<joelteon>
calculates playoff chances for haxball teams
<joelteon>
a win is worth 3 points. an ot win is worth 2 points. an ot loss is worth 1 point. a loss is worth 0
<joelteon>
each team has 3 games remaining in the season
<joelteon>
top 3 teams get into the playoffs
<devyn>
what did GHC profiling tell you?
<joelteon>
allocation is all in main.\.res.\
<joelteon>
which is pretty helpful I know
<devyn>
yeah I think I know why
<devyn>
looks like you're allocating a fuckton of lists
<joelteon>
that's what probability does
<joelteon>
not sure if it tries to evaluate the whole thing at once or what
<devyn>
main.res has no mention of Numeric.Probability.Distribution (D) or Math.Combinatorics.Exact.Binomial (B)
<devyn>
though
<devyn>
that just looks like a bunch of list logic to me
<devyn>
lists are very easily abused in Haskell :/
<devyn>
er yeah main.\.res.\
<joelteon>
well it's allocating a bunch of 10 element lists
<devyn>
well and sortBy...
<joelteon>
yeah, sorting a 10 element list
<devyn>
true
<devyn>
how many 10 element lists
<joelteon>
i dunno, like 1.2 million
<ELLIOTTCABLE>
OH
<ELLIOTTCABLE>
MY
<ELLIOTTCABLE>
GOD
<devyn>
?
<ELLIOTTCABLE>
I MANAGED A SUCCESSFUL LUNAR ORBIT
<ELLIOTTCABLE>
and, for once, retrieved the damn thing
<ELLIOTTCABLE>
totaled up to, like, 307 science points, all told :D
<ELLIOTTCABLE>
DAMN NUG
<ELLIOTTCABLE>
This is unreasonably fulfilling.
<devyn>
lol
<purr>
lol
<whitequark>
ELLIOTTCABLE: see, that's what I don't like about KSP
<ELLIOTTCABLE>
??
<whitequark>
well, let me explain
<whitequark>
I see KSP as a space simulator. and I bump into its limitations all the time, and that is annoying. have you ever looked into how it simulates aerodynamics? it's so horrible it's almost funny. then, there's the orbits.
<whitequark>
see, the problem with orbits is KSP is that they, well, aren't. KSP is fundamentally restricted to analytical solutions for orbits, because it needs to have time warp. as a result of a lack of analytical solution for 3-body problem KSP is limited to conic sections.
<whitequark>
moreover, it doesn't even solve the *2-body* problem, as planets don't affect each other gravitationally. but that is not important.
<whitequark>
what is important is that KSP's poor orbit simulation prevents some of the more interesting and important phenonena from ever appearing, for example, there are no Lagrange points in KSP;
<devyn>
no lagrangian points
<devyn>
yeah
<devyn>
and you can do some damn interesting things with those
<devyn>
very useful
<whitequark>
additionally, you cannot make a sun-synchronous orbit for your mapping satellite, because KSP simulates planets as point attractors, and sun-synchronous orbit is dependent on presence of an equatorial bulge.
<whitequark>
then there's autonomy of satellites, or lack thereof. you can get some rather interesting and engaging plugins, for example, MapSat. but you can't just leave it on, and go do something else, because KSP wouldn't calculate *anything* for the satellite while you're away from it.
<whitequark>
as a result, I would have to wait 14 hours for Kerbin to be mapped, which is not exactly my idea of "fun".
<whitequark>
then there's the fact that KSP, as it is built, has no lightspeed. this can be fixed with advent of communication satellites (another plugin), but then you yet again bump into the fact that non-focused satellites don't do anything at all; and as result, the notion of a directional antenna, positioning the satellite or orbits with precession don't make sense. which is, again, un-fun.
<joelteon>
ELLIOTTCABLE i killed the ender dragon and leveled up to 67
<whitequark>
their attempt at simulation of planes is simply laughable, being even further from any reasonable simulation by an absurd amount
<whitequark>
some of those things could get better with applied work and slightly less crude models--not scientifically accurate, but you can definitely make it much more realistic if you move from "any part regardless of position or direction generates lift" to even a simple precomputed matrix
<devyn>
>No one wants to start WWIII over this. Including Russia. They just want their freaking warm water port.
<devyn>
lol
<purr>
lol
<whitequark>
some of these things, sadly, would never improve, because it is incompatible with basic goals in game, e.g. have a time warp, not to mention the mournful lack of resources on contemporary PCs for more precise simulation
<whitequark>
well, lack of conics would not kill just time warp, it'll also kill trajectory planner and absolutely any notion of a stable orbit
<whitequark>
and it's not exactly fun to sit there all the day and correct your orbit too
<whitequark>
and this is why I will never play KSP anymore.
<whitequark>
</rant>
<whitequark>
before I gave up, I built a network of a dozen communication satellites and mostly mapped two planets, not to mention boring things like lunar landing and such
<whitequark>
of course, all that was done with stages with precisely calculated ΔV and an autopilot, not some wanky manual control. I mean, even Apollo had a guidance computer.
<whitequark>
(also, AGC's instruction set is entirely nuts. Entirely.)
<devyn>
FTL the game is alright, but honestly I'm not a huge fan of really frustrating games anyway
<whitequark>
frustrating? dude
<whitequark>
nethack is frustrating
<whitequark>
ftl is a children's walk
<devyn>
true
<devyn>
honestly games like SpaceChem are more my kind of game
<whitequark>
ugh
<devyn>
that or exploration type games
<whitequark>
spacechem lost me after one level.
<devyn>
really enjoy Starbound
<whitequark>
oh?
<whitequark>
isn't it, like, boring?
<whitequark>
it's like MMO without the whole online multiplayer part
<whitequark>
but you have to grind resources all the like
<devyn>
it has online multiplayer
<whitequark>
well, okay, massive part
<devyn>
yeah I suppose
<devyn>
but there are also really interesting things to discover
<devyn>
the procedural generation is pretty good and massive
<whitequark>
well yes, it's a good show-off of the procedural generator
<whitequark>
but not a very good game sadly
<devyn>
from your perspective :p
<whitequark>
well what is there to play with in starbound
<whitequark>
I'm serious too, I kind of want to like it
<devyn>
there are lots of really strange biomes and civilizations to discover
<devyn>
you can build shit
<devyn>
idk that's pretty fun for me
<devyn>
:p
<whitequark>
to be honest I'm starting to think I stopped liking games in general
<devyn>
take a break then
<devyn>
I did
<whitequark>
like, I can't remember when was the last time I liked a 'game' game (not a 'should've made a movie' game) at all
<ELLIOTTCABLE>
reading
<whitequark>
no, it's not about a break
<whitequark>
more like, why make virtual masturbatory crap when you can go and make real slightly-less-masturbatory crap
<devyn>
reality takes so much work
<devyn>
games just give it to us
<joelteon>
i found out about a lot of minecraft biomes yesterday
<joelteon>
speaking of biomes
<whitequark>
each game is a universe with its own laws. sadly however you train your brain to follow those laws, it's not really applicable anywhere else
<whitequark>
and if you grok them really well, you see at how constraining and pathetic they are, and that makes you sad
<whitequark>
well, not constraining. primitive? not allowing for enough emergent behavior?
<whitequark>
idk, minecraft has a lot of emergent behavior but it is social behavior, not minecraft-emergent behavior
<whitequark>
technically that last thing happens too, with all redstone turing machines and all, but that's incredible exceptions. no way I'm actually making something bigger than, idk, one redstone trigger
<devyn>
I made part of an ALU once
<whitequark>
half-adder?
<devyn>
well a whole bunch together
<devyn>
honestly though I don't really enjoy playing with pure redstone most of the time
<devyn>
I really like RedstoneChips
<devyn>
it's a basic server mod
<devyn>
that brings integrated circuits in
<joelteon>
i like killing the ender dragon
<whitequark>
did I already rant about minecraft here?
<joelteon>
pretty sure you did
<joelteon>
it's a thing that exists
<joelteon>
so you must have
<devyn>
hahaha
<devyn>
ouch.
<whitequark>
it's basically the same thing as KSP, except I was complaining about the industrialcraft mod
<whitequark>
I tried to make a power grid and was unable to, because there's no way to make a PID controller turning an engine on/off in it
<whitequark>
or in fact no feedback mechanism on grid load whatsoever
<whitequark>
I briefly considered implementing it but then java
<ELLIOTTCABLE>
interesting
<ELLIOTTCABLE>
I suppose I'm exactly KSP's target-market
<whitequark>
and apart from that, it really doesn't feel like a game anymore when you prod it and poke from inside and have to write fucking java. it feels like work.
<ELLIOTTCABLE>
it's *too* realistic for Devyn, and *not realistic enough* for whitequark
<ELLIOTTCABLE>
personally, I think you're intentionally missing the point, whitequark
<whitequark>
ELLIOTTCABLE: oh? intentionally?
<devyn>
yeah, I fully admit to really enjoying fantasy bullshit
<ELLIOTTCABLE>
we all get that you want a realistic space-flight sim. That's cool, I'd try that out, too. I don't think it'd be remotely as fun, though.
<ELLIOTTCABLE>
But KSP is, directly, a game. It's a game whose primary mechanic is, as devyn says, being way more realistic/simulated than almost any other “games” out there, but it's still clearly a game.
<ELLIOTTCABLE>
As they develop career mode, it'll become *more* game-like, not less.
<whitequark>
it's in an uncanny valley of simulators
<ELLIOTTCABLE>
I agree re: the drag factor thing, it's ridiculous. And spaceplans are currently Just Dumb.
<whitequark>
s,currently,will ever be,
<ELLIOTTCABLE>
But the orbital mechanics as a whole, the system? Definitely in the sweet-spot, *for what it is*.
<ELLIOTTCABLE>
The perfect balance of complex, but also learn-able with no background in physics, in a day or two.
<whitequark>
no way their team is pulling a remotely good space simulator in a sane timeframe.
<whitequark>
(orbital mechanics) I kinda agree
<whitequark>
it's not just in sweet-spot, it's in a global optimum.
<ELLIOTTCABLE>
nah. I think spaceplanes will be brought up to an equitable level of “just right” soon enough. They're not *focused* on spaceplanes, for obivous reasons: their only real use is servicing stations, and the current system/setup for stations is A) underdevloped, and B) also relatively pointless.
<whitequark>
I don't think you can make a game with orbital mechanics better than in KSP due to goddamn physics being a bitch
<whitequark>
and *that* is what makes me sad.
<ELLIOTTCABLE>
yeah. exactly. then it's no longer a game; it's a torture device (at the closer-to-KSP levels of accuracy) or training tool (at the closer-to-NASA levels of accuracy).
<ELLIOTTCABLE>
but I *definitely* want to see KSP developed more, personally, I'd love to see it stay in the fun-but-rewardingly-difficult sweet-spot.
<whitequark>
absolutely.
<ELLIOTTCABLE>
in the order in which I'd like to see them,
<whitequark>
training tool, there's Orbiter.
<whitequark>
but it's also not focused on what I want, which is building huge satellite networks
<whitequark>
it's also about micromanaging shit, and I *hate* micromanaging shit
<ELLIOTTCABLE>
A) more developed career goals, something to make further missions after the basic landings *exciting* still, *purposeful* still (I really, really like some of the exploration/mining type mods' idaes).
<ELLIOTTCABLE>
Then B), a *real* system for space-stations / support-sattelites, better than hacking together shit with docking ports.
<whitequark>
B) would maybe even make me reconsider KSP again.
<ELLIOTTCABLE>
then, C) spaceplanes improve to be actually a real game mechanic instead of an afterthought, to support aforementioend stations.
<ELLIOTTCABLE>
yeah.
<ELLIOTTCABLE>
the stations are the exciting part, to me.
<whitequark>
because that sounds like a subset I could actually enjoy.
<ELLIOTTCABLE>
the entire fucking point of KSP, as a whole, if we back away from all this<
<ELLIOTTCABLE>
is to RUN A SPACE PROGRAM.
<ELLIOTTCABLE>
if a space program weren't abortive, à la NASA, then the entire purpose would be setting up an orbital infrastructure to support further exploration and explotation. *That's* what's exciting.
<whitequark>
lol, abortive.
<purr>
lol
<ELLIOTTCABLE>
Not this simple, purposeless extrapolation of what NASA *has* done, which is simple exploration. That's only fun up to a point.
<ELLIOTTCABLE>
personally, I think the point where KSP would be, er, ‘complete,’ so to speak,
<ELLIOTTCABLE>
is when there's:
<whitequark>
I concur on the infrastructure part, that's what attracted me as well
<ELLIOTTCABLE>
1. a way to construct ships *in space*, as an end-game,
<ELLIOTTCABLE>
and 2. an end-game aspect that *needs* that to be anything less than insanely difficult. (Inter-system travel, or something?)
<whitequark>
have you seen garry's mod?
<ELLIOTTCABLE>
really wish there were Steam acheivements for this.
<whitequark>
I think garry's mod did satisfy some of those insane requirements of mine
<ELLIOTTCABLE>
It'd have been unreasonably exciting if a plaque had popped up and congratulated me when I achieved stable orbit around the moon. I was *already* really excited, and it would have been great to have the game acknowledge that, encourage it. (=
<whitequark>
I never got around running it because I was running linux on a potato back then
<ELLIOTTCABLE>
Linux on a Potato™
<joelteon>
Potato on a Linux
<ELLIOTTCABLE>
anyway. Back to KSP with me. love you all. REALLY want a moon landing, and then a simple space-station. After that, meh.
<joelteon>
Potato Simulator 2013
<ELLIOTTCABLE>
devyn: IMHO, knowing you, I think you should give it a second try. Did you learn to get into orbit reliably and easily before you quit? The learning-curve *before* that part is kind of ridiculous. I tried really hard to figure it out by trial and error, but I was unable to master it by myself, without giving up and reading a bunch of tutorials and
<ELLIOTTCABLE>
wiki-pages.
<devyn>
ELLIOTTCABLE: yeah I did, and I definitely had to read the wiki
<ELLIOTTCABLE>
Once you have an intuition for getting into orbit and the necessary power to get a given payload up, though, the game is a solid 90% more fun. (=
<devyn>
but I managed to get into a stable orbit several times
<devyn>
I made it to the moon
<devyn>
but not stable
<ELLIOTTCABLE>
also, how long ago? there's a relatively new flight-computer thing
<ELLIOTTCABLE>
not in the sense that whitequark means,
<whitequark>
transfer orbits are rather complex...
<ELLIOTTCABLE>
it's all still manual,
<devyn>
what's it called
<ELLIOTTCABLE>
but it lets you calculate / set up orbits.
<ELLIOTTCABLE>
look at the maneuvers page on the wiki
<devyn>
oh, okay, no, I didn't play with that
<devyn>
don't think it was there
<ELLIOTTCABLE>
it's faaaaar superior.
<whitequark>
yep, that thing is kinda awesome
<whitequark>
and its existence hinges on conics. :/
<ELLIOTTCABLE>
because you can arrange and perfect your burns *before* they happen, instead of psychotically munging with the navball in real-time
<whitequark>
FUCKING CONICS
<ELLIOTTCABLE>
whitequark: explain this to me simply? no idea what you're talking about.
<devyn>
the model of orbital mechanics
<devyn>
it uses
<devyn>
I guess
<whitequark>
ELLIOTTCABLE: conical sections
<whitequark>
what devyn said
<ELLIOTTCABLE>
splain
<whitequark>
well imagine slicing a cone with a knife^Wplane
<whitequark>
you can get a circle, an ellipse or a hyperbola
<whitequark>
that's basically three possible orbits in 2-body system where one body is far bigger than the other.
<ELLIOTTCABLE>
wat
<ELLIOTTCABLE>
first two sentences, I got. Last one threw me.
<whitequark>
there's also the notion of Sphere of Influence (SoI), describing the area where a body's gravitational attraction overwhelms attraction of other bodies
<whitequark>
okay well, your station is attracted to the planet, that's significant
sharkbot has quit [Remote host closed the connection]
<whitequark>
planet to the station, not really
<ELLIOTTCABLE>
yep
sharkbot has joined #elliottcable
<whitequark>
so if you solve the simplified equations for this case, turns out the orbits you get correspond to conical sections.
<whitequark>
I guess you can think of the cone as of "spacetime curvature" (know that demo with a rubber plane and heavy balls? yeah, kinda), if you want to know physical sense of it
<whitequark>
devyn or alexgordon may correct me
<devyn>
lol, me correcting whitequark on physics
<purr>
lol
<devyn>
that would be pretty amusing
<whitequark>
anyway. when the station doesn't leave the body's SoI, you have an ellipse or a hyperbola
<whitequark>
first for a stable orbit, second for an unstable (escape) one
<whitequark>
if the body then enters SoI of something else, you patch the corresponding conics for interaction with first and second body together
<ELLIOTTCABLE>
yah
<ELLIOTTCABLE>
makes sense, of course
<whitequark>
hence, patched conics.
<ELLIOTTCABLE>
why “fucking conics?”
<whitequark>
because that's not how gravity works :p
<ELLIOTTCABLE>
well duh
<ELLIOTTCABLE>
what does it make so terrible, I mean?
<whitequark>
no lagrange points, no non-point attractors
<ELLIOTTCABLE>
k.
<whitequark>
I mean, you could salvage those within KSP's model
<ELLIOTTCABLE>
and with other models that don't pretend everything's a space-time cone, you can't do fast-forwarding, why?
<whitequark>
well, because of accumulation of error in calculations
<whitequark>
but
<whitequark>
I think that's kinda bullshit, because you have what, 10-body problem?
<whitequark>
those can be solved *very* quickly on contemporary CPUs to any precision one'd want
<whitequark>
and if you throw CUDA in, you can fucking *demolish* that problem
<ELLIOTTCABLE>
okay? and?
<whitequark>
more or a problem would be this:
<whitequark>
with nbody, there are no stable orbits.
<ELLIOTTCABLE>
why not fake it?
<ELLIOTTCABLE>
do all the calculations as accurately as possible, and then *downgrade* the accuracy, so to speak (cheat.), to make it fun.
<ELLIOTTCABLE>
if an orbit is predicted to have smaller than N variation for X time, then cheat and make it perfectly stable.
<ELLIOTTCABLE>
… and other such solutions in that area.
<whitequark>
that'll make lagrange points impossible as well
<ELLIOTTCABLE>
why? :x
<whitequark>
because they're metastable
<whitequark>
you have to continuously correct your orbit in order to stay here
<whitequark>
it's just easier than in other places
<whitequark>
well, what would work is an "in-game flight computer" performing those correction burns
<whitequark>
that considerably adds to complexity to the game and likely would require major refactoring
<whitequark>
the authors said we're stuck with patched conics forever.
<whitequark>
:(
<whitequark>
I'm 100% sure that given current hardware, you can solve nbody to any desired precision ever relevant in-game quickly enough to have warp.
<whitequark>
hell, just link to that Fortran library, how was it
<whitequark>
LAPACK?
<whitequark>
nbody is one of its motivating examples :D
<whitequark>
<3 FORTRAN
<purr>
Let it be known that whitequark hearts FORTRAN.
<ELLIOTTCABLE>
go write a competitor. :P
<joelteon>
that would take effort
<whitequark>
that would take several man-years of effort
<whitequark>
*shrug*
<purr>
¯\(º_o)/¯
<whitequark>
I'll rather go play with my 12 balls, they're so round and shiny