Re: [boost] [Review] GGL review starts today, November 5th

Subject: Re: [boost] [Review] GGL review starts today, November 5th
From: Bruno Lalande (bruno.lalande_at_[hidden])
Date: 2009-11-13 04:38:04

Hi Pierre,

Thanks a lot for your feedback.

About remark A:

I'm also more interested in real time 3D and if I had had more time to
spend, the algorithms you're talking about would be there already. I
mostly participated in the design, not algorithms. Barend worked a lot
on algorithms so he brought his part, that is GIS. This doesn't
prevent from bringing other application domains. We constantly made
sure that the design allows to add that easily.

Also, note that we mentioned a project structure based on extensions.
The kernel is "agnostic" regarding any precise application domain,
while extensions allow to propose more specific stuff. Although the
limit between both is sometimes hard to define, this normally avoids
having many people wanting to see in the kernel the algorithms *they*
are interested in. So typically, some of the things you are asking for
could be included in the kernel (lines, rays...) and other ones would
rather come as specific extensions (frustums, meshes...). Hope this
makes sense to you?

About remark B:

Good point about the fact that documentation is more user-oriented
than extender-oriented, this is something we should try to improve.
Regarding the user part, I sincerely think that what we have is
sufficient (I mean, adapting legacy geometries and using algorithms on
them). Regarding the extender part, it will indeed require us to give
better insight about the actual design.

However, I would like to be clear about this sentence: "can it be
extented by someone who do not use more templates than the stl". The
answer is no. The library has been designed to be easily *used* by
someone who basically doesn't know template. But to extend it (e.g.
add mesh-related concepts and implement algorithms on them), knowledge
is obviously required. You'll have to specialize some traits, play
with tags, etc... the point is rather to ensure that the design gives
sufficiently precise directions and instructions to keep extensions
consistent.

About remark C: Indeed, several persons have made the same remark already :-)

Regards
Bruno