Ykkrosh

Oops, careless mistake. Should be fixed in SVN now - thanks

Following a single person lets you tell a better story, since you can make the player care about the characters (if you do it well enough) - it's much harder to care about an abstract notion like the fate of an empire - and I think a compelling storyline is an important way to make it feel like a proper campaign rather than just a sequence of skirmishes.

Transport ships don't figure into it. (Should they? Seems more like a job for player micromanagement or for higher-level AI, not for the pathfinder.) Group movement isn't handled specially yet (it'll just do fifty separate pathfindings). There ought to be a group/formation system so it computes a single long-range path and then each unit uses the short-range obstacle-avoiding pathfinder to chase its position in the group, or something like that. There's no optimisation for unreachable destinations yet. That probably would be good, although if you select an unreachable target then the unit should move to the closest reachable point so I think it'll still need to do some kind of non-trivial search to find the goal tile. Also I expect it's a bit trickier when the goal is a circle/square (e.g. when an archer moves into attack range) rather than a point, because you can't simply check reachability of a single tile and there's too many different ranges to let you precompute it all. The complexity may still be worthwhile for performance, though. There's loads of algorithmic optimisations that are possible - currently it's just a dumb brute force tiled-based A*, which is adequate for now since computers are good at brute force (it only takes milliseconds) but it really ought to be replaced with something cleverer. (Plus there's a dumb brute force visibility-graph A* for short-range obstacle avoidance, which is probably harder to optimise (everything keeps moving so you can't precompute much) but still has plenty of scope for improvements). My focus was on getting the gameplay functionality and the interfaces with the rest of the system working, so we can get on with testing the game, and hopefully someone with more time and/or pathfinding experience can improve the implementation in the future

Half the fun of GTA is the unrelenting puns in the naming of shops and radio stations and locations and pretty much everything else - it would be terribly dull if they started using real brands instead . And it wouldn't be good if developers stopped making historical or fantasy or science-fiction games because they don't provide enough opportunities to insert modern-day adverts. But it seems developers are often happy to trade creative freedom for money so things like product placement are already become more common, even if people complain about it occasionally.

SpiderMonkey (as used in Firefox etc). I committed a fix in r7550 a while ago since I encountered a similar problem when making the game launch directly into the loading screen skipping the menu - I don't think that's the same problem (so I don't know what the original problem was) but it's the same solution so hopefully it'll work okay.

Oh, okay. Glad it works now!

Hmm... Could you try running some commands like file test_dbg file /usr/X11R6/lib/libGL.so file -L /usr/X11R6/lib/libGL.so locate libGL.so|xargs file(from binaries/system/)? It sounds like the game was compiled as 64-bit but it's picking up a 32-bit libGL, but that's distro-specific and I'm not sure if that's really the case or why it's happening or how to fix it

That looks like you need wxGTK-devel and libjpeg-devel.

Sounds like you probably don't have OpenAL installed - it should be in /usr/include/AL/al.h. What OS / distro are you using?

Random animation selections are already supported . Just list multiple animations under the same name when editing the actor files and it'll pick one at random each time it loops. We don't have any way of blending multiple skeletal animations, though - that would be mostly useful for transitions between different animations but I expect it's very hard to make the blending look good.

The code should probably be considered a bit more stable now. It does automatic out-of-sync checks (which works pretty nicely with the design of the new simulation system) and complains whenever there's any difference - if anyone tests multiplayer and gets an OOS error then it'd be good to know exactly what actions the players were doing, what OSes/etc they're using, and the oos_dump.txt files that it generates on each machine. (You can also test with multiple instances on a single computer, though then it'll annoyingly overwrite its own log files...) I'm not currently aware of anything that causes OOS errors, but I expect there's bound to be a few still hiding in there

The game engine requires all music and sounds to be Ogg Vorbis. Currently we just have 'peace' and 'battle' tracks, but we don't have any dynamic music selection implemented in the engine so I'm not sure how it'll end up working. (Probably depends largely on how wide a variety of decent music we can get made, and whether we can blend between them sensibly without it sounding terrible, but I don't know any details.)

No . Modifying another game engine to support the full gameplay design of 0 A.D. would take a lot of effort and involve a lot of compromises, since they aren't designed to support all the features we want and they have a lot of hard-coded assumptions (e.g. Spring forces buildings onto a grid pattern, instead of allow free placement/orientation). It would probably still be easier than starting from scratch, but we already have most of a game engine that lets us do what we've always wanted to do and I believe it's the best way to get a decent-quality game based on our design finished and working

Well, if someone nukes a city near me then I'll probably no longer care so much about data files for computer games . But I occasionally copy really important things to servers in other countries, so I'd still have my Fallout 3 saved games.

Is the output of years of your work worth so little that you can't spend more than a few dollars on a decent backup solution for it? Hard disks will die, and if there's anything important and irreplaceable then it pays in the long term to prepare for that. (My current approach is a 1TB external hard disk (which is now ~$80) so I can just copy my entire computer onto it without having to worry about space, and if my main disk dies then I can copy everything straight back and carry on as if nothing happened. And I hope it's unlikely that the backup disk will die at the same time...)

Yes, that would be fine now - thanks

I get that too - looks like it's just a problem with the data files, so it's not my problem . (Probably Matt's, but I don't think it's urgent.)

Just a quick note about some recent changes: I've been integrating the new simulation system with the old networking code, to allow multiplayer games. The current status is that it seems to work at a basic level (e.g. I can run a server (on 64-bit Linux) and connect with a client (on 32-bit Windows) and move units around on one and see them move on the other), but it needs out-of-sync checks and it needs to stay in sync more (e.g. projectile attacks break because they currently use non-synchronised random numbers) and it needs to not leak memory and it needs a lot of testing and it could benefit from a lot of optimisation. So if anyone wants to test it, it might be best to wait a few days until it's a bit more stable but it ought to be runnable now. Also I deleted about twenty thousand lines of old code (mostly networking and simulation) in an attempt to clean up the codebase a bit. Should improve compile times a bit, too. But I haven't it thoroughly - hopefully it shouldn't have broken anything, but let me know if it's introduced any new bugs

Is this the 0.0.0+r07539~pre-alpha-1~getdeb1 package from PlayDeb? I don't understand why you get this problem when it appears to work fine for me and (as far as I'm aware) for most other people . Implicit global variables are yucky but it shouldn't fail randomly - functions_page_pregame.js sets curr_music which I think should make it available to other GUI scripts. Do you have other problems with music/sounds not playing, which might make this work differently? (I could apply this fix to the code in SVN but I'd prefer to understand the problem first.)

If you look for files like binaries/data/mods/public/simulation/templates/template_unit_support_female_citizen.xml, there's a <Sound> section which associates sound group files (in binaries/data/mods/public/audio/) with various actions for each unit type. Currently there's only very basic audio integration so you can't easily e.g. make a unit say something when you order it around - you'd have to edit the gameplay scripts for that (binaries/data/mods/public/simulation/components/, call a function like 'PlaySound("death", this.entity);' to play the <death> sound).

Yes. (Currently there's basically no unit AI, just a badly-designed state machine for basic testing - it'll need to be rewritten to handle order queues, combat stances, etc. Preferably it would be rewritten by someone with AI experience who can design it better than my current attempt )

Yes.

Hmm, I can't find a way to get that error myself. I've committed r7546 which fixes a couple of bugs that can crash if you're unlucky and the JS garbage collection runs at the wrong time, but I don't know if that's the same as this problem. If it still fails, can you run in gdb to see where it's failing? ("gdb ./pyrogenesis_dbg", "r", run until it hits an error and freezes, "bt full")

What series of actions did you take before that error occurred? Can you reproduce it by doing the same actions?

So it works now on 10.6 straight from SVN with no local modifications? This is the first time we've ever had it running there, so thanks for helping debug it