Dwarf Fortress: Very Ambitious Roguelike

[Hanacker]

I haven't played DF in ages, but yeah, I agree that I thought the idea was to play the game at a macro level, and then have the game simulate all the micro level and allow you to zoom in and see how many toes your head marksdwarf was missing if you wanted.

[Quaro]

After reading that NYT article I was, for some odd reason, having daydreams of volunteering myself to build a better UI for DF. I live two hours away from Silverdale, so meeting in person would actually be conceivable, and I think I have the coding chops to do something useful there.

But seeing some of the more recent posts in this thread -- in which even Calistas, one of the guys who's done the most to promote DF via tutorials, has given up on the more recent builds -- clarifies that it's kinda pointless.

If Toady is consistently adding more buggy features without fixing old bugs ever, then any attempt to work with him would be doomed; it'd be like trying to remodel a sandcastle while someone keeps pouring more and more sand on top. The key to long-term large-scale software development is to work clean -- you have to have something solid, or over time entropy will kick your ass. Sounds like Toady doesn't even have a test suite per se, nor any kind of effective bug tracking system ("effective" in the sense of "leading to fewer bugs over time"). And yeah, it also sounds like he really doesn't want anyone entering into his cozy and familiar interpersonal dynamic with his brother.

Not a whole lot of room for fruitful collaboration there....

Well, it'd be good story however it turned out.

Toady needs to decide min/max his development skills. There's no free respec on a software project -- the longer it goes on, the more expensive it is. The way he's playing now, his build is gonna hit a wall in the end game and the rest of us can see it coming. Then he'll be stuck grinding for incremental gains.

AFAIK, he's not using version control. He's mentioned that he'd be embarrassed to show to code to anyone before, and I can understand that feeling, as I've been there. Maybe he could do a small side project instead, and work with some people on that. As a break and as a learning tool, then later he can bring he new skills to DF. Just adding version control and some amount of automated testing would improve his quality of life, every day, by reducing the stress of fixing bugs. Even just version control.

[RepoMan]

"improved mythologies" is awesome, even though I'm sure you meant "improved methodologies" :-)

And if he's not using source control... well...



Still, it's not like Toady's alone in being a mathematician/scientist who drifted into programming with no opportunity to learn software-engineering-specific skills. That, plus the fact that he's the only developer, makes it at least barely feasible for him to work with no actual process or tool support.

But the big problem for him is that it seems his player base is suffering significant attrition (as witnessed by this thread), due to his long-running and unfixed quality issues. Unless he switches his focus from features to quality, that trend will very possibly continue, which threatens his long-term plans of making a living indefinitely from DF.

[Serial Kicked]

Where do you see the attrition? The official board is still extremely active as far as i can tell. I don't think that judging from the activity on an external forum is the best method. And according to the donations it's still going well for him. People come and go, yes, but those people are not necessarily those who are paying him regularly. He has a relatively large base of loyal core users who don't find anything wrong with his schedule.

I'd say that he is pretty safe until he hit the inevitable "oh shit, I have to rewrite at least 50% of my code to implement feature X or Y" moment.

The sweet spot was 40d19 (2) for me, now my interest is decreasing, despite some good ideas there and there.

[Marcin]

I'd say that he is pretty safe until he hit the inevitable "oh shit, I have to rewrite at least 50% of my code to implement feature X or Y" moment.

This is my opinion too. He'll be fine until the codebase implodes under its own weight of unfixes.

[fenrir]

All that stuff about quality is probably true. But at least for me personally the real problem with DF is that he doesn't really want to make a game - he wants to make a world simulator. While the two have some overlap and I've enjoyed DF in the past, it's becoming very apparent to me that eventually they end up in very different places. A lot of the design decisions that sound reasonable for a world simulator make absolutely no sense in a game. I guess I could understand it more if all the depth was under the surface which might make for a richer and more real world, but I'm expected to bandage my dwarf squads one toe at a time as well as make sure the supply of left socks matches the supply of right socks.

(Standard disclaimer about how the game is free and I have no right to be bitching about it, etc. etc.)

[gameoverman]

Most of everyone's ideas regarding improving this game revolve around making it more playable now.

I'm willing to bet that he views this, his labor of love, as a life long project. So something like tossing out the entire program and starting from square one probably will happen and he is probably okay with that. How important can squashing a substantial list of bugs in this game be in 2011 if you're a guy planning to still be working on this game in 2040?

I'd guess from a pro software developer point of view that sounds at best like doing things the hard way, and at worst crazy, but I don't think this game is about making him a professional software/game developer. I've read interviews with people like Garry from Gmod and the Minecraft guy and they NEVER sound like this DF guy when it comes to their approach to their game development. Those other guys always sound like making a living developing games is the goal.

[Disconnected]

I guess I could understand it more if all the depth was under the surface which might make for a richer and more real world, but I'm expected to bandage my dwarf squads one toe at a time as well as make sure the supply of left socks matches the supply of right socks.

The problem you're having, at least as you describe it, isn't with the scope of the mechanics. Rather, your problem is the interface doesn't provide you adequate control.

If it makes you feel any better, I very much doubt that anyone who has actually played the game, disagrees with you. Even the creators. Pretty much every interface update since... Before Boatmurdered, I think, has included (also totally inadequate, but...) various attempts at providing better, faster, simpler and less anal control.

[mostlytigerproof]

I don't see how Tarn could be as naive a coder as y'all seem to believe.

The fact that DF has gotten this large without collapsing under its own gravitational pull indicates that it's fairly modular, and that he has testbeds for individual aspects of the simulation. The fact that the patch notes are mostly hilarious system interactions instead of 'fixed memory trample and crash to desktop' x 50 indicates that he has a reasonable amount of coding experience.

There's ample explanation for the bugginess. It's one of the most complex game simulations ever conceived*, and it's specifically designed to produce emergent behaviour. There is no upper limit on the game duration, you could cycle a world between adventure and fortress indefinitely.

DF is QA-proof. 100 Zachs would not be enough.

* Out of curiosity, can anyone name a game with more complex internal logic?

[mostlytigerproof]

What's more, I also suspect that DF may have become UI-proof.

Back in the early 2D days, I could easily imagine what a workable UI would look like. But now he's added the Z-axis? Now that half of the tables in the UI are growing foreign keys? I don't know what I'd do with that even if I had all the resources in the world. I fear it'd look like a cross between MS Access and Maya.

[Teiman]

The fact that DF has gotten this large without collapsing under its own gravitational pull indicates that it's fairly modular, and that he has testbeds for individual aspects of the simulation. The fact that the patch notes are mostly hilarious system interactions instead of 'fixed memory trample and crash to desktop' x 50 indicates that he has a reasonable amount of coding experience.

http://df.magmawiki.com/index.php/DF...m_requirements
CPU: AMD 7750x2 BE @ 2.7GHz
MBO: Gigabyte GA-MA78G-DS3h
RAM: 4GB DDR2 800
GPU: XFX GTX260

Number of dwarves: 140
Average fps: 6-14


So you get 10 FPS for 140 dwarvens. In a machine that problably have more than 5 TeraFLOPS ( no idea really how much ).


Warning, totally random article follows
http://www.gaslampgames.com/blog/201...t-a-bad-thing/

[Serial Kicked]

I don't see how Tarn could be as naive a coder as y'all seem to believe.

Because you probably haven't spent years debugging others' messy code. After a while you don't even need to see the code itself to know it's a mess.

The fact that DF has gotten this large without collapsing under its own gravitational pull indicates that it's fairly modular, and that he has testbeds for individual aspects of the simulation.

The fact that he was unable to (multi) thread his application to at least somewhat take advantage of multi core CPU tells me the opposite. Well, yeah the "display engine" is in another thread, but that's 5% of the CPU and it's not coded by himself iirc.

The fact that the patch notes are mostly hilarious system interactions instead of 'fixed memory trample and crash to desktop' x 50 indicates that he has a reasonable amount of coding experience.

Wrong priorities. So low level issues, like the fact that the more you go, the more the game gets sluggish, and even that killing all dwarves, objects, and closing pathfinding heavy regions doesn't really fix the slowdown, are ignored. I'd love to see a 10 page long memory, behavior, crash fix pass instead of your face-palming changelog once in a while.

There's ample explanation for the bugginess. It's one of the most complex game simulations ever conceived*, and it's specifically designed to produce emergent behaviour. There is no upper limit on the game duration, you could cycle a world between adventure and fortress indefinitely.

No one said the game isn't complicated, the point was because this is such a complicated game he should really get its priorities straight before hitting a brick wall. And, also, UI = crap.

DF is QA-proof. 100 Zachs would not be enough.

Nothing is with proper priorities and timing from day one.

* Out of curiosity, can anyone name a game with more complex internal logic?

Not really if you take the game as a whole.

A note, though, all the reactions between elements to make "adamantine axes encrusted with jewels menacing with spikes of bones" are (or at least should) be database driven. It's tedious, sure, but complex, not so much granted you've got a plan.

The dwarf individual AI is not that complicated either, a merge of various already well documented methods (need driven, order stacks, smart terrain).

What's more, I also suspect that DF may have become UI-proof.

Please... Especially not when you talk about modularity in your first post. I love this game, but no need to be blind either.

[rezaf]

What's more, I also suspect that DF may have become UI-proof.

Back in the early 2D days, I could easily imagine what a workable UI would look like. But now he's added the Z-axis? Now that half of the tables in the UI are growing foreign keys? I don't know what I'd do with that even if I had all the resources in the world. I fear it'd look like a cross between MS Access and Maya.

Many pages back we assigned the perfect UI to DF: that of the original Rollercoaster Tycoon.
It's simple enough to reasonable, could be coded by a single indivudual and could cover most, possibly all situations in DF - just open one of them windows.

Anyway, I guess he'd need to do a couple of years of priorization and bugfixing before getting DF to a state where investing lots of time in an UI
would be feasible. And that problem is only going to get worse the more partially-broken crap he adds on top.
_____
rezaf

[SpoofyChop]

I've always felt that the fundamental disagreement between the Toady apologists (myself included) and the Toady critics (myself included) comes from the fact that DF is so awesome. If it wasn't so awesome it simply wouldn't create such an intense level of feeling.

I think we're all in agreement that in a perfect world DF would both be less buggy and have a more consistent and usable text based UI AND would still have all the crazy features. The question then turns to why some people defend and some people criticize given that we don't have that.

I think it's pretty simple:

The critics are generally of the opinion that they want DF approximately the same but they would be willing to give up some of its features in order to get to the ideal picture.

The defenders are generally of the opinion that they would not be willing to give anything up of DF to fix those issues. That even if they personally don't care if he simulates every bone in the dwarven body and left and right socks separately, they just can't imagine a case in which we would have a DF with the issues fixed...that the essential qualities of the game are so tightly coupled to the quirkiness that it's basically impossible to have both of those things simultaneously. Basically, that if Toady were the kind of game dev that was concerned about bugs and modular code he wouldn't be the kind of game dev that would have written DF.

The critics tend to think that the worst possible outcome of this is to have something that's as awesome as DF conceptually but that's as buggy and inconsistent as DF is actually. So they tend to feel like we're in the worst possible situation. They have a game they want to love but can't.

The defenders tend to think we're in the best possible outcome. We have DF, it's awesome, and you simply have to appreciate it for what it is.


Personally I'm more of a defender but I see where the critics are coming from.

[zengonzo]

But some of that doesn't even have to do with complexity, but with inconsistency. Stuff that is obviously the result of a system developed of patches rather than builds.

Some things are just convenience items that shouldn't be impossible with existing code, but don't make sense in absence. I can view artifacts in a simple listing, but if I want to locate them, I have to back out and hunt through the inventory - which itself would be vastly more usable if it had some form of filtering and categorization ..

That same goes for viewing and locating dwarves.

[Madmarcus]

The defenders tend to think we're in the best possible outcome. We have DF, it's awesome, and you simply have to appreciate it for what it is.

That's a fairly good statement of my position. DF is like folk art; sometimes ugly, often jarring but containing a charm and freshness that arises from the fact that it isn't polished or steeped in 'correct" procedure.

A long while ago when I was a grad student I did some hobby programing of a much less ambitious but similar nature (baseball simulation). In the past 10 years or so I've dabbled in writing IF. In both cases much of the fun, for me at least, was in working with the code as it came up. In both cases I stopped because I realized that the options were to head in the direction Toady is demonstrating or to turn it into real work, stop treating the process as more then half the fun and do it right. Ultimately I didn't want to go either route so I dropped the projects.

Someone steeped in correct software development might not find that to be a valid distinction since to them writing bad code is probably close to a painful experience. But for a code tinkerer, a code boffin, a code McGyver the path Toady is taking isn't that hard to understand or sympathize with.

Which changes nothing about how horrible parts of the UI are. Yet somehow DF Tycoon has little appeal to me.

[mostlytigerproof]

The critics are generally of the opinion that they want DF approximately the same but they would be willing to give up some of its features in order to get to the ideal picture.

Well put.

I'm certainly not defending Tarn's priorities. I wouldn't ever take a 'features above all else' approach to any of my projects. It's sort of interesting to watch though. Niche userbases are much more loyal and adaptable than I ever thought possible.

I'm just pointing out there's a point at which DF's featureset becomes so large that it's infeasible for polish to take place. If DF hasn't reached it already, it will eventually.

[DeepT]

What I do not understand is why another gaming company hasn't done a DF rip-off with a good UI. There are lots of companies out there making very derivative games in markets saturated with other derivative games. You do not even need to be a genius game designer. Just download DF play it and write your game design document. Start with a small group of coders to implement the game engine to run your world and once that is done spin up with the artists and GUI development team. You will be printing your own money before you know it.

[DocLazy]

A long while ago when I was a grad student I did some hobby programing of a much less ambitious but similar nature (baseball simulation). In the past 10 years or so I've dabbled in writing IF. In both cases much of the fun, for me at least, was in working with the code as it came up. In both cases I stopped because I realized that the options were to head in the direction Toady is demonstrating or to turn it into real work, stop treating the process as more then half the fun and do it right. Ultimately I didn't want to go either route so I dropped the projects.

Someone steeped in correct software development might not find that to be a valid distinction since to them writing bad code is probably close to a painful experience. But for a code tinkerer, a code boffin, a code McGyver the path Toady is taking isn't that hard to understand or sympathize with.

I don't think there is anything wrong with programming in that fashion. As long as your code is readable it doesn't really matter. It does pay to have modular code, but that shouldn't really be an issue these days with modern languages. Besides, writing modular code requires minimal effort and can be really handy when you come back in a few years and are able to extract that algorithm and put it into your new code with minimal effort.

The real problem is the game runs like a dog. I can't play it on my Q6600 as it gets bogged down, before I can really set up a decent size fortress. As a programmer who has studied a bit of computer science, it can be quite frustrating. I'd like to know what's sucking up all this processing power. The dwarfs themselves are fairly simple compared to the Norns from the creatures series, which would happily run on a low end pentium. I'd presume he's doing something weird with path finding, but as stated earlier, reducing the path finding load doesn't really help.

[rezaf]

Deep games aren't made any longer and/or don't sell, that's why.

Even (at their core) pretty simple games such as Master of Magic or X-Com have always been greatly simplified in their mechanics when somebody attempted a remake because, hey, those things are too complex for modern gamers.
If no company is willing to make such game, which should (try to) remake a monster like DF?

And even if one did, I think you're underestimating the "write your design document" part. Also, DFs own design document is IMMENSE, so which part do you want to remake?
_____
rezaf

[ShivaX]

What I do not understand is why another gaming company hasn't done a DF rip-off with a good UI. There are lots of companies out there making very derivative games in markets saturated with other derivative games. You do not even need to be a genius game designer. Just download DF play it and write your game design document. Start with a small group of coders to implement the game engine to run your world and once that is done spin up with the artists and GUI development team. You will be printing your own money before you know it.

There have been a few attempts, but invariably people screw it up or realize its too hard.

[Sir Digby]

Yeah, remember that game release recently, Dungeons? Supposedly the successor to Dungeon Keeper? Turned out to be... well... utter crap.

[RepoMan]

Also, DFs own design document is IMMENSE, so which part do you want to remake?

Wait, what? DF has a design document? Can't find any such thing on bay12games.com. The features page is... well... not immense, at least bytes-wise :-)

Also, BAAA ha ha ha HAAA! Toady has a great sense of humor :-D



Edit: Never mind, the upcoming task list is considerably more voluminous. It doesn't matter that he hasn't done all that yet, it still counts as a design document!

[John Many Jars]

Toady ain't gonna change no matter how the vanillas and developer-by-numbers rage, because he has the manner of one fey and menaces with spikes of code.

So go back to raging about Elemental or something, or else the Royal Guard will hit you with obsidian swords.

[Gus_Smedstad]

he has the manner of one fey and menaces with spikes of code.

I thought it was the UI he was menacing us with.

[rezaf]

Never mind, the upcoming task list is considerably more voluminous.

I dunno where it's located these days or if Toady has outright removed it, but there used to be a massive document with numerous interconnected goals (I think he calls them arcs), their details and implications.
_____
rezaf

[Sir Digby]

I dunno where it's located these days or if Toady has outright removed it, but there used to be a massive document with numerous interconnected goals (I think he calls them arcs), their details and implications.
_____
rezaf

Ya, here you go:

http://www.bay12games.com/dwarves/dev_single.html

Reading all that sure makes it look like the game will be a life time project if it's a one man show.

[Arbit]

In the NYT article he gave an estimate of 20 years for version 1.0. He's been working on it for ~10 years. I assume releasing version 1.0 wouldn't preclude patches, possible content expansions, etc.

So yes, it's a lifetime project.

[extarbags]

Ya, here you go:

http://www.bay12games.com/dwarves/dev_single.html

Reading all that sure makes it look like the game will be a life time project if it's a one man show.

Heh, I think I remember seeing that on a different site a few years ago...

[extarbags]

Also, based on the NYT article, I can't tell if this is meant to apply to the dwarves or to Toady himself:

Bloat170, ROOM CLEANING, (Future): Cleaning rooms more responsibly, perhaps before sleeping.