Re: [TuxKart-devel] Re: [Tuxkart-cvs] tuxkart/src Driver.h,1.14,1.15 joystick.h,1.3,1.4 PlayerDriver.cxx,1.12,1.13 sdldrv.cxx,1.16,1.17 sdldrv.h,1.6,1.7

[James G. <jam...@bt...>]

On Sun, 2004-08-08 at 00:08, Steve Baker wrote:

> The whole full screen business is a really trivial matter - if the PLIB
> windowing library can't do it then let's make it do it rather than tossing
> it out and starting over.  However, this isn't something that in any way
> stops people from doing what GoTM was supposed to be doing - it just
> causes all of the ridiculous SDL hassles - which is now infecting the
> joystick interface and who-knows-what else.

Saying that fullscreen is "trivial" might be relatively true from a
technical point of view, but from a player's point of view it's totally
vital. The idea of GOTM is to do whatever is necessary to make a game
more polished and playable so that at the end of the period you can say
"download this and look at it as a game, not a tech demo of what might
be". With TuxKart, lack of fullscreen was a major barrier to this.

> The whole problem with GOTM (from the maintainers perspective) is that you
> set out to do a relatively modest amount of work just where it was most
> sorely needed - something that could have been managed in two months if
> everyone had stayed focussed on the original objectives.  This was all agreed
> before the voting and with my 100% backing.
> 
> What's actually happened is that you've ended up attempting to do virtually
> a top to bottom rewrite with all new models, new libraries, new GUI - the whole
> works.  You are dinking unnecessarily with things that are really working just
> fine.
>
> Frankly, it would have been easier for you to start afresh with a completely
> new game...but that would be ridiculous because you KNOW you can't write a
> complete new game in two months.

"Top to bottom rewrite" is rather an exaggeration. Except tweaking the
camera somewhat, none of the OpenGL/PLIB scene graph stuff has been
touched, and that's by far the largest part of the game. I've seen
people spend months writing simple 3D engines. I don't know whether
you're ignoring this fact through modesty or a wish to make your
argument seem less weak or what.

Switching to SDL only took one evening, and much of that was spent just
getting to know the TuxKart code, which I'd not previously seen.
Switching the joystick code to SDL took like maybe an hour or something.

Adding the new GUI took quite a bit longer than switching to SDL,
because the code was taken straight from another game rather than a
documented library designed to be used in other places. However, TuxKart
/really/ needed a new GUI. Maybe if you're used to making flight
simulators a pretty GUI doesn't seem very important, but TuxKart is
supposed to be an arcade game, not a flight simulator. Can you imagine a
Nintendo game with a GUI like something out of an office application? It
totally ruins the experience.

The same polishing concept applies to models - it doesn't matter how
technically great a game is if the models don't look very impressive.

You say that we are "dinking unnecessarily with things that are really
working just fine", but that's the entire point. TuxKart already worked
just fine, you could download it and play it and it worked. But games
are about things not only working, but working nicely - looking pretty,
feeling right, etc. We're making it work a bit finer.

> The GoTM concept is a good one - but failing to stay focussed on what was
> the original scope of the work is what undoubtedly lead to your failure
> to complete Pingus - and with only a few weeks to run, the same will
> happen with TuxKart.

The original scope of the work was to take a game which was already
functional and polish it - making lots of smaller changes with players
in mind rather than showing off mad coding skillz by trying to add new
features. By making changes to the models, GUI etc, that's exactly what
we're doing. We haven't touched the underlying engine, or tried to add
network play, or anything else as major as that.

Also, as far as I can tell Pingus already is largely complete code-wise,
it "just" needs more levels and artwork.

	James