[haiku-development] Re: OpenGL Kit maybe moving into Mesa
- From: Paul Davey <plmdvy@xxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Fri, 4 Oct 2013 17:35:16 +1300
>>
>> Are there any known programs from BeOS days that require OpenGL, and for
>> which ones do we no access to source code? And if so are those program
>> worth the effort that would be required to keep them running on a gcc2 only
>> system? What I'm saying is, is there even a good reason to support OpenGL
>> for gcc2? I mean we waited and waited for the original OpenGL kit to come
>> out from Be but was it ever released in an official release? If not then
>> maybe keep the focus on 9.3 for gcc4?
I would support not doing extra work to make gcc2 work here because
gcc2 is becoming a larger and larger burden as time goes by and we
will also eventually not be using it for the system compiler so there
are diminishing returns to having the GL kit build with gcc2.
>>
> From my in depth testing about 2 years ago, there are a handful of demo
> apps, but hardly much of anything else and that it. I think it'd be sane to
> stop supporting gcc2 in opengl at all. Whatever apps are lost so be it.
If there are no significant BeOS apps that use GL then there is little
point to having gcc2 GL work on haiku.
I can however see a possible compromise, would it be possible to write
some kind of gcc2 wrapper library that would use the gcc4 swpipe
renderer but provide a gcc2 C++ OpenGL kit interface. With such a
wrapper library the actual GL renderer could be gcc4 compiled mesa
using the up to date version and we would be left with the task of
maintaining a wrapper for gcc2 rather than making changes to an out of
date mesa implementation. Additionally if we are eventually going to
relegate gcc2 to being a compatibility layer on top of gcc4 code then
whatever work needs to be done to produce this wrapper library would
also be useful for producing such a compatibility layer.
I hope I am not the only person who is dismayed at the prospect of
having another indeterminable period of having a system compiler that
is actually capable of compiling a rapidly shrinking set of current
code. I personally hope that we can pass R1 soon if for no other
reason than to get a system compiler that isn't 10 years old.
My 2 cents
Paul Davey.
Other related posts:
