Opinion: The way forward

[tomreyn]

It is, so far, my understanding that we have a very stable release now, and that it may be possible to head towards a release which can take a while but would introduce larger improvements and features which require more time for development / testing than it was possible lately.

I think there are some areas which really need dedication, and i think we should try to collect those to form some milestones.

To me, these are:

• Widgets / GUI, modal interfaces (would SDL2 make this easier?)
• Better game graphics (many techniques have been discussed here and I know very little about it so I won't get into it)
• Performance analysis (increasing our insight into how certain code affects certain systems, learn what we keep CPU and GPU busy with, why, whether it's needed or whether there are alternatives which may perform better on current hardware) resulting in performance improvements (there is no reason to keep supporting OpenGL 1.x hardware, shifting minimum requirements to 2.1 should be perfectly fine as a first step, then replace old OpenGL callsand CPU-bound routines by better performing alternatives, mostly by preventing to create bottlenecks by replacing code which is problematic on some systems by other code which is not as problematic there)

To me, these are really the three most important things we need to look into. There are more, but they aren't as important / urgent from my perspective. I also feel these are obvious and everyone else will agree on at least two of these marking the most important milestones. If meaningful improvements can be achieved in just one of those areas, this would definitely provide a good reason to name this release 4.0.

I'd love to hear what Softcoder, Titi, Will and everyone else think about these milestones, and what seems most important to you for the game itself (not just for personal interests).

Edit:

Before this is forgotten, I'll list the topics Titi brought up during the discussion as most important to him:

• a better performing / less demanding headless server
• the (yet unreleased) feature to rejoin running games

Also, Softcoder came up with criteria for choosing an OpenGL UI library/framework:

• cross-platform
• good performance
• easy to use

[Omega]

Curiosity: Why SDL instead of, say, SFML, which would have much cleaner code (C++ instead of C)?

One change I think a lot of people would like to see is making the AI more versatile. In particular, a Lua AI. This is a major change, as all the existing AI logic (which is a lot) would have to be converted into Lua code and appropriate handlers would have to be created to allow the AI to use this Lua code. I think the end result would be fantastic, however, as the AI would be extremely customizable and we'd be able to customize the AI to factions and scenarios.

[tomreyn]

Curiosity: Why SDL instead of, say, SFML, which would have much cleaner code (C++ instead of C)?

I can neither confirm or deny your statement nor answer this question. What I think I know is that most games which got recently ported to Linux (and often OS X, too) use SDL 2.
For what it's worth: most discussion on SFML vs SDL I can find on the web compares SFML 2 to SDL 1.2, predating the SDL 2.0 release on August 13. And the SDL 2.0 feature list is impressive.

One change I think a lot of people would like to see is making the AI more versatile. In particular, a Lua AI. This is a major change, as all the existing AI logic (which is a lot) would have to be converted into Lua code and appropriate handlers would have to be created to allow the AI to use this Lua code. I think the end result would be fantastic, however, as the AI would be extremely customizable and we'd be able to customize the AI to factions and scenarios.

It would also be a lot slower and at least a third of those who currently enjoy MegaGlest would no longer be able to, nor would large games with many AI controlled units be possible (this is, of course, a guess). This said, making it possible to reconfigure the AI using a simple programming language would be a nice feature (apparently Spring does it this way). Maybe that's something to discuss after performance of the current game has been improved (and if this succeeds).

[Ishmaru]

I do agree with Tomreyn on these Priorities,

Especially On Gui + Graphics as that is what attracts players to the game.

[softcoder]

As discussed on IRC I am currently investigating GUI replacement, looking at CEGUI for now.

[kratos]

I think the better graphic is a secondary problem with a game like Megaglest...

[tomreyn]

I've split the discussion on fuzzy graphics to form a separate thread because it was getting detailed and would have deviated from discussing overall milestones here.