[Freeciv-Dev] Re: freeciv 2.0 spec

[Stepan Roh <stepan@xxxxxxxx>]

On Mon, 3 Dec 2001, Andrew Sutton wrote:

> never mind... apparently my ideas about development processes are just plain
> crazy. its odd. some people have said we need this, and then others have
> turned around and said that its bad.
>
> maybe somebody can explain the prejudice against process - however light -
> and validate their claims.
>
> if there's actually historic evidence, can somebody please point me to an
> open source project that has failed utterly because the developers wanted to
> take the time to think about the *next* version and write their ideas down?
> there seems to be some concensus that having a requirement/design phase is so
> bad that anything written that way isn't good (enough? raimar?)

One example of much-design-early-die is Freedows95 or how it was named.
They designed and designed and designed and finally they split to smaller
groups which split again to smaller groups and whoever knows (well, I
don't) where they ended.

I'm not against design before code, but Freeciv is strictly open source
project. It is not funded like Gnome, it has no FSF and RMS behind its
back like GCC. No organization pays Freeciv developers to work on Freeciv.
Everything is done by volounteers. There is no small group of developers
who are assigned to design Freeciv. Freeciv can't have a strict
company-like design or development process, because it will take very
long. You seem to be the one behind design. That's OK. But if you will
quit for some reason? There are no pushes from outside. No deadlines, no
budgets, no customers, well, developers heaven :-) Even if project leader
quits from commercial company, project goes to trash in 80% of cases
(guessing). No leader, no project. Same applies to Freeciv.

You are proposing to make a framework or platform (I like these terms,
they're so vague I can name with them almost everything - useful at work)
for general (read 'customizable') turn-based game with clever AI. Well,
that's a big amount of work. And I don't think this goal can be achieved.
It's too much complicated and I can see lot of problems in the future. In
fact you're trying to do a simulation software. Agents living in artifical
world act according to rules. Well, that was stupid sentence, but I can't
help myself. First you want to build microkernel. Than you want to program
components implementing rules which interact with it and finally you want
to create AI which can handle all possible cases. AI is important, because
it (besides multiplayer) makes Freeciv a good game (or it should; playing
with current AI is boring - it is like playing chess with Stalin).

There is also a little connection to current Freeciv besides planned
default rulsets and maybe network protocol and client reuses.

Freeciv is a game, it's not research project ((c) Trent). For the record -
by research I don't mean educational or something like that.

Maybe I wanted to say something more (or different or maybe clever, who
knows), but I'm very tired. I'm shutting up :-) Good night.

> this isn't sarcasm. i'd really like to know.
>
> and for the record... no i'm not in college. i'm an actual software engineer
> and i design software for a living (a slave to the process :) and my rants
> aren't coming from idealistic innocence fresh from a college course, but hard
> learned lessons in practical development. i've written code both ways and i
> have to say that the best software i've ever written has gone a much longer
> development process than what i've proposed. so please, enlighten me.

Have a nice day.

Stepan Roh