This happens because pix_coordinate nulls the texture coordinates in it's postrender method. I have no idea why it does this since it does not create or properly deallocate texture coordinates. It's also probably a good way to have another memory leak since setting a pointer to NULL without freeing memory is the one half of the memory leak equation.

I just commented out both lines in pix_coordinate :: postrender and it still seems to work fine. I won't commit the change until I test it more and/or hear from someone that this is not breaking something else.

cgc

On Aug 31, 2005, at 3:06 PM, B. Bogart wrote:

Hey all,

I realized some weirdness with textcoords in my patch the spreads one texture over many geos.

My chain looks something like this:

[gemhead] [pix_image] [loadbang] [pix_texture]-[mode 0, repeat 1 < [s group]

Then there are 40x30 geos each with:

[r group] [pix_coordinate] [separator] [square]

I've moved separator AFTER pix_coordinate as per cgc's instructions. As far as I know this is odd in Gem, since the pix_coordinate for each square is before the seperator they may interfere?

Anyhow When I just this patch the text-cords are all screwed up, (shifted slightly, actually looks like they are doing square texturing when they should be doing rectangular). Even once I send the proper coordinates using my python script.

In order to get everything to display right what I need to do is load the patch and then manually reinstanciate the pix_texture (clearing the texture) and then reloading it with pix_image.

Actually now that I think of it that is effectively removing the "mode 0, repeat 1" thing. Hmmm, actually I don't need to set "mode 0, repeat 1" now that I put separator after pix_coordinate...

Anyhow I answered by own question, maybe someone can enlighten me as to why this is happening.

Thanks all.

b> _______________________________________________ GEM-dev mailing list GEM-dev@iem.at http://lists.puredata.info/listinfo/gem-dev