Re: Overhaul FLTK toolkit resize/redraw functions

[sergey plotnikov]

On 25/07/14 10:15, Andreas Weber wrote:

On 25.07.2014 09:57, sergey plotnikov wrote:

24.07.2014 21:44, Ben Abbott пишет:

On Jul 24, 2014, at 2:56 PM, sergey plotnikov <address@hidden> wrote:

I've tested this, since I'm able to replicate the bug. And actually Ben was right, lines are kind of hidden, but in my case, if I don't set axes color to 'none' they are hidden just partially (see junk2.pdf in attachment). 

And to my understanding it might be somehow related to the sorting algorithm we are choosing in gl2ps_renderer.
<junk2.pdf><junk.pdf>

Sergey,

Can you try the attached version?  This is the same as Adreas' but keeps the duplicate "Fl::check ();"

Ben

Ben,

I've just tried it, but it has the same result for me -- lines are
partially missing (see junk.pdf). As a matter of fact, it was happening
even before applying initial Andreas' patch. So, it looks like your
problem has a different origin than one from the bug #42534, which I'm
able to replicate.

I may probably repeat it here as well. In my case problem *almost*
disappears if I use 3D rendering options (e.g., commenting out a place
in __fltk_print__.m where "is2D" mark is being added). "Almost" means
that even in this case parts of patches maysometimes still be missing
after chart is printed (see junk33.pdf, just an illustration, I will
post simple code leading to the same result later).

Sergey

Hi Sergey,in a previous mail (some minutes ago) I asked you to change
gl2ps-renderer.cc

Could you please try something more?
I would set gl2ps_sort = GL2PS_NO_SORT; for testing because the
description for GL2PS_SIMPLE_SORT (when correctness is not crucial.)
doesn't sound appropriate for us.

Then please remove GL2PS_SILENT and GL2PS_OCCLUSION_CULL and cout the
return value of gl2psBeginPage, perhaps we see an error here?
-- Andy

Hi Andy,

I've just found some time for testing when you already have pushed csets you asked me to test (sorry). I've given it try cross-compiling your cset c9f960441513 "Overhaul FLTK plotting and printing ..." using the latest MXE-Octave (9985fed66b1c, "index.html: update for current package list...").

On my Win7 (x32) virtual machine there are still some missing primitives even with glFlush() added. E.g., the following piece of code produces the file you may find attached:

graphics_toolkit fltk
figure; for ix=1:6, subplot(3,2,ix); plot(sin(1:10)); legend('sin'); end
print -depsc junk55.eps

And again, if I commented out "is2D" adding in __fltk_print__.m, the most of primitives are shown.

Sergey

junk55.eps
Description: image/eps