[Soc-2014-dev] Weekly Report #7 Viewport FX III

[Jason Wilkins]

It took a bit more time than I expected but I have a GLX implementation
now.  I actually got it to work on OSX, so no need for me to setup a Linux
box. (It didn't take much work to use x11 on osx actually)

I'll be committing it later today after cleaning it up.

This is a heads up for Antony since he indicated he wanted to look at the
GLX context class.  It actually works on OSX but there may be a surprise on
Linux so that would be a good idea.

On Sunday, July 6, 2014, Jason Wilkins <jason.a.wilkins at gmail.com> wrote:

> Beside GLX support I just need to get feedback.
>
> On Sunday, July 6, 2014, Antony Riakiotakis <kalast at gmail.com
> <javascript:_e(%7B%7D,'cvml','kalast at gmail.com');>> wrote:
>
>> Hi,
>>
>> It's not entirely clear to us what the state of the project is. We had
>> agreed before GSOC to have GHOST initialization reviewed and merged as soon
>> as possible, so that we could at least have the basis on which to port the
>> new vertex streaming code.
>>
>> As far as I know this is not achieved yet.
>>
>> I would really rather one target is finished, reviewed and merged first,
>> then move on to other targets.
>> Not doing concentrated work will surely result again in unfinished work
>> and the code in an uncertain state.
>> Let's avoid that this time.
>>
>>
>>
>> On 5 July 2014 16:36, Jason Wilkins <jason.a.wilkins at gmail.com> wrote:
>>
>>> I set a few modest goals last week and I was a little disappointed that
>>> that is about all I was able to do.  Then I realized I lost a couple of
>>> days due to having to go home early one day (a pipe burst) and of course,
>>> the July 4th holiday.
>>>
>>> This week:
>>>
>>> * Extension Shims - This is just a way of reducing the amount of
>>> conditional compilation and branches that can happen due to the different
>>> possible OpenGL APIs that may be available.
>>>
>>> * State Policy - Reducing the number of OpenGL API calls by choosing a
>>> "policy" for each state that results in the fewest calls.  For example I
>>> eliminated a few dozen calls to glBlendFunc by just assuming that the state
>>> is GL_ALPHA/GL_ONE_MINUS_SRC_ALPHA.
>>>
>>> At this point I regretted that I ignored the Game Engine last year
>>> because I'm not completely sure that all my state policies can hold across
>>> all of Blender, BlenderPlayer, and the embedded Game Engine without more
>>> testing.
>>>
>>> Next Week:
>>>
>>> One reason I was conservative about goals last week was that I wanted to
>>> think about my next move, since it wasn't completely clear to be at the
>>> time, but I think I figured it out.
>>>
>>> There is a module called "gpu_immediate" that is meant as a replacement
>>> for immediate mode.  Not a great name since it is more like a vertex buffer
>>> builder and is capable of saving vertex buffers for reuse.  My goal for
>>> this week is to get that module into a reviewable state.
>>>
>>> There is some overlap between it and gpu_buffers and some of the cases
>>> where Blender uses vertex arrays instead of immediate mode that should be
>>> combined to remove the redundancy.  Also I've been meaning to use vertex
>>> array objects for over a year now and I want to get to that.
>>>
>>>
>>> _______________________________________________
>>> Soc-2014-dev mailing list
>>> Soc-2014-dev at blender.org
>>> http://lists.blender.org/mailman/listinfo/soc-2014-dev
>>>
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2014-dev/attachments/20140709/e5c8f3f6/attachment.htm