On Mon, 15 Oct 2001, Rodolphe Ortalo wrote:
>  Specifically, the complexity of the design and the fact that it prevents
> people from trying to help on a specific area annoys me (especially as,
> sometimes, I share this feeling also). But I don't really know how this
> could be improved (more comments in the code, more README.internal
> files?).

I am always willing to answer questions on IRC or by e-mail on the
docs for the new libraries, especially LibGAlloc, as it is the 
stumbling block (see below).  Questions help me figure out what's
not clear in the docs, and improve them.

So folks, please do read the docs and ask lots of questions.

> [2] E.g.: texture allocation deserves to be done cleanly

Here you hit the nail on the head.  All of the GGI Project's problems
with staying in the race vs. other APIs has been with the inability to
allocate and use various types of storage i.e. DMA/AGP and VRAM that
for one reason or another cannot be gotten by simply using virtual
screen area.  (Look at the GGI-native fbdev renderers.  Notice the
lack of a accelerated Put* primitives.)

Basically, we tripped over that stumbling block, got up, dusted
ourselves off, and atypically "Did the Right Thing" by walking back to
the shed for an axe to chop it into kindling rather than just
sidestepping it like everyone else does.  Hopefully we can finish the 
chopping job soon.

--
Brian

Reply via email to