Back to Home / #openttd / 2015 / 05 / Prev Day | Next Day
#openttd IRC Logs for 2015-05-31

---Logopened Sun May 31 00:00:09 2015
00:30-!-guru3-vps [~guru3-vps@] has joined #openttd
00:52<Jimmy_Bignuts>i would like to host one of these games
00:53<Jimmy_Bignuts>can you run a server version of this game on linux?
00:56-!-Eddi|zuHause [] has quit []
00:56-!-Eddi|zuHause [] has joined #openttd
01:02-!-guru3-vps [~guru3-vps@] has quit [Ping timeout: 480 seconds]
01:23-!-roidal [] has joined #openttd
01:24-!-Supercheese [] has joined #openttd
02:10-!-roidal [] has quit [Quit: WeeChat 1.1.1]
02:12-!-roidal [] has joined #openttd
02:23-!-andythenorth [] has joined #openttd
02:24<andythenorth>can CivilAI occasionally build some pointless road bridges ?
02:24<andythenorth>it spoils my shipping empire :P
02:25<Pokka>how pointless did you have in mind?
02:25<andythenorth>I have no serious suggestion :P
02:25<andythenorth>but the roads block canals / lowering land for ship canals
02:26<Pokka>oh, right
02:26<andythenorth>doesn’t stop me using it in games
02:26<andythenorth>but eh
02:26<Pokka>no more than any other AI's routes, I guess
02:26<Pokka>with the update, all you have to do is build some tracks over its roads and squash a car on it
02:27<Pokka>then it will remove the road and build a bridge over your tracks. remove tracks, replace with canal?
02:28<andythenorth>nice hax
02:42<andythenorth>should I company colour the vehicles?
02:42<andythenorth>also buildings?
02:42<Pokka>why not?
02:42<Pokka>perhaps not the buildings entirely, give them different coloured roofs or stripes or something. Or too much work? :)
02:44<andythenorth>I’m pretty quick at colouring in buildings
02:44<andythenorth>done a few now :P
02:45*andythenorth must also make a power station
02:46<Pokka>I have some bits of an oldeworlde one lying around, if it's any help... but you've probably got enough industry bits of your own.
02:46<andythenorth>is it base set sprites?
02:47<andythenorth>or you drew some?
02:47*andythenorth needs a cooling tower probably
02:47<Pokka>actually, never mind, it's got too many bits of base set sprites
02:47<Pokka>and no cooling tower. :)
02:47<andythenorth>there’s always OpenGFX
02:48<Jimmy_Bignuts>is there any way to exit a game but have your company continue to grow? then re-enter and continue where you left off providing you dont get bankrupt?
02:49<Pokka>on multiplayer? depends on the server whether your company gets wiped or not.
02:52<andythenorth>also this builders yard
02:55<Pokka>internet is playing up, doesn't want to load openttdcoop
02:55<Pokka>nor google, apparently
02:55<andythenorth>australia has special internet
02:55<andythenorth>sometimes bits of internet aren’t there for you
02:56<andythenorth>it’s extra special for those of us who have servers in australia :P
03:02<Pokka>even those of us in australia are sensible enough to keep our servers somewhere else :)
03:04-!-Flygon_ [] has joined #openttd
03:11-!-Flygon [] has quit [Ping timeout: 480 seconds]
03:13<andythenorth>eh, so power plants
03:14<andythenorth>must to have an acceptance limit?
03:14<andythenorth>tied to town population?
03:14<Pokka>does anything else have an acceptance limit?
03:14<Pokka>I thought that stockpiling wasn't generally a FIRS mechanic :P
03:17<andythenorth>also don’t really want to add it :P
03:17<Pokka>I guess that's the answer then
03:17*andythenorth must try harder
03:17<andythenorth>produce building materials?
03:17<andythenorth>bit odd tbh :P
03:20<andythenorth>generate electricity?
03:20<andythenorth>moved by a new railtype?
03:24-!-wicope [] has joined #openttd
03:29-!-namad7 [] has joined #openttd
03:29<Supercheese>tried that, was sort of eh
03:30<Supercheese>general consensus seemed to be "needs a GS to make things work nicely"
03:31*Supercheese is too lazy to write an electricity GS
03:31<andythenorth>eh, it’s a bad use of a GS
03:31<andythenorth>because there’s only one GS per game
03:31<andythenorth>using it to do a show-pony cargo trick is daft
03:31<Supercheese>eh I don't use any at the moment
03:31<Supercheese>my slot is free
03:31<Supercheese>still too lazy
03:35-!-Alberth [~alberth@2001:981:c6c5:1:be5f:f4ff:feac:e11] has joined #openttd
03:35-!-mode/#openttd [+o Alberth] by ChanServ
03:48<supermop>needs to be a better way of seeing topography through trees
03:48<andythenorth>transparency :P
03:49-!-OsteHovel [] has quit [Ping timeout: 480 seconds]
03:50<supermop>better than that
03:50-!-OsteHovel [] has joined #openttd
03:50<supermop>currently the only reason to have trees visible to to avoid destroying them,
03:50<@Alberth>you know transparency includes making trees invisible?
03:51<@Alberth>ah :)
03:51<@Alberth>just avoid towns :p
03:51<supermop>if they are visible the land looks ugly because they completely hide any contours - so the whole map looks flat
03:52<supermop>current map has some nice mountains - but cannot see them if trees are on
03:52<supermop>and if trees are off, then the mountains look too bare
03:53*andythenorth always hides
03:53<andythenorth>can’t build anything with trees on
03:53<andythenorth>too hard to see
03:53*Supercheese agrees
03:54-!-sim-al2 [~chatzilla@2602:306:cdd9:e4c0:a43e:d2c6:73c1:a7f8] has quit [Ping timeout: 480 seconds]
03:54<supermop>i guess a more sparse set of trees would help
03:55<supermop>but i think shading trees based on tile slope would make landscapes look pretty nice
03:55<andythenorth>use the original tree algorithm
03:55<andythenorth>better results
03:55<andythenorth>the ‘improved’ algorithm isn't
03:55*andythenorth just tested
03:58<supermop>the problem isn't really that i can't build stuff - it's that trees look ugly when they become essentially just green noise
03:59<andythenorth>tried both algorithms?
03:59<supermop>it's been awhile
04:01<supermop>man there are a lot of north american diesels that i have a really hard time bringing myself to care about
04:02<supermop>like I had no idea that this thing existed:
04:02-!-Pensacola [] has joined #openttd
04:02<supermop>looks pretty underwhelming
04:03-!-sla_ro|master [slamaster@] has joined #openttd
04:30-!-Progman [] has joined #openttd
04:42<@Alberth>it has good colours
04:42-!-namad7 [] has quit []
04:43<@Alberth>and I have been using the "none" tree algorithm, works quite nicely, although eventually, you get a few more trees than "none" :)
04:44<@Alberth>also, playing other than temperate climate helps :)
04:44<@Alberth>hmm, maybe also not toyland, it's also quite tree-ish
04:45<andythenorth>I like Evilweed
04:45<andythenorth>nice location for the port
04:45<andythenorth>looks like Canadian northwest coast
04:45<@Alberth>nice :)
04:46<@Alberth>would look good in the screenshot section :)
04:47*andythenorth rarely goes there :)
04:48<@Alberth>no need, you have your own pretty little towns :)
04:49<@Alberth>now you only need a few stations in the big city, and you have covered it
04:57-!-guru3-vps [~guru3-vps@] has joined #openttd
05:00*andythenorth refactors
05:00<andythenorth>refactoring is like doing work without having to do work
05:00*andythenorth should refactor more
05:02<@Alberth>indeed :)
05:03-!-sla_ro|master2 [slamaster@] has joined #openttd
05:03-!-zeknurn` [] has joined #openttd
05:04<andythenorth>no design needed, very few decisions, lots of commits :P
05:04<andythenorth>way easy
05:05<@Alberth>you should have an idea of direction, but other wise, yep :)
05:07-!-HerzogDeXtEr [] has joined #openttd
05:08-!-Netsplit <-> quits: Pikka, sla_ro|master, OsteHovel, guru3-vps, zeknurn, ccfreak2k, wicope
05:08-!-zeknurn` is now known as zeknurn
05:11-!-OsteHovel [] has joined #openttd
05:13-!-ccfreak2k [] has joined #openttd
05:18-!-wicope [] has joined #openttd
05:35-!-JacobD88 [] has joined #openttd
05:45-!-Pokka [] has quit [Read error: Connection reset by peer]
05:46-!-Pokka [] has joined #openttd
05:51-!-andythenorth [] has left #openttd []
06:12-!-Speedy [] has quit [Remote host closed the connection]
06:15-!-Hiddenfunstuff [] has joined #openttd
06:25-!-Speedy` [] has joined #openttd
06:26-!-Speedy` is now known as Speedy
06:27-!-Myhorta [] has joined #openttd
06:42-!-Suicyder [~Suicyder@] has joined #openttd
07:01-!-nikow_ [] has quit [Ping timeout: 480 seconds]
07:06-!-JacobD88 [] has quit [Quit: JacobD88]
07:09-!-Alberth [~alberth@2001:981:c6c5:1:be5f:f4ff:feac:e11] has left #openttd []
07:13-!-Progman_ [] has joined #openttd
07:17-!-Progman [] has quit [Ping timeout: 480 seconds]
07:17-!-Progman_ is now known as Progman
07:31-!-Supercheese [] has quit [Read error: Connection reset by peer]
07:31-!-Supercheese [] has joined #openttd
07:40-!-Pensacola [] has quit [Remote host closed the connection]
07:53-!-Wolf01 [] has joined #openttd
07:53<Wolf01>hi hi
08:05<jjavaholic>what is the basic construction of a side chain
08:08-!-Ketsuban [~ketsuban@] has joined #openttd
08:17-!-frosch123 [] has joined #openttd
08:21-!-heffer [] has joined #openttd
08:22-!-kamnet [] has joined #openttd
08:22-!-Pokka [] has quit [Read error: Connection reset by peer]
08:24-!-heffer_ [] has quit [Ping timeout: 480 seconds]
08:26-!-HerzogDeXtEr [] has quit [Quit: Leaving.]
08:30-!-andythenorth [] has joined #openttd
08:42<andythenorth>‘industries for the sea'
08:42*andythenorth ponders
08:43<Jimmy_Bignuts>how do i exit a game, and return tomorrow with it still running?
08:45<supermop>on a server?
08:46-!-Myhorta[1] [] has joined #openttd
08:46<kamnet>I think you got all the sea industries covered, andy
08:47<supermop>or just leave your computer on all night
08:47<kamnet>unless you want to make a cultured pearl farm. :D
08:47<supermop>floating fishing village
08:48<supermop>for vietnam economy
08:48<supermop>metabolist floating cities
08:49-!-Jimmy_Bignuts [] has quit [Quit: HydraIRC -> <- *I* use it, so it must be good!]
08:50<kamnet>kelp fields
08:51-!-Myhorta [] has quit [Ping timeout: 480 seconds]
08:53<andythenorth>deliver nothing
08:53<andythenorth>produce nothing
08:54<Eddi|zuHause>that's called an "object"
09:02<supermop>good night
09:03<andythenorth>over-sized spritesheets aren’t going to make much difference to compile time?
09:03<andythenorth>probably negligible or undetectable?
09:06-!-guru3-vps [~guru3-vps@] has joined #openttd
09:06<frosch123>all images files are loaded onlyonce
09:06<frosch123>encoding of sprites is sorted by source file
09:06*andythenorth files refactoring spritesheets under ‘tmwftlb’
09:07<frosch123>anyway, if you compile multiple times, you use the cache anyway
09:07<frosch123>so, maybe separate those sprites, which you change every 5 minutes :p
09:08<andythenorth>I did consider again an nml->nfo compile
09:08<andythenorth>for cases where only the graphics change
09:09<Eddi|zuHause>but that means it can't use the graphics cache, which may make things actually slower when you change code and no graphics
09:10<andythenorth>I’d have to have two compiles :P
09:10-!-liq3 [] has joined #openttd
09:10<andythenorth>and choose the right one for the work that day
09:10<andythenorth>seems lame
09:10<Eddi|zuHause>i don't think "lame" is the right word :p
09:11<andythenorth>more superlative than ‘lame'?
09:12<andythenorth>those trucks
09:13<andythenorth>should be removed?
09:14*andythenorth is going to tweak a few things, but lacks inspiration
09:18<andythenorth>dragline gravel pits are pretty minimal
09:23<andythenorth>considered animating the crane, but I’d have to include multiple turning angles
09:23<andythenorth>total PITA :)
09:34-!-Alberth [~alberth@2001:981:c6c5:1:be5f:f4ff:feac:e11] has joined #openttd
09:34-!-mode/#openttd [+o Alberth] by ChanServ
09:40-!-krinn [] has joined #openttd
09:40<krinn>hi guys
09:40<krinn>Alberth got a min? (i don't want report a non bug again)
09:42<krinn>looks like i'm unable to build a bridge from slope_N to slope_E
09:43<krinn>but that's not a define limit in the api (of course slope_E and slope_N are opposite tiles, same tileX...)
09:43<krinn>while by hands i'm able to build it
09:44<krinn>(and yes, build on slope is on of course)
09:44<Eddi|zuHause>i'm sure dozens of AIs managed to build bridges
09:46<krinn>lol Eddi|zuHause sure, but they only do it because the pathfinder hint them on where to build it, and the pathfinder ONLY gave them a tile source or target slope_SE/SW/NE/NW
09:46<krinn>never a S/N/W/E
09:47<Eddi|zuHause>that seems like a shortcoming in the pathfinder
09:48<krinn>maybe, but that's looks fine then if you cannot have the BuildBridge function to build a bridge on a tile that isn't NW/SE/NE/NW
09:48<Eddi|zuHause>anyway, what do you want us to do now?
09:48<@Alberth>give magical answers :)
09:49<krinn>well, confirm it's a limit from openttd or a bug?
09:49<krinn>i don't see why i could do a bridge by hands and by API i cannot do the same
09:49<@Alberth>but I don't have them, I don't have an AI that shows it fails
09:49<krinn>can pack mine, but its state is really poor now
09:49<@Alberth>krinn: I agree, an AI should be able to do what a user can do
09:50<Eddi|zuHause>krinn: try to make a reduced example that only tries to build such a bridge
09:50<@Alberth>as simple as possible, so your AI code doesn't contain errors
09:51<krinn>Eddi|zuHause, erf, we need a special map/condition, it's not common to find an area slope_N slope_E to build a bridge on
09:51<krinn>a screenshot and the code that try to build the bridge would do?
09:51<Eddi|zuHause>krinn: make a savegame/scenario with the right condition then
09:51<@Alberth>make a map, or test specific tiles to have some confiuration
09:52<@Alberth>or tarreform tiles to the right form
09:52<@Alberth>it doesn't have to be complete, it can be mostly manual, whatever, the code just has to be as easy as possible, so you can concentrate on the bridge build problem
09:54<@Alberth>if you can show that the simple code fails, and manually it doesn't fail, it's a bug
09:55<@Alberth>eg in my signal tester, I had hard-coded positions for the test area
09:56<krinn>ok, going to build a tiny ai that build bridge only
09:57<@Alberth>ok, thank you
10:00<@Alberth>It's often a useful exercise with these problems. If the tiny example fails, you have a piece of code to demonstrate the problem (just like I did, "my code works, could you have a look?")
10:00<@Alberth>if the tiny example works, you know the problem is not the bridge itself, but elsewhere
10:01<krinn>i know, just that this time i have check more than once (ok i might still be wrong more than once)
10:03<@Alberth>another tactic that often works is to post the question at the forum, but also there you need to provide something to show the problem
10:05<krinn>i'm doing the test ai, so my ai code will be out of the problem
10:20-!-nikow [] has joined #openttd
10:30<krinn>ok done, testmap + testai
10:31<krinn>the ai fails on two bridge while is able to build a 3rd test one
10:32<krinn>wants bugreport with the map and ai to test?
10:41-!-roidal_ [] has joined #openttd
10:46<@Alberth>sounds good
10:48<krinn>weirds, the same function is able to build one, and fail on the two other
10:48-!-roidal [] has quit [Ping timeout: 480 seconds]
10:52<krinn>Alberth, confirm at least code "seems" ok for ai part?
10:57<@Alberth>looks ok-ish at first sight
10:58<@Alberth>a and b are a bit magic, you may want to compute the tile index the next time (making it easier to adjust the code for further experiments)
10:58<@Alberth>"c == false" is normally expressed as "!c" :)
10:58<@Alberth>but for testing it's all fine enough
10:59-!-tokai|mdlx [] has joined #openttd
10:59<krinn>i've put sign to help see start & end and find them fast on the map
10:59<krinn>and i tooked a & b to be close middle map, so they should be easy to get when loading it
11:00<@Alberth>I saw, but TileIndex(100, 100) is easier to understand
11:00<@Alberth>but just for a next time, no need to change this any more
11:01<krinn>ah yes, will try keep it in mind for next time
11:01<krinn>i know i should kept some lower openttd version to test, but i only have 1.4.4 right now
11:02<@Alberth>ha, I test in trunk :)
11:03<krinn>2 fails on trunk too?
11:03-!-tokai [] has quit [Ping timeout: 480 seconds]
11:04<@Alberth>don't know, didn't actually try to run the code yet
11:06<krinn>was trying to optimize path result of pathfinder, i think i know why pathfinder was sucking at building bridge now :)
11:14<andythenorth>my compile got slower
11:18<andythenorth>about 14s :(
11:18<andythenorth>just for changing one industry
11:19<V453000>world ends right the hell now then
11:20<andythenorth>it’s ~28% slower
11:21<andythenorth>just due to refactoring
11:21<andythenorth>that’s insane
11:23-!-guru3-vps [~guru3-vps@] has quit [Ping timeout: 480 seconds]
11:23<@Alberth>you have this too?
11:24<@Alberth>I have no signs
11:25<krinn>enable seeing other company sign :)
11:25<@Alberth>nvm thats my fault\
11:25<krinn>or switch to the ai company to see them
11:25<krinn>i'm glad only one is build for you too :D
11:27<krinn>the signs are good help, at least they shown the S & E points, and it's easy to see they are good points to build a bridge on
11:27<@Alberth>looks like it
11:27<krinn>what doesn't really help is the ERR_UNKNOWN
11:30<krinn>it's also fail if you set currentroadtype to ROAD and try build road bridges (so at least it's not VT_RAIL depend)
11:31<@Alberth>good to know
11:39<andythenorth>ground aware spritelayouts are insanely inefficient to compile
11:41<krinn>i've heard Eddi|zuHause saying he has put plenty noop() and sleep() in it to bug newgrf makers to death
11:44<Eddi|zuHause>that's libel!
11:46<Eddi|zuHause>andythenorth: of course optimizing for worktime means that the optimal solution is doing no work at all.
11:46<Eddi|zuHause>you're done the quickest
11:47<andythenorth>depends if you do achievement = work / result
11:47<andythenorth>or such
11:47<andythenorth>or result / work
11:48*andythenorth wonders why the damn thing is so slow
11:48*andythenorth reads the nml
11:49<andythenorth>eh, does nml incur any time parsing empty lines?
11:49<andythenorth>if that was significant, I would be surprised
11:49<andythenorth>but eh
11:50<Eddi|zuHause>my experience in that matter is very old...
11:51<Eddi|zuHause>back then, the biggest impact was complex formulas e.g. in templates
11:51<Eddi|zuHause>empty lines should be handled in ply
11:51<Eddi|zuHause>nml won't ever see those
11:52<krinn>might also be a good idea to really drop that zx81 and change your computer andythenorth
11:54<Eddi|zuHause>andythenorth: but i have never seen you complain about that ratio. you only ever complain about compile time
11:54<andythenorth>slow is bad
11:55<krinn>oh speaking off, would you go for a 6xcores i58xx or 4xcores 4970 intel cpu? (i'm unsure the ref, the first is newest gen with 6cores, the other previous gen but at 4ghz)
11:55<andythenorth>stick with the zx81
11:57<andythenorth>the nml compile time is only 35s on average
11:57<andythenorth>so wtf is eating the rest of the time
11:57*andythenorth explores
11:58<@Alberth>krinn: max speed for single core is what counts :)
11:59<andythenorth>1s to run chameleon templating
11:59<andythenorth>2s to render docs
11:59<andythenorth>there are 20-30s missing somewhere
12:00<krinn>Alberth, i think so, alas the extra cores won't beat up the extra ghz on most apps
12:00<andythenorth>ach, how do I profile a makefile :(
12:01<krinn>andythenorth, did you try use a tmpfs to see if may help?
12:01<andythenorth>is that a ramdisk or what?
12:02<andythenorth>eh, no idea how to set that up on OS X
12:02<andythenorth>doubt that IO is the problem
12:02<@Alberth>simple way is to add "time" to the commands, time nmlc ....
12:02<krinn>well, if you keep loading the same groundpic other & other it could
12:02<@Alberth>which dumps timing of the executed comand
12:03<andythenorth>so I need to find all the commands :)
12:03<@Alberth>krinn: nmlc uses lots of caching
12:03<@Alberth>no variable for that?
12:03<@Alberth>NMLC = nmlc or such
12:03<@Alberth>and in the rule $(NMLC) ....
12:04<@Alberth>then you can change to NMLC = time nmlc
12:04<@Alberth>you can also do verbose build _V= make
12:05<@Alberth>which should dump all commands as they are executed, that may give a hint what takes time
12:05<krinn>andythenorth, dunno for os/x look at /proc/filesystems if you can use tmpfs
12:05<andythenorth>all I found so far was a thing saying “don’t bother with this for BSD-like systems” :)
12:05<andythenorth>it can be done, but apparently it’s a bad idea
12:05<@Alberth>it's probably different for osx, as it's BSD-based
12:06<@Alberth>don't you have ssd ?
12:07<@Alberth>I seem to remember that, at least
12:07<andythenorth>it puts about 600MB/s through on average
12:08<@Alberth>disk times are probably negligible then
12:08<andythenorth>also I know the main factor is the nml source
12:08<@Alberth>seek times are more relevant, as you have lots of small files
12:08<andythenorth>for nml compile time
12:08<krinn>did you then try see if you can alter ccache to handle nmlc?
12:09<@Alberth>time make firs.nml ?
12:09<andythenorth>IO isn’t going to affect advanced varaction 2 expansion times in spritelayouts ;)
12:09<andythenorth>it’s absolutely not disk bound
12:10<andythenorth>with everything cached, a make run is 11s
12:10<@Alberth>nml is quite inefficient in expression simplification afaik
12:10<andythenorth>of that 11s, 2.5s is spent in nmlc
12:10<andythenorth>so the makefile has some base overhead somewhere
12:10<andythenorth>I can’t understand the makefile yet
12:10<krinn>dunno could also look at your makefile and ignore some steps if you are sure the result won't change from previous
12:10<andythenorth>it is...sophisticated
12:11<Eddi|zuHause>andythenorth: that means the inefficiency is in your generator script?
12:11<andythenorth>Eddi|zuHause: plausibly
12:11<andythenorth>although the code generator is < 1s to run when cached
12:11*andythenorth adds a time to that
12:14<krinn>that's a kickass makefile
12:14<krinn>do everything for you
12:15<V453000>spend years writing it :P
12:16<andythenorth>is it just expected that make has a significant overhead?
12:16<andythenorth>it has to dep check, yes/no?
12:17<Eddi|zuHause>it has to check the date of every file it is supposed to touch
12:17<Eddi|zuHause>also, build a dependency tree
12:17<krinn>maybe you can try /bin/sh if there's no bashism, people keep saying sh is faster
12:18<@Alberth>I don't trust lines like 219
12:18<Eddi|zuHause>if you have a fractured directory tree, checking the dates may take some disk access time
12:18<andythenorth>Alberth: +1
12:19<@Alberth>nested for-loop, long pipe lines
12:19<andythenorth>but I think that’s turned off
12:19<krinn>it's also not that good to assume sed == gnu/sed for people using bsd/sed
12:20<andythenorth>what is custom_tags.txt?
12:20<krinn>and gnu/sed kick bsd/sed ass by far :)
12:21<Eddi|zuHause>andythenorth: thing that puts hg revision into the version info of the grf, and stuff
12:21<andythenorth>how long is CPP spending?
12:21*andythenorth tests that
12:21<andythenorth>historically, CPP has been treated as near-instant
12:22<Eddi|zuHause>andythenorth: shouldn't be significantly longer than reading in the file and writing it out again
12:23<krinn>oh! football on tv, later guys
12:23<Eddi|zuHause>that's kinda the point of a simple text replacement engine
12:23<Eddi|zuHause>andythenorth: i'd put a time on that line Alberth mentioned
12:24<Eddi|zuHause>andythenorth: also, try "_V=time make"
12:24<andythenorth>L219 is wrapped in an ifdef
12:24<andythenorth>I don’t think it’s called
12:24<andythenorth>but I’ll test
12:26<andythenorth>the echo line above is never printed
12:26-!-sim-al2 [~chatzilla@2602:306:cdd9:e4c0:795f:b09c:8603:7604] has joined #openttd
12:28<Eddi|zuHause>then try that last line i mentioned
12:28<Eddi|zuHause>it will time each individual command
12:28<Eddi|zuHause>you can then sum that up, and the difference to a "time make" will be the time spent in make itself
12:39<Eddi|zuHause>you can also do "_V=time time make" to do that all in one go
12:39<Eddi|zuHause>try not to combine this with a -jX :p
12:57-!-smoke_fumus [~smoke_fum@] has joined #openttd
13:07*andythenorth finds nothing obvious
13:11-!-glx [] has joined #openttd
13:11-!-mode/#openttd [+v glx] by ChanServ
13:15<@Alberth>so dependency calculation or time-stamp checking perhaps?
13:16<@Alberth>or perhaps something as simple as collecting all files that are part of the project
13:18<@Alberth>hmm, it does an awful lot of $(shell ... )
13:25<frosch123>so, processing .bashrc :p
13:26<frosch123>not sure whether $shell would trigger reading that
13:26<@Alberth>probably it will
13:26<@Alberth>make help is quite fast for me though
13:27<@Alberth>less than a second
13:29-!-jjavaholic [] has quit [Read error: Connection reset by peer]
13:30-!-jjavaholic [] has joined #openttd
13:32<andythenorth>pretty certain the earlier (10-15s slower) results were an aberration
13:32<andythenorth>think my laptop was running some system task
13:33<andythenorth>compile times now are back where they were yesterday
13:33<andythenorth>still, the nmlc part takes 35s, and the makefile takes 48s total
13:33<andythenorth>so something is filling that gap :)
13:34<@Alberth>the "" line was quite slow here
13:34<@Alberth>but I couldn't pinpoint the program
13:40<andythenorth>FIRS can do single industry compiles
13:40<andythenorth>for those, nmlc takes 3-4s
13:40<andythenorth>makefile takes ~16s
13:41<andythenorth>-include Makefile.dist
13:41<andythenorth>there is no Mafkefile.dis
13:42<andythenorth>or even Makefile.dist :P
13:42<@Alberth>that wouldn't take 12 seconds :p
13:43<frosch123>maybe it scans the network drives for it :p
13:43<@Alberth>but to get a handle on things, maybe write a shell script that runs the computations?
13:44<@Alberth>although that may take more time than you ever gain :)
13:51<andythenorth>maybe my nmlc timing runs are wrong
13:59<andythenorth>oh nvm :)
13:59*andythenorth got bored finally
14:04<Eddi|zuHause>andythenorth: lines starting with "-" mean "don't error out when this doesn't work"
14:05-!-HerzogDeXtEr [] has joined #openttd
14:06<Eddi|zuHause>andythenorth: did you read the log about me telling you something, then arguing about forgetting it, and now i forgot it?
14:08<andythenorth>can’t remember
14:08<Eddi|zuHause>was yesterday, i think, and you went away before i said it
14:19<andythenorth>floppy versions? <-- at least provide sloped versions of the outer rim tiles
14:26<Eddi|zuHause>that seems like it.
14:28<andythenorth>the inner tiles of the pit will build on slopes
14:28<andythenorth>which will be hidden
14:29<andythenorth>the processor buildings also build on slopes
14:29<andythenorth>I think the placement issue is broadly solved
14:29<andythenorth>I’m getting the expected number on mountainous maps
14:35-!-guru3-vps [~guru3-vps@] has joined #openttd
14:42-!-Plaete [] has joined #openttd
14:58<andythenorth>so if I replace L161 with the following line, it saves about 10s:
14:58<andythenorth> REPO_VERSION_STRING ?= "FOO"
14:59<andythenorth>I get a 39s compile, of which 35s is nmlc
14:59<andythenorth>which is appropriate
15:00<andythenorth>there’s a stack of hg stuff, from L129 or so
15:19<Eddi|zuHause>so, make here took 40 seconds, 25s of which were "render pnml", then nmlc crashed with "no module named 'ply'"
15:24<andythenorth>that’s exciting
15:24<andythenorth>pip install ply? :P
15:26<Eddi|zuHause>already beyond that
15:26<Eddi|zuHause>nmlc takes 1m17s
15:27<frosch123>wrt. hg you might try to replace some ?= with :=
15:27-!-sla_ro|master2 [slamaster@] has quit []
15:29<Eddi|zuHause>total time 2m1s
15:29<Eddi|zuHause>so there's definitely stuff happening inbetween
15:30<Eddi|zuHause>from my limited experience i'd also guess it's the mere fact of calling $(shell ...) that's slow
15:32<andythenorth>Eddi|zuHause: second run is significantly faster btw
15:32<andythenorth>even with changes (limited changes)
15:32<Eddi|zuHause>andythenorth: but i'm not interested in second runs.
15:32<andythenorth>first run, chameleon is very slow
15:33<Eddi|zuHause>andythenorth: they make very terrible benchmarks
15:33<andythenorth>should benchmark common case no?
15:33<andythenorth>dunno, the engineering priniciple here seems to be use caching :)
15:34<andythenorth>any non-primed cache will be slow
15:34*andythenorth wishes it were otherwise :|
15:35<Eddi|zuHause>andythenorth: but... you're always interested in the cases where you changed something that's not in the cache.
15:37<andythenorth>caches are fine-grained afaict
15:37<andythenorth>typically only some small % of stuff is changed in FIRS between compile runs
15:37<andythenorth>chameleon cache is fine-grained, compiled .pyc per template
15:37<andythenorth>nml caches _seem_ to be quite fine-grained
15:38<andythenorth>what is all that hg stuff actually for?
15:39<Eddi|zuHause>andythenorth: still the same thing. putting the version into the grf
15:39<Eddi|zuHause>revision hash, branch/tag name, age, ...
15:40*andythenorth looks for an equivalent to git describe
15:40<frosch123>you proabably want the script for the nml make framework
15:40<andythenorth>ho maybe
15:40<frosch123>instead of those calls within the makefile
15:40<andythenorth>sounds highly plausible
15:40<frosch123>and you likely want to use many := instead of ?= or =
15:41<andythenorth>what’s the difference? o_O
15:41<frosch123>:= evaluates once and stores result
15:41<frosch123>?= and = reevaluate on every usage
15:46<andythenorth>just changing those can’t make 8s speed difference :o
15:46<frosch123>how often is hg called and asked whether the working copy is modified?
15:47<frosch123>by using = i would not be surprised if you get to numbers like 32 times :p
15:51<andythenorth>primed caches, 38s compile, 35s of which is nmlc
15:53<andythenorth>single-industry compile, 6s
15:54<andythenorth>that’s more like it :P
15:58<andythenorth>now all I need is a flag to tell nmlc that only pngs have changed
16:00-!-Plaete [] has quit [Quit: Nettalk6 -]
16:15<andythenorth>so for FIRS 1.4.4 (current trunk), best-case compile was ~1m15s
16:15<andythenorth>and single-industry compiles weren’t possible
16:30-!-roidal_ [] has quit [Quit: WeeChat 1.1.1]
16:34-!-Ketsuban [~ketsuban@] has quit [Quit: To robbery, slaughter, plunder they give the lying name of empire; they make a desert and call it peace.]
16:37-!-sla_ro|master [slamaster@] has joined #openttd
16:39-!-sla_ro|master [slamaster@] has quit []
16:42-!-Ketsuban [~ketsuban@] has joined #openttd
16:42-!-krinn [] has quit [Ping timeout: 480 seconds]
16:49<andythenorth>also bedtime
16:49-!-andythenorth [] has quit [Quit: andythenorth]
16:50-!-Alberth [~alberth@2001:981:c6c5:1:be5f:f4ff:feac:e11] has left #openttd []
16:51-!-krinn [] has joined #openttd
16:52-!-efess [~Efess@] has quit [Ping timeout: 480 seconds]
17:03-!-krinn [] has quit [Ping timeout: 480 seconds]
17:04-!-frosch123 [] has quit [Quit: be yourself, except: if you have the opportunity to be a unicorn, then be a unicorn]
17:04<Eddi|zuHause><andythenorth> now all I need is a flag to tell nmlc that only pngs have changed <-- that exists. it's called a dependency.
17:12-!-krinn [] has joined #openttd
17:17-!-glevans2 [~glevans2@] has quit [Quit: Leaving.]
17:18-!-wicope [] has quit [Remote host closed the connection]
17:22-!-glevans2 [~glevans2@] has joined #openttd
17:42-!-krinn [] has quit [Quit: Leaving]
18:02-!-HerzogDeXtEr [] has quit [Quit: Leaving.]
18:05-!-heffer [] has quit [Ping timeout: 480 seconds]
18:06-!-supermop_ [] has joined #openttd
18:07-!-Progman [] has quit [Remote host closed the connection]
18:11-!-Wolf01 [] has quit [Quit: Once again the world is quick to bury me.]
18:13-!-heffer [] has joined #openttd
18:13-!-supermop [] has quit [Ping timeout: 480 seconds]
18:45-!-efess [] has joined #openttd
18:56-!-Suicyder [~Suicyder@] has quit [Quit: HydraIRC -> <- Wibbly Wobbly IRC]
19:04-!-FLHerne [] has joined #openttd
19:31-!-argoneus_ is now known as argoneus
19:52-!-Ketsuban_ [~ketsuban@] has joined #openttd
19:52-!-Ketsuban [~ketsuban@] has quit [Read error: Connection reset by peer]
20:26-!-Hiddenfunstuff [] has quit [Quit: HydraIRC -> <- It'll be on slashdot one day...]
20:26-!-Ketsuban_ is now known as Ketsuban
20:58-!-glevans2 [~glevans2@] has quit [Quit: Leaving.]
20:59-!-Ketsuban [~ketsuban@] has quit [Quit: To robbery, slaughter, plunder they give the lying name of empire; they make a desert and call it peace.]
21:27-!-liq3 [] has quit []
21:33-!-glx [] has quit [Quit: Bye]
21:34-!-Ketsuban [~ketsuban@] has joined #openttd
21:34-!-JezK [~jez@2407:7800:400:107f:3db5:daca:8457:e66a] has joined #openttd
21:46-!-FLHerne_ [] has joined #openttd
21:48-!-smoke_fumus [~smoke_fum@] has quit [Quit: KVIrc 4.2.0 Equilibrium]
21:50-!-FLHerne [] has quit [Ping timeout: 480 seconds]
21:53-!-glevans2 [~glevans2@] has joined #openttd
22:20-!-Pikka [] has joined #openttd
22:27<kamnet>Good morning Pikka
22:28<Pikka>good afternoon kamnet
22:37<kamnet>how goes it in your part of the world?
22:39-!-Pokka [] has joined #openttd
22:39<Pokka>It goes alright :)
22:45-!-Pikka [] has quit [Ping timeout: 480 seconds]
22:48-!-Biolunar [] has joined #openttd
22:55-!-Biolunar_ [] has quit [Ping timeout: 480 seconds]
23:03-!-Myhorta[1] [] has quit [Ping timeout: 480 seconds]
23:18-!-CompuDesktop [~quassel@2604:6000:120a:a013:a8a5:633d:8dc7:9664] has joined #openttd
23:21-!-Compu is now known as Guest116
23:21-!-CompuDesktop is now known as Compu
23:25-!-Guest116 [~quassel@2604:6000:120a:a013:e9f3:630c:2692:e3e0] has quit [Ping timeout: 480 seconds]
23:33-!-Pokka is now known as Pikka
23:50-!-DDR [] has joined #openttd
---Logclosed Mon Jun 01 00:00:10 2015