Yeah- this is the sort of thing that I was very prone to doing in Diablo 2 (not so much in Dwarf Fortress, but people do resort to it.)  I wanted the ability to experiment with various approaches to be built into the game, without permitting victory-through-repetition.  I'm glad you like the idea in any case.

(The other question would be how you'd make this work in MP- particularly deathmatch multiplayer.  I'm a long way from actually implementing that, of course...)


Oh, in other news- I know I said I was taking some time off, but I've re-implementing mining and added terrain generation systems for rock outcrops and mineral deposits.  You can see some preliminary results here and here.  (And, of course, up on github.)

Sure, by all means.  I'm probably getting a little ahead of myself in any case- I'll want to have the more mundane economic, combat and diplomacy systems up and running first before I work on anything fancier, and that's a ways off.

Looking well.  Also fireballs!  Fireballs are good.

I appreciate the warning, though I've generally found background auto-saving to be too punishing and arbitrary save/load options too forgiving, so I thought this might be interesting as a middle ground.  Are you aware of other citybuilder/strategy titles I could look at that attempted this?  Not trying to second-guess, or anything, just curious.

EDIT:  Okay, how's about this for a tweaked alternative:  by default, the game uses the automatic-save-on-quit-persistent-world approach.  However, the player can spend psi points/mana/spice capsules/whatever to create an explicit 'save point', and then gets a minute or two of subsequent play before their 'precognitive vision' fails, and they have to either save their progress (walk the path) or revert to the last save (deny the vision.)

It would work in basically the same way, except that the player controls when the system kicks in.  Does that sound any better?

Actually in the meantime, I might explain myself a little further.  One of the things I'd like to do with indistinguishable-from-magic psyonic mojo would be to take a leaf from, say, the Prince of Persia reboot, and make functions like saving/loading the game a kind of rationed resource.

One of the major concepts behind the Dune and Foundation series (which are two of my biggest templates for the setting,) is the idea of using hypercognitive forecasting of the future to enhance decision-making.*  The application here would be that saving/loading the game is actually the player-character exploring alternate futures, and selecting the course of action which produces the results they desire.  The drawback is that such forecasting eats up mental stamina, so that every time you quit-and-reload, you lose a chunk of mana based on how far back you retrace your steps.

Actually saving (and/or quitting) the game, on the other hand, gives you back a chunk of mana (or whatever) as the player-character pauses to meditate.  Mana also regenerates slowly over time, and can be spent on a bunch of other FX, like time dilation (game speed settings/bullet-time), issuing psychic suggestions (unit micro) and other more traditional forms of direct intervention- healing, damage, recon, transport, et cetera.

I'd like to have this kind of mechanism because (A) I notice I'm personally terribly prone to abusing stuff like game speed settings and save/load features, and (B) because the game will probably have a pretty slow tempo otherwise.  But the idea is to complement automated behaviours rather than overshadow them, and the mana economy should ration the amount of micro the player can really afford.

I'm not sure how much, if any of that I'll get around to implementing, and it's likely that some of the Schools would provide special abilities of their own.  But, while I'm fantasising idly and borrowing liberally from genre precedent...

*  In the literature, there's something of a question mark as to whether this involves jedi-style magical clairvoyance or simply logical extrapolation from past trends like a chess player seeing ahead 12 moves.  ...Don't ask me about shooting lightning from your fingers.

I think I'll have to take a break from working on this for the next week or two, since other commitments have come up.

Interesting.  One thing that occurs to me is that you could have root attributes increase along with practice in the skills they boost, and/or possibly allow PCs to train eachother in particular skills when their players are offline?

One complaint I've heard about practice-based progression is that PCs can eventually become omnicompetent in all areas.  Is that part of why you want to have a class system?

@ags1

Glad you enjoyed it.  I actually find the art & background easier/more fun to work on, which might be the main reason I presented it.  Which isn't to say the technical aspects were trivial, but in the final analysis they're sort of invisible to the player.

Don't hold me on the 'real science fiction' thing, though.  I might be adding psychic powers yet...


@ Eli Deventhal

It all sounds very nice in theory, but if I wind up maxing out the processor with just one settlement at normal speed (which might happen yet,) I don't think simulating multiple settlements at greatly sped-up rates is going to be viable, leave alone the question of exactly how I write governor AI in the first place.  I could approximate what happens based on broad trends, but then I need to be able to translate from that approximation down into the fine details, and again, I don't really know how to do that.  Is it possible?  Maybe.  But it's not my priority for the moment.  Likewise, I could pretend that all my walkers are representative of bunches of people, but then I can't simultaneously assign them individual personalities or relationships.

Now, I do have some ideas for having a limited degree of interface between your local settlement and the 'big picture' view of the larger setting, but the player's ability to affect interstellar politics is probably going to be pretty restricted.

That said, I do appreciate the suggestions, and will try to get around to incorporating as much of them as I can.

Looking over that reddit thread, the algorithm sounds quite interesting, and I'd like to see some efficient implementations of delaunay triangulation/minimal spanning trees (seeing I've never done it myself.  Hell, I hadn't even heard of the former before.)

I guess I'd just say that guaranteeing and identifying loops in the room network would be especially useful, if you want to allow multiple approaches for dealing with a given mob of monsters.  Looks cool in any case.

Hmm.  I actually reduced the colour saturation by about 30% compared with earlier drafts because the colours were too intense on my laptop screen.  I'll look into it.  Shading might help too, now I think about it.

Offworld immigration and supplies should now be working, along with extra work on maglines and shield walls.  I've disabled mining facilities for the moment while I work out the mechanics in more detail.

Looks like strong progress.  I like the top-down isometric views in particular, with the sense of ambience coming from the lighting.

Well, shoot.  ...I'm sold.

Those look gorgeous.  (I'd eventually like to implement a day/night cycle, so it'd also be great to vary the intensity of shading and make interior lights stand out.)  I might PM you about this later.

In other news, some final tweaks to terrain generation, and basic versions of the mag line and shield wall are up on github.  (If you're looking to run it, TestGame has been renamed to WorldDebug, so watch that.)

Well, you're talking about gaining XP, and the player engaging in various activities like mining, woodcutting, agriculture, combat, etc. etc. etc.  So, I was wondering if the character had associated stats or attributes for activities like these?

(Also, reference.)

...we'll go to Very Distant Lands?

What's the range of skills you'd imagine having in the game?

If it's going to be purely text-based, perhaps you could put it in a browser interface, make it easier to get feedback on?

Also, grues.

Looks like I'll have to check this out at some point.  The art is super-adorbz.

If you're letting players allocate skill points directly rather than having a class system, are you planning to go bethesda and have skills advance through practice?

Ehh, I can't really take credit for that.  Adding shadows (including for the background) is something I'd like to do, it just complicated the production and rendering, and I happened on the vegetation outlines/palette largely by accident.  If I ever scrape together the time/money, I'll have 'em rendered properly in 3d, highlights and all.  But I'm glad it worked out. 
Thanks for that.  The package layout was something I changed a few months back, and I think it actually made things much easier to organise.  I might try slapping together a basic tutorial level (and I'll see if something's hindering the menu interface.)

Digging the voxel art by the way.

Sorry, one more point that just occurred to me: If you possibly can, grab a copy of Myth 2- it's a real-time-tactical game that excelled in getting the player to use differences in height and terrain to defeat enemies.  (e.g, setting fire to patches of grass, laying fields of explosives, setting fire to grass which then ignites fields of explosives- all with gorgeous physics-simulation of the attendant carnage and mayhem.)

I really want to start a project based on relatively heavy-duty AI like that, since it opens the possibility of some truly dynamic NPC behaviour and dialogue, but I'm afraid of getting badly distracted. *sigh*
Well, I was talking more about the situation you'd have in, say, Final Fantasy, or most of the Mass Effect trilogy- where you have freedom, and you have plot, but the two are at complete cross-purposes.

Without getting into a discussion about theme or drama or whatever, what I'd honestly like to see is where you have some long-term objective (get off the planet would be fine,) but the implementation of that goal is as free-form as possible.

I only played Space Crusaders once, but me and my siblings went through every mission in the MB Hero Quest game.  (Ah, fond memories.  ...If I selectively black out all the bickering over who got the chainmail.)  I did enjoy the sense of exploration and gritty detail in the line-of-sight/fog-of-war mechanics.  (I can't say I''ve played the other Roguelikes you mentioned, but I did play the hell out of blades of exile, which I now realise owed a great deal to the Ultima series.)

I 'd suggest that if you're making it single player, there may not be a particular need for niche protection based on a class system.  (And also that single-player doesn't mean single-character.)

But anyway- I'll leave you to focus on lighting/mapping, and if you like, we could discuss skills/spells/classes afterwards?

Much appreciated.  Was it able to run alright on your system?
I won't make any particular claim to outstanding originality in terms of inspiration for art or mechanics here, but that particular game I never played.  I'm not sure it's a very close resemblance, though.

I have noticed I have this odd habit of making every variable I can 'final'.  Hmm.

What I actually found most intriguing about clojure was the built-in default assumption that variables are immutable.  That's handy for multithreading, but in theory it would also adapt well to incrementally simulating various state-based scenarios, which is a large part of AI programming.  (GOLOG is explicitly designed around that idea, but I think that's pascal-based...)
*shrugs*  Works for me.  I just get a little tired of the situation you find in a lot of games, where you can go wherever you want and do whatever you please... as long as it's irrelevant.

I appreciate laying out the technical details like that (on a similar note, props to HGD.)  I've been trying to think of ways to efficiently allocate space to AI-generated structures in my own game, so this kind of breakdown on the algorithms involved is very valuable.

I could comment more extensively on the design details, but aside from 'Roguelike', I guess I'd ask what's your model or inspiration here?  Thief?  Diablo?  Tomb Raider?  Orcs Must Die?  Would this be a single-PC affair, or would companions/hirelings be involved?  Is this gritty survival horror, a hack'n'slash power-trip, or something between?

I would tentatively suggest that, if you want player tactics to be primarily about exploiting terrain differences, then it might be best to ensure that the dungeon layout allows enough branches that differing approaches to given enemies are viable.  (Relatively safe methods for advance recon would also be needed.)  Finally, the ability to modify or circumvent terrain- stockades, traps, stealth, teleports, etc.- would probably be interesting, as would aggro or crowd-control powers or spells to herd, draw or repel enemies around the dungeon, and perhaps regenerating resources like herbs or fountains could be another element to consider.

I should just mention that obliging the player to explore every nook and cranny of the level for XP was one of the principle complaints made against, e.g, velvet assassin, so be careful there.

Not much to report in terms of gameplay function, but I've gotten large-scale pathfinding and path-caching to work, which should substantially increase efficiency on larger maps with lots of citizens.  (This also means that fine-tuning terrain-gen is less critical, since the pathfinding can handle island effects, but I might revisit it later anyway.  I might try integrating it better with road-generation, actually.)

Next up:  mag lines and shield walls.  ...and maybe offworld freighters for supplies and immigrants.

I quite like the look of the lighting FX.  I remember doing some procedurallly-generated-architecture a few years back, but never entire dungeons, so I'd be interested in how you're going about it.

What kind of PCs, spell mechanics, skill systems, etc. are you planning to implement?

Javascript arguably has a number of advantages over Java, by having first-order functions and lack of static typing.  (Strict typing is a solution for a problem that largely disappears if you do thorough unit-testing... and if you don't, you have larger problems to worry about.)  I've been meaning to learn Clojure one of these days.

Did you have any specific ideas on the story framework?  Open world + story driven can certainly be done, but genuinely integrating the two is relatively rare.

I have about half of the changes to terrain generation and road-networks implemented.  I'd also like to use the road network to determine distribution of power, water and life support when the time comes.  (I've also reverted to an older flora set for the time being, since the newer versions were more colourful/diverse but didn't really 'sit' well with the terrain.)

At any rate, now that the basics are in place, I'm also open to suggestions as to what features people would like to see implemented first-

  Housing and economic supply chains?  Detailed construction and salvage?
  Fog of war, exploration and combat?  Different creature species and lairs?
  Citizen life cycles and dialogue?  Foreign trade, espionage and diplomacy?
  Tutorial/tooltips or better UI?  A video demo?  Executable .jar/.exe/.dmg?

All comments/suggestions are welcome.

OMG!  It has a panda!

On the topic of funding for art assets- you might want to post a more finished version to some relatively topical blog?  It might be your best chance for getting noticed.