Henrik Grubbström (Lysator) @ Pike (-) developers forum wrote:
a) module.pmod as file name is not very helpfull. Would it be possible to back-up one step in the directory tree when the node is module.pmod to get a sensible filename in the log?
Sure, modify interpret.c:do_trace_call().
Nice :) Here's a very simple diff for it:
Index: src/interpret.c =================================================================== RCS file: /cvs/Pike/7.6/src/interpret.c,v retrieving revision 1.353 diff -r1.353 interpret.c 1425c1425,1429 < while((f=STRCHR(file,'/'))) ---
while((f=STRCHR(file,'/'))) { if (!STRNCMP( f, "/module.pmod", 12)) { //f[0] = 0; break; }
1426a1431
}
Index: src/interpret.c =================================================================== RCS file: /cvs/Pike/7.7/src/interpret.c,v retrieving revision 1.361 diff -r1.361 interpret.c 1445c1445,1449 < while((f=STRCHR(file,'/'))) ---
while((f=STRCHR(file,'/'))) { if (!STRNCMP( f, "/module.pmod", 12)) { //f[0] = 0; break; }
1446a1451
}
But I've not been able to cut off the "/module.pmod" which is a bit redundant, but removing the comment for the if-statement in the diff above seems to result in a bad pike. Guess I'm not allowed to fiddle with the char * we get from pike_string->str.
b) Is there any tools around for parsing the trace output into some more readable format/graphs?
Not yet (AFAIK), but it shouldn't be that hard to do.
Thought so too.. see if I ever get around to do anything about it.
//K