Open Source vs Quasi-Open (what we're doing now)

[F_Smith]

Hi, all:

I would just like to repeat my agreement with Stephen.

To my mind, that is largely the reason coders are coming and going. I understand if you still disagree with me, but the basic framework was there over a year ago and should have been designed and coded up already. Once that architecture is done, then functions (individual models) can be added/improved far more easily, hung on the frame in a 'component' approach. The house can only be as solid as it's foundation.

And in my experience, without an architecture, there is very little for a coder to even do . . . for fear that lots of what they work on will have to be changed once an architecture is built.

[roquijad]

Hi, all.

IMO the real problem is deciding when a model has what's needed to go into coding phase. I could just invite Stephen to work with me coding the govt, social and riots models, but we still haven't yet discussed these models, not even in a general way. The bad thing is when discussions start, they never get to a point where you can say "well, at least we have the core of what we want... the rest are details we'll need to solve in the way". If we could reach that point, we can start coding and get the most from available coders. My suggestion would be to create a simple and quick method to take that decission. IMO our team lead, Sir Mark Everson, is the right person taking the decission. I know Mark'd prefer to know everybody's opinion on a given model before taking a decission and in that case some kind of poll or short thread can be made. But someone HAS to take the decission at some point. Otherwise we'll never advance to coding phase and we'll keep losing coders.

What do you say, Mark?

Rodrigo

[Mark_Everson]

F. Smith:

If I understand what you're saying, then I think we already have a framework for much of the game. Things like the government and characters parts of it would be an exception. If you would like, I can send you a current version of the source code, and you can generate the Javadoc for it. Then perhaps you could look at the class structure and tell me if what we already have is what you are talking about or not. Alternatively, if you would be interested in working with Rodrigo and axi to specify what the architecture for the social/government/riot models is, then you would both do the project a service, and give me a direct example of what you're talking about.

Rodrigo:

We have cut off debate on several models, or at least the basic aspects of them, when they reached a sufficiently mature state. It just hasn't happened in the government model yet, although I think it will soon with respect to the framework! The reason I have not cut off debate in more areas, is that we haven't had the coders to actually do the programming. So, my reasoning has been why stop discussion and exploration of the model when nothing solid is going to be done in the near future anyway... In my opinion, we have never lost a coder because of lack of some work for them to do. However, I can't say this with certainty because a lot of them disappear and never say anything! Even when I asked them why they split. But I always give them a choice of what they would like, and state that I can push to have something sufficiently finalized in any given area they choose, so I don't Think that's the problem.

In terms of S. D., the thing that he expressed most interest in coding was the diplomacy model. So unless he has changed his mind, I think he should probably stick to that. I also think it will be much less complicated to code the diplomacy model since there are far less interconnections with the rest of the models. However, if a programmer should turn up that wants to work on the social and/or government models, I would be overjoyed. I think there is already enough meat in your proposals, that unless someone points out a huge flaw in the next few days, a programmer could quickly move from architecture into coding some of the classes (classes here means methods, not like classes in the government model).

[Richard Bruns]

I'd like to say that I support the open source concept, and I personally would have no problems with turning Clash into freeware. I don't expect any money from this; I view the project as a fun hobby. I am fairly certain that most other people feel the same way.

[F_Smith]

Actually, the design of any single model is not what I mean. I'm really talking about the design of the actual program that these models will hang on.

I did help do an OOA for the new combat model (which everyone seemed to like, altho they then went back to discussing in terms of formulae) and for the Tech model (which was soundly rejected -- remember the 'we don't do object design' discussion?). A search should turn those up easily, if you're interested in an example of what I mean. But still, I'd be happy to help do an OOA of any individual model, if anyone would be interested. It's just frustrating when all that gets done and then never used . . .

If the code is significantly different from when I saw it last, then perhaps I'm mistaken. But the design I last saw could have really benefitted from more eyeballs. Behaviors (interactions between objects) were not modelled as interfaces, I think, and some liberal use of the Mediator and Command pattern would likely have helped simplifiy much.

And again, I'd be happy to do an OOA on the whole package, and then open a thread to begin some group OOD. But you indicated to me when this last came up that you would not adopt any major structural changes, so I'm not sure why it would help to go to all that trouble.

[Mark_Everson]

F. Smith:

Yep, I remember the discussions. I would characterize the Tech model one as more "why do we need to do object design Now?"... I think your going ahead in working with the government model people on the OO design is great. Hopefully I can learn something from what you are doing about the way it should be done! It's also an ideal place for you to do it, because no coder is yet lined up for it, so there is no possibility of conflict with the coder...

On the overall code structure, it's probably still pretty much as bad as you remember it! At least all the stuff you say about interfaces is not something I understand. Before demo 4 I did go ahead and rework the higher-level structure of the existing code more or less according to your suggestions, so at least something should be better than way back when... and since demo for there has been a lot of improvement in the area of adding getters and setters, privatizing variables and other low-level things.

I am game to take a look at some possible structural changes in the object models at this point. At a minimum, I think if you can give us a succinct list of things that have previously been done incorrectly in your opinion, we can try using your approach as things go forward. I value the fact that you are a pro, and so consequently know a lot of stuff that I'm completely ignorant about in the area of OO design. I think our metric just needs to be whether re-doing it right is worth the effort. I think I'd argue that for things that are going to undergo fairly radical change, it's definitely worth going forward with a better approach. For the others, I just don't know. I guess it depends on the trade-off of how much improvement we gained from doing it right, and the time investment it will take.

[This message has been edited by Mark_Everson (edited June 19, 2000).]

[F_Smith]

Mark:

What little I know about design could fit on the head of a pin, but I'm more than willing to take a stab at organizing a Linux-style OOA hashfest. With enough eyeballs, someone out there can fix almost everything. I'll try and start one for a few of the models, and we'll see how it goes.

The basic rule to designing and coding any big program (which has many objects interacting in complex behaviors) is to list all the 'objects' in the program, then to list all the 'behaviors' of those objects. The idea is that 'behaviors' of objects (like a govt -- the behavior of people or social classes) should work like 'listeners' do in Java. Or like 'Runnable'. After doing this analysis and design, coding becomes a short, simple thing that anyone can do quickly. And the objects and interfaces can evolve, and be improved, from that point, as the model evolves. And you automatically have a prototype available to test ('release early, release often').

And then there are 'patterns' -- more about that later. That's a topic in and of itself.

I've been the beneficiary of someone else's design -- I'm much happier (and more productive) when I have been able to code from a good design.

[F_Smith]

Bump:

anyone know how we can get ahold of 'Stephen_Dedalus'?

[Mark_Everson]

S_D has the current diplomacy code, and is in principal working away at it... I don't know how far he may have gotten, but I presume not too far since I haven't heard from him in several weeks. If he were actively working I would have expected to get questions by now. If you want his email, I have it at home, and can send it to you.

[F_Smith]

Either you or I, your choice. Might be a good idea, just to point out the thread on teh coding standards/architecture. From his comments above, in this thread, he might have some feelings on the MVC architecture approach. It would simplify his job immensely.

I also came across the name 'mca' -- is he still with us? Can we get ahold of him?

There was also a 'shimmin', but I'm not sure how involved he ever got . . .

[Mark_Everson]

I'll contact S_D on this.

mca pulled out long ago, after making really good progress on a map generator using plate techtonics... Henrique said he would try to pick that up, but we haven't heard much from H in the last several months.

shimmin is working on the merchant model, if he ever gets free time again. He isn't a coder that I know of.

If you know of a place to advertise for java programmers, please do so . TK had a friend go on the mirc channels (programming I think) and got some interest, but not java coders, just people who though java was the anti-christ. That might be worth a shot by someone more knowledgable (read you ).

[alms66]

BTW, just so you both (Mark and F_Smith) know, I had this friend go to the Mirc channels because I detest Mirc, and while I trust him, as far as I know he could have lied about the whole thing (he went to Programming channel got interest, but once he said it was in Java they kicked and banned him). So, in other words, I'm sure Mirc is a good place to recruit, especially for a professianal programmer like you, F_Smith, you just need to know what you're talking about, as you do (my friend knows nothing about programming, and that may be the reason he was banned, I don't know).