United Bot

[stefanhendriks]

A very , very ambitous idea but it would be great if it could work out.

1. A 'United Bot', a bot that is made by the comminity (meaning , everyone who can code and has some serious contributions) for the community
2. A bot that supports a lot of mods (since the community decides)

What needed.

1. A place where the source is available
2. A program that can get your source updated without destroying your own made changes (and vice versa)
3. A good set of rules so the 'official bot code' gets updated with proper code.

I wonder how the guys at linux do this, they have millions of coders, yet they seem to create distribution packages which are continiously updated.

Note. If needed, i would like to strip down my code to a minimum to some sort of template... Or , just use the source as it is now. Anyway, i need some input/replies on this one.

[Nova]

I can setup a CVS servber for you guys ...

[Pierre-Marie Baty]

This is without any doubt THE goal to be achieved sooner or later, however it'll need a lot of discussion before even thinking about it at the algorithmic level. Let's take this by the right end.

The goal must be clearly defined, and all of us must agree on the features.
Then the general architecture of the bot (data structures) must be agreed upon.
Then each of its features has to be discussed at some technical level.
Then a work plan has to be devised.
ONLY then it'll be time to worry about CVS and who codes what...

Anyway I believe it'll be safer for each of us to bring our own projects to completion beforehand. We will then free our mind of them, and we will have gained even more experience.

For what it's worth...

[stefanhendriks]

i would be willing to stop realbot for this (or atleast, not working that much as i used to) to code/design something for our bot.

[Onno Kreuzinger]

Originally Posted by Pierre-Marie Baty

The goal must be clearly defined, and all of us must agree on the features.
Then the general architecture of the bot (data structures) must be agreed upon.
Then each of its features has to be discussed at some technical level.
Then a work plan has to be devised.
ONLY then it'll be time to worry about CVS and who codes what...

Anyway I believe it'll be safer for each of us to bring our own projects to completion beforehand. We will then free our mind of them, and we will have gained even more experience.

For what it's worth...

i'd rather agree with PM, better have one thing finished and gain some kind of same level. then make a real project, from first hand experience i would suggest this as a commercial setup:

- one is lead-programmer (usually no specific coding, just getting all work together and fixup small errors)
- most others have fixed fields of responsibility
- one makes prototype testing, regression test, test cases and perhaps the build, pakage and user doc's stuff

and the leader (el ultimo) makes the design stuff, preferably using some software desing software (there are simple UML editors or Toghether ($$$$$)) so incase someone leaves/idles/is bussy the code maintanance is not so painfull. prefeably you have one quality management guy who documents the process and enforces the usage of a intrgrated bugtracking system.

for the open source software it looks like this:
- if it's small one make all, a simple download and some basic "get it to work" doc's, after a longer time perhaps a mailinglist is started
- if gets moderate one make some sort of core coding [starting with a working proof if concept] and volunteers start to fix bugs and add/demand features, a mailinglist is started, cvs is usually not need, the community will start boards with tips/faq/patches; a very strong community
- if it gets popular more than one mailinglist is formed, a cvs is common, a team is anounced which has csv write access, one ore all are input point for the community for patches/wishes/help, good documentation and samples are needed to reduce questions, forums will become popular and need much care, if mentioned in a pc magazine the slashdot effect might come; but fame is certain
- if it grows real big and people are more and more relying on it you will need a core team which manages more than it codes, you will have branches and probably interessting job offers

it will be easier if you make clear guidelines and strict principels. my linux tv software (name it my tv does it *g*) happens still to be only 366 kb source code, since the main invetor made clear code, a plugin interface and never integrated stuff he didn't like or needed.

Cheers MeMeD

[botmeister]

IMO a united bot, should be like a template bot, where components are plugged in, or default ones removed and replaced with alternate versions. The template idea may be best, because programmers can create their own alternate versions of the bot (diversity breeds innovation), and still retain a good standard to work with. Default components can be replaced with newer enhanced versions, allowing the bot to evolve to ever better forms. "Better" is an opinion, so anyone with coding skills could in princible replace or modify components to suit their vision of better. I think we'll have much less arguments of opinion over how the bot should perform and what it should do.

This type of project was attempted a few years ago, and there was a breif talk about it in botmans forums. It should be in the archives for reference. Someone (can't recall the details) had (or still has) a website with the details of the project.

Personally, I like the idea of us working on such a project. There's a lot of effort going into reinventing the wheel, but there's also a need to allow plenty of freedom to explore alternate methods - in a united way of course.

[stefanhendriks]

apart from the question how much this bot should support (in terms of games/mods) it would be indeed superb if you could use plugins.

So, you kinda have a code structure which loads DLL's which are specific parts of the bots brain. Like a pathfinder DLL for example.

[Pierre-Marie Baty]

Righto, that's why we need to agree on a common data structure and the APIs that come with them first.

botmeister, I wonder how interesting it would be to "integrate" this concept with the little things we were talking about by email... or the contrary.

[stefanhendriks]

yes, a working 'template' in this one would be neat. Perhaps using metamod can simplify things? hmm, nahh..

[botmeister]

Originally Posted by Pierre-Marie Baty

botmeister, I wonder how interesting it would be to "integrate" this concept with the little things we were talking about by email... or the contrary.

What we've been talking about would use the exact same concept, but would be a much larger project. I really would like to explore working on a bot first, because it is a relatively smaller project. We can gain experience with the concept, develop the needed techniques and experience, then move ahead with the larger project with a much better idea what to expect.