Re: [Gimp-developer] GSoC Status



Hi Carlos,

While I appreciate the work you did, it really falls short of what is
expected of a GSoC.

That by itself wouldn't' be a problem if you had applied yourself to the
project but it happened that you weren't able to deliver what you promised.
Much to the contrary, you had few coding issues and was able to catch up
stuff quickly at the beginning, so there was a lot of potential.

GSoC is not like a contract where you deliver stuff and that's it. You
almost didn't interact with the GIMP community at all and just now appeared
after mid-term. Frankly, it was embarrassing to send you emails asking
about the work and have them unanswered, that's not acceptable.

That said, it's nothing personal and if you decide to go forward outside
GSoC, which would be really great, I'll help you out.

Victor Oliveira
On Sep 18, 2013 11:10 PM, "Carlos Zubieta" <czubieta dev gmail com> wrote:

Hi Victor,

I know it's been a long time since the last time I wrote to you. So I'll
list all the progress I've had since then:

* I did a gegl-cl-random re-structuration so it fits better the gegl
directory structure, as you mentioned.
* I ported to gegl-random
  - noise-hsv
  - noise-cie-lch
* And added OpenCL support to
  - noise-hurl
  - noise-slur
  - noise-pick
  - alien-map
  - red-eyed-removal
  - dot
  - posterize

* I fixed a bug in box-min and box-max causing a segfault.
* Also, I've already ported noise-hsv and noise-cie-lch to OpenCL. But
when I wanted to test them I realized that there is no CL support for
those color spaces, since there are extensions in babl. So I think we
need to discuss if there will be OpenCL support for the babl extensions.

* I tried to use gegl-tester but it's kind of limited (you can see the
options in the gegl-tester.c file in the bin folder), so for now I'm
still using a simple diff operation. I think I could extend gegl-tester
to suit also our needs but I think we need to discuss this in the IRC.

* I'm working on add OpenCL to:
  - Checkerboard
  - Grid
  - Stretch-contrast
  - Remap
But I have some questions. Since Checkerboard and grid are
point_render_class and remap is point_composer3_class, I'm unsure if it
is better to hide the iterators as you did in the point_filter_class or
do it the same way we usually do in the filter_class; letting them
inside the cl_process function. What do you think?

Also, this week I'm supposed to document the work tha needs
documentation but I don't know if we really need documentation for this
ports. What do you say?

The are various tasks to finish, but I'll be working on them even after
the SoC, so don't hesitate to "let me homework" :P .

Cheers,
Carlos Z.









[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]