PDA

View Full Version : GLIntercept 1.0



sqrt[-1]
09-28-2011, 01:35 AM
I have just posted GLIntercept 1.0 to the website:
http://code.google.com/p/glintercept/

Note that GLIntercept was mainly designed as an OpenGL 1.0-2.1 debugger. So while basic function logging should work on all OpenGL versions, the more advanced features (image/frame buffer/shader editor) may or may not work in OpenGL 3.0+. (especially when using a core OpenGL profile)

I have mostly just added the logging for newer functions and fixed some minor bugs. (note you can add you own functions definitions to the text file parsed on startup - so you can log extensions as they come out)

Thanks to alfonse for his XML listing of extensions - made my job much easier.

If someone has an OpenGL 4.0 card, can they test the OpenGL samples pack 4.0+ examples?
http://www.g-truc.net/project-0026.html

I only expect basic logging to work with 4.0, but I want make sure errors are not generated. (yes, I should get myself an OpenGL 4.0+ card - but have been on consoles for the last few years - and now unemployed since THQ-OZ went down)

If you test with you application and find any obvious bugs, please let me know.

Groovounet
09-28-2011, 06:00 AM
It seems that the installer is corrupt... at least I have an error when I launch it.

sqrt[-1]
09-28-2011, 06:58 AM
Humm, what is the error?
I just downloaded and checked both versions - they both seem fine.(32/64 bit versions)

What OS are you running?

Groovounet
09-28-2011, 08:04 AM
A meaningful less error 216 at 76A6BBCC

Seven 32.

I just downloading again to double(or at least triple) check but I still have the same error when I launch the installer...

mhagain
09-28-2011, 08:43 AM
I've just downloaded and installed the 32-bit version without trouble.

Many many thanks for this tool by the way, it's been very valuable to me in the past. :)

Groovounet
09-28-2011, 02:21 PM
It worked at home. I wonder if it had to do with administrator writes.

sqrt[-1]
09-28-2011, 04:25 PM
OK - I re-packed the installer with the latest version of inno Setup and re-uploaded the files.

Let me know if they now work at your work.

Groovounet
09-28-2011, 05:35 PM
I did some quick tests with the OpenGL Samples.

I am not sure everything is perfect. Do you handle the differences between core and compatibility profiles?

Also I had an error store in glilog.txt "GL ERROR - Function glDrawElementsInstancedBaseVertexBaseInstance generated error GL_INVALID_ENUM" with ogl-420-image-load but I have no error without GLIntercept.

In the AMD ARB_debug_output, this error is "Using glIsEnabled in a core context with parameter <cap> and enum '0xde1' which was removed from Core OpenGL (GL_INVALID_ENUM)".

sqrt[-1]
09-28-2011, 06:16 PM
So you are testing on AMD?
It seems that I have missed removing a lot of glIsEnabled calls when using a core profile (in this case GL_TEXTURE_2D)

Seems that Nvidia does not generate errors for these cases?

Do basic logging (not image or shader dumping) work OK?

sqrt[-1]
09-28-2011, 06:32 PM
Oh I think I see what the issue is, a lot of the OpenGL 4.0 examples use glBindProgramPipeline, instead of a standard program binding, so the code is falling back to test for fixed function enables. (error on my part)

sqrt[-1]
09-28-2011, 10:57 PM
OK I have uploaded a new version (1.0.1) that should not have the glIsEnabled(GL_TEXTURE_2D) errors in core profile - strange that Nvidia drivers did not flag this as an error?

Groovounet
09-29-2011, 03:58 AM
Still installer error at work...

Will try version 1.0.1 at home and let you know.

Groovounet
09-29-2011, 11:40 AM
Otherwise, a zip file is good!

kyle_
09-29-2011, 12:02 PM
Still installer error at work...

Will try version 1.0.1 at home and let you know.

Strange stuff.
Works like charm on my side. What OS you're on?



Otherwise, a zip file is good!

Compile from sources ;)

Groovounet
09-29-2011, 12:17 PM
Seven 32, see up ;)

That's another option but I guess if I have a problem, I am not the only one so I just report the bug. :p

sqrt[-1]
09-29-2011, 05:12 PM
Yes, I will do a zip file - the only reason for the installer is so that the paths in the sample config files are updated to the install location.

Might actually wait a little to see if any more bugs fall out - have already found a few other minor ones.

Groovounet
09-29-2011, 05:19 PM
I haven't done a full test yet but I will and let you know on time.

sqrt[-1]
11-04-2011, 10:03 PM
OK I have just uploaded a new version 1.0.2
This fixes some minor bug in the function parsing and adds the NV_path_rendering extension

The NV_path_rendering extension was actually a pain to support as it has the first OpenGL call to return a float value. (glGetPathLengthNV) I ended up having to do a manual logging of it.

I also included a manual install zip file for people that have trouble running the installers.

http://code.google.com/p/glintercept/