[Bf-committers] Hello and questions

[Ton Roosendaal]

Hi,

> - strong knowledge of C/C++ ;
> - good knowledge of Python ;
> - strong knowledge of the Qt library, but weak about OpenGl ;
> - currently working on a CAD software - nothing to do with art, but an
>   industrial-oriented, closed-source software ;
> - work almost only on Win2k at the office, only on Linux at home ;
> - I also write each month an article in the french magazine  
> LinuxMagazine.

You're welcome! :)

> 1- have someone ever considered to "convert" the whole source to C++ ?
>   Stronger type checking as its advantages (and its constraints) ; this
>   could also provide many interesting tools, such as those provided by
>   the STL for basic data structures.

Better not to start the debate on that again. Blender just has a C core  
out of historic reasons, which - as key maintainer - I am still very  
happy with. Lack of structure in the C code can be solved in many ways,  
also with sticking to regular C.

In the past we've defined C Apis to be default in Blender, but inside  
of a module/library people can choose whether to use C++ or C. That's  
the responsibility of the developers. One of the ways to move on with  
Blender is take out parts of functionality in Blender and turn into  
decent modules/libraries... so also for C+++ enthusiasts a means to  
contribute.

> 2- I noticed global variables where used quite often (like the "Global"
>   struct) ; is there a special, precise reason for this ?

Well... back then I already decided global variables were ugly and  
should be banned. :)
Forcing them in a struct is a bit like a punishment (dependency forces  
all code to recompile) and a means to have them all located together,  
so everyone knows where the globals are.
Another reason was - for a future - to move parts of the struct to a  
dynamic one, allowing more instances (real windows) of Blender to be  
open at once.

> 3- I praticed a little with the Python API, and I found it lacked some  
> easy
>   mean to create primitive meshes (cube, cone, monkey...). Looking at  
> the
>   code, it seems to me it would need some restructuring to provide such
>   functions, for example passing the mesh as a parameter to the  
> addverlist()
>   function, instead of using the global editted mesh, or extract some  
> code
>   out from the add_primitiveMesh() function. Is there any plan for  
> this,
>   or does it sound useless (or event stupid) ?

There's a very active python team, gathered in the bf-python mailing  
list. I guess you can best have it discussed over there...
But yes, adding primitives would be an excellent job for python, also  
allowing it to be customized.

If you like to be involved with Blender development why not just choose  
an area out of personal interest or competence?
You can also aks us for a nice todo list of course. :) Some quick hints:

- audio in Blender is totally unmaintained
- the Booleans library needs maintainer (c++ andd heavily STL usage)
- game engine would need more developers (c++ entirely!)
- physics/collision detection maintenance, maybe check on ODE,  
integrate it in Blender tools (not only engine)
- the GHost library needs maintenance (also c++)
- in general we can use competent experienced Windows developers,  
especially on platform related issues like quicktime installs, path  
conversions (C:\\ etc). Check our tracker!
- same as above for Linux too.

-Ton-


>
> Last word, I would like to congratulate and thank a lot all Blender  
> coders
> for their incredible work.
>
> Best regards,
>
> --  
> (°< | Yves Bailly  : http://kafka-fr.net   | -°)
> //\ | Linux Dijon  : http://www.coagul.org | //\
> \_/ |                                      | \_/`
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at projects.blender.org
> http://projects.blender.org/mailman/listinfo/bf-committers
>
>
------------------------------------------------------------------------ 
--
Ton Roosendaal  Blender Foundation ton at blender.org  
http://www.blender.org