Qbasicnews.com

Full Version: Possible profiler bug
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
In some cases the profiler fails to output the profile.txt file.

As I'm trying it in a fairly complicated program, I can't say why does it happen. In fact it's a port of my jpeg viewer (coming soon..).
If I run it without a command line parameter, so the program ends with an error, profiler creates its file. But if a jpeg file is passed, and all the jpeg decoding is performed, there is no profile.txt at all.

If it's needed for testing I can post somewhere the present version of the viewer.

In fact the viewer does ntot seem more complicated than Jark's raytracer, where the profiler worked ok...
I found it!

Code:
f$=command$
print "hello"
sleep

If something is passed in the command line, profiler does not output its
file. If the command line is blank, profiler works.
humm, works just fine here, profile.txt is generated in any case, as long as the program is compiled with -profile... No matter if the compiled program receives any arguments or not.
I checked and I'm, running the prerelease version 0.14....I should'nt erase the stable versions so quickly...

My command line is fbc is <file> -profile -g
If -g is missing no profile.txt is issued at all.

BTW: WHAT A GREAT TOOL!!!
Angelo: I reinstalled V 0.13 and i still have the problem...

Edited: I have the same problem at home where I never overwrote V0.13
Windows 2000 SP4 both places

BTW: I'm using the profiler in spite of the bug...I'm only willing to help discover what's happening.
Humm, I've had not this problem, but I'll try something more when I'm at home later.

Anyway, I'm in the process of rewriting the whole profiling code in order to better support recursive functions... I'll try to fix the issue while I rewrite the beast.