have a real pickle in trying to debug Pd. Namely, when I grep-ed every single instance of pdtk_canvas_getscroll and commented it out, it still gets invoked somehow and I am at a loss how. I know it occurs as many times as there are objects selected that have been moved (so it is definitely per-object)
The only thing I managed to isolate so far is the following code: g_editor.c
void gobj_displace(t_gobj *x, t_glist *glist, int dx, int dy) { if (x->g_pd->c_wb && x->g_pd->c_wb->w_displacefn) (*x->g_pd->c_wb->w_displacefn)(x, glist, dx, dy); }
I am however unable to decipher what function is exactly being called by (*x->g_pd->c_wb->w_displacefn). I only found in g_canvas.h definition of w_displacefn as part of _widgetbehavior struct but nowhere (except in iemgui objects) is it linked to anything I can actually understand. Where does this struct get initialized?
Any help on this one would be most appreciated as that will put a stop on me pulling my hairs out :-)
Best wishes,
Ico
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On Thu, 18 Nov 2010, Ivica Ico Bukvic wrote:
I have a real pickle in trying to debug Pd. Namely, when I grep-ed every single instance of pdtk_canvas_getscroll and commented it out, it still gets invoked somehow and I am at a loss how. I know it occurs as many times as there are objects selected that have been moved (so it is definitely per-object)
Use <execinfo.h>'s backtrace() so that sys_vgui prints the name of the calling function just before each line is printed while in -d 1 mode.
| Mathieu Bouchard ---- tél: +1.514.383.3801 ---- Villeray, Montréal, QC
On Nov 18, 2010, at 12:50 PM, Mathieu Bouchard wrote:
On Thu, 18 Nov 2010, Ivica Ico Bukvic wrote:
I have a real pickle in trying to debug Pd. Namely, when I grep-ed
every single instance of pdtk_canvas_getscroll and commented it
out, it still gets invoked somehow and I am at a loss how. I know
it occurs as many times as there are objects selected that have
been moved (so it is definitely per-object)Use <execinfo.h>'s backtrace() so that sys_vgui prints the name of
the calling function just before each line is printed while in -d 1
mode.
That sounds useful, do you have an example of how that you could share?
.hc
Programs should be written for people to read, and only incidentally
for machines to execute.
- from Structure and Interpretation of Computer Programs
On Thu, 18 Nov 2010, Hans-Christoph Steiner wrote:
That sounds useful, do you have an example of how that you could share?
There's short_backtrace in gridflow/src/gridflow.cxx
And there's another similar thing in segfault.cxx
They also use <cxxabi.h> to decode C++ symbols (those that start with _Z)
| Mathieu Bouchard ---- tél: +1.514.383.3801 ---- Villeray, Montréal, QC
Hi Mathieu,
This is a great advice! However, when adding blargh() function from the gridflow.c inside the sys_vgui (s_inter.c) so that when the debug is on it outptus a trace for each call, it appears it only manages to do one level which is obviously the sys_vgui call itself. Why is it not able to trace further back?
void blargh(void) { #ifdef MACOSX fprintf(stderr,"unhandled exception\n"); #else int i; void *array[25]; int nSize = backtrace(array, 25); char **symbols = backtrace_symbols(array, nSize); for (i=0; i<nSize; i++) fprintf(stderr,"%d: %s\n",i,symbols[i]); free(symbols); #endif }
void sys_vgui(char *fmt, ...) {
<snip>
if (sys_debuglevel & DEBUG_MESSUP) {
blargh();
fprintf(stderr, "%s", sys_guibuf + sys_guibufhead);
}
<snip>
}
Before you invest too much time in that, you should know that that
stuff has changed a lot in 0.43, so its likely anything you did in
0.42 would not apply to 0.43
.hc
On Nov 21, 2010, at 10:40 AM, Ivica Ico Bukvic wrote:
Hi Mathieu,
This is a great advice! However, when adding blargh() function from
the gridflow.c inside the sys_vgui (s_inter.c) so that when the debug is
on it outptus a trace for each call, it appears it only manages to do one level which is obviously the sys_vgui call itself. Why is it not
able to trace further back?void blargh(void) { #ifdef MACOSX fprintf(stderr,"unhandled exception\n"); #else int i; void *array[25]; int nSize = backtrace(array, 25); char **symbols = backtrace_symbols(array, nSize); for (i=0; i<nSize; i++) fprintf(stderr,"%d: %s\n",i,symbols[i]); free(symbols); #endif }
void sys_vgui(char *fmt, ...) {
<snip>
if (sys_debuglevel & DEBUG_MESSUP) { blargh(); fprintf(stderr, "%s", sys_guibuf + sys_guibufhead); }
<snip>
}
"A cellphone to me is just an opportunity to be irritated wherever you
are." - Linus Torvalds
-
Hans-Christoph Steiner -
Ivica Ico Bukvic -
Mathieu Bouchard