Re: [glob2-devel] Creature AI

[MOA3333]

I dont know why the first email i sent using postfix was not recived,
but anyway, here it is again, and i am using the web interface now.


Andrew Sayers wrote:

> To be honest, one of the things that attracted me to this project was the
> thought of being able to contribute without wading through lots of
> undocumented code.
>

 The game was writen mostly by nct (Staphane Magnenat, most of the game)
 and also Nuage (the algorithmes and many other).  It is an old project,
 but only in the past 12 months some other people from the comunity got
 more involved, since it was put on savannah i guess.
 Many sugestions have been made (you can see the wiki), but we concluded
 there are only a few important:
  1)  Documenting the code so that the comunity gets involved.
  2)  Making a solid-and-stable, "state of the art" 1.0 release so that
  whatever hapends, the project survive, and the comunity can enhance it
  more.
  3)  Improving the CPU usage so that the we can play it on a Pentium2
  or old Pentium3. And also to play it on big 512/512 maps. (Adding new
  features,etc...)
  4)  Improving the web site, screnshots, publicity, debian inclusion,
  anouncements, etc... (but also better graphics an images)

  It was soon obvious creating a comunity that can help is not as easy
  as it seemed first. (4) was easy to do without help, (and you can
  apreciate the new web site :) ). But since nct and nuage know the code
  (they did it from the begining), but they dont have time to work on
  (1) and (2) and (3).

  Untill now all the efforts were on (2).  You cannot imagine how many
  bugs were found in the past 6 months that noone else could fine even
  working very hard.

  There is work done to identify what takes time in (3), and the code is
  now as fast as it can with the current implementation. Also some
  "work" was done to identify the feature requests for after 1.0 on the
  wiki/forum but no implementation yet.

  For (1), until now no people were realy so intrested as you (i think).
  If you want to help commenting the code, there is a place on the wiki
  where everythink should converge and Where you can scratch ideas :
http://epfl.ysagoon.com/wiki//index.php?n=CollaborationPlatform.TheCodeBazaar
Also some files are on the CVS, you only nead to link to then. Of couse,
somments on the code are also usefull but even more difficult to make.


> On the other hand, if you're willing to put the time into explaining how
> things work in simple terms, I'd be happy to try and do some
> documenting. 

 Since (2) and (4) are going on,  helping with (1) should be of great
 intrest to all of us.

> Admittedly, this significantly increases the chances I'll
> get burnt out before doing anything interesting, but IMHO Glob2 would
> attract a lot of developers if the code were easier to dive into.
>

I agree, working with (3) or with  new features or with better AI is
much intresting, but since (3) is optimal on current implementation,  it
is recomanded to finish (1),(2) and (4) before merging any progress in
(3) that can be done at the same time on a secondary branch for release
1.2 or 1.4

If you work on (3) and new features on some other branch, and help with
(1) at the same time would be optimal.

PS: If you have any questions about the web site or about what takes so
much CPU time, you can ask on the forum.

Mostly CPU is used with:
- Making hash-codes so the network game stays in sync (~40% CPU more or
less)
- Calculating the path the units should folow when they move from one
point to the other. (~50%CPU or more)
- Other are not important with OpenGL.