IRC logs for #openttd on OFTC at 2016-01-26
⏴ go to previous day
00:29:01 *** Wormnest has joined #openttd
01:21:24 *** peter1138 has joined #openttd
01:21:24 *** ChanServ sets mode: +o peter1138
02:14:28 *** Hiddenfunstuff has quit IRC
04:49:52 *** Biolunar has joined #openttd
05:52:57 *** tt_johannes has joined #openttd
06:09:10 *** tokai|noir has joined #openttd
06:09:10 *** ChanServ sets mode: +v tokai|noir
06:45:50 *** mescalito has joined #openttd
06:56:16 *** Eddi|zuHause has joined #openttd
08:15:10 *** oskari89 has joined #openttd
09:59:56 *** Myhorta has joined #openttd
10:30:36 *** tycoondemon has joined #openttd
10:43:16 *** Hiddenfunstuff has joined #openttd
10:44:03 *** glevans2 has joined #openttd
11:04:13 *** Conductor_Cat has joined #openttd
11:27:53 *** smoke_fumus has joined #openttd
11:29:15 *** sla_ro|master has joined #openttd
11:43:50 *** Clockworker has joined #openttd
12:01:35 *** dihedral has joined #openttd
12:40:59 <Wolf01> where's the cat person?
12:46:40 <Wolf01> „Confidential Technic Ultramotive“ (42056) £178,99 £1249,99- neuer Sportwagen ähnlich einem McLaren (the first price has strikethrough) but I think they put a wrong "1" there
13:31:33 *** Supercheese has joined #openttd
13:35:02 *** Clockworker_ has joined #openttd
14:12:43 <argoneus> good morning train friends
14:45:41 *** Clockworker has joined #openttd
14:45:44 <V453000> this just in: forum moderators to prove that they are assholes yet again \o/
14:46:12 <V453000> why even interfere in a thread where somebody is trying to do something constructive
14:47:07 <V453000> and I wouldn't call asking for constructive feedback rude at all
14:57:35 *** Myhorta has joined #openttd
14:59:08 <Wolf01> bah, that's the artist's fault too, I can like one thing and I'm free to not explain why, maybe I can like it because of colour while the artist drew it so because it shows a "story" behind it
14:59:58 <Wolf01> (I know because I had to do with a graphics designer for 4 years)
15:03:01 <V453000> yes but in terms of actually bringing it anywhere, derailing the discussion is not great
15:03:51 <V453000> and not explaining why I don't like something is just plain not useful
15:04:02 <V453000> the only thing he is asking for is a fruitful discussion
15:04:30 <V453000> sure because you are used to it
15:04:57 <V453000> just like me, I don't mind it ... mostly because I am used to seeing it for years now
15:05:11 <Wolf01> I like Hydros' logo, but I don't think it's suited for OTTD
15:05:18 <V453000> but if you try to be unbiased and decide if it is actually Really better than his suggestions, ...
15:06:03 <V453000> sure if you describe OpenTTD as amateurish and lame, just recycling ideas from TTD, then the logo fits perfectly Wolf01
15:06:22 <V453000> because that is what it is, very basic logo with no real thought in it, just copypasting the shape of TTD logo
15:06:56 <Wolf01> because OTTD is not a TT C&P?
15:07:10 <peter1138> nothing wrong with a basic logo
15:07:16 <V453000> exactly, then why should hte logo be C&P
15:07:33 <V453000> well one thing is basic and one thing is amateurish
15:07:45 <V453000> a logo can have little content and be super well thought through
15:08:30 <Wolf01> it's like what they did with the IntelliJ softwares, every year they changed the logo, the first ones were almost the same and recognisable, now the PHPStorm one looks a Photoshop one and it looks like a full rebrand of the suite
15:09:56 <V453000> what do you suggest then?
15:10:16 <Wolf01> For example I like more the Chrill's ones, I'm used to associate OTTD with the diamond shape
15:10:28 <V453000> ok peter1138 then perhaps we can build upon that and improve it?
15:10:34 <V453000> instead of "I don't like X" ?
15:10:52 <peter1138> what needs improving?
15:11:53 <Wolf01> also TT is not all about trains, why not shape the logo like a ship?
15:12:28 <argoneus> >TT is not all about trains
15:13:10 <Wolf01> or maybe I'm not aware that the acronym changed to Open Train Transport Deluxe?
15:13:35 <Wolf01> which might solve some copyright problems
15:14:04 <V453000> acronym and meaning of words is one thing, what people actually do in the game is another
15:14:15 <argoneus> Only Train Transport Done
15:15:38 <peter1138> trying to make a train shape out of the letters "ottd" is first mistake
15:16:20 <peter1138> well, assuming the current logo is needs replacing is the first one :p
15:17:13 <Wolf01> I'm the opinion that logo and name should be two distinct figures, the name should be well readable and the logo could call some elements of the name, but not with the shape of the name
15:17:31 <argoneus> the most important part of a logo is that it's simple and recognizable
15:17:33 <argoneus> and should be easy to draw
15:17:36 <Wolf01> yes, that's my opinion
15:17:46 <argoneus> if you stylize the logo with trains and ships
15:17:49 <argoneus> it will be too cluttered
15:18:05 <peter1138> yeah where's the buses and lorries?
15:18:08 <Wolf01> for example death metal band logos are a bad example
15:18:18 <argoneus> peter1138: planes are banned
15:20:04 <Wolf01> you could stylize the logo with ships, aircrafts and lorries for a splash screen
15:20:28 <Wolf01> you have plenty of space
15:20:47 <Wolf01> (I still want ekranoplanes)
15:21:42 <Eddi|zuHause> but ekranoplans are essentially ships?
15:21:57 <Wolf01> yes, but ships are too slow in the game
15:26:22 <Wolf01> let's get a distinctive tract of the game and make a logo from it, for example a depot with a road and a rail crossed behind it and OTTD printed on the side, it's better than a logo which doesn't mean anything until you tell me that O-||-D means OTTD
15:26:43 <Wolf01> (and I have bad taste for sure, but at least I could recognise it)
15:31:40 <Eddi|zuHause> that's pretty much my main point as well...
15:31:58 <Eddi|zuHause> the rotated Ts are just a tiny bit "too clever"
15:35:33 <V453000> even when OPENTTD is written under it?
15:35:48 <V453000> where is this world going to
15:41:06 <peter1138> if you write it under it
15:44:29 <Eddi|zuHause> V453000: well, but the point is, if you cannot clearly read the logo, it's not better than any random shape
15:44:59 <Eddi|zuHause> and then we might as well keep a random orange rotated square
15:45:17 <V453000> hence the text to explain the shape?
15:46:09 <Eddi|zuHause> the random orange square also has the text to explain it.
15:46:32 <V453000> yes except it has nothing interesting in it and looks bad
15:47:13 <Eddi|zuHause> but... transition costs are huge. and none of the proposed logos are better by that much of a margin to justify it...
15:47:16 <V453000> a modernized logo would be a lot better, especially in making the project look like a serious thing and bring new players
15:47:54 <V453000> well neither of them probably is final, they are suggestions on which direction to work further
15:47:57 <V453000> refining is the last part
15:48:30 <Eddi|zuHause> uhm, yes. but i don't see how "refining" will do much to that margin
15:50:08 <V453000> well at a point when some direction where to go is decided, then none of the results will be worse than the original
15:55:47 <Eddi|zuHause> well, my point was that it doesn't suffice that it is better than the old one. it must be better by a LARGE margin
16:23:42 <Rubidium> V453000: "proper" 3D is probably even better to attract people
16:24:35 <V453000> proper anything is the best way to attract people but we should probably remain in reality, aka areas that some people are putting effort into
16:25:32 <V453000> if anybody starts coding 3D OpenTTD, I will happily support graphics
16:25:54 <Rubidium> in any case, I think we can say OpenTTD is currently dead or maybe dormant at best
16:26:19 <V453000> :d ok, and it will wake up when/how? :D
16:26:27 <Rubidium> nobody seems to be interested in resolving issues, only some are interested in adding new features
16:26:44 <Rubidium> V453000: at Sankt Nimmerlein with a big bang
16:26:54 <V453000> honestly, a new project sounds to me like the best solution lately
16:28:03 <Rubidium> but a new project is, by definition not OpenTTD anymore
16:29:09 <Rubidium> ergo, a new project is no solution to revive OpenTTD
16:30:55 <V453000> if it is basically a successor, as in heavy inspired but building upon it, it indeed is not reviving OpenTTD, but I believe many people would be interested in it
16:31:02 <V453000> the biggest shame is the newgrf content
16:31:54 <V453000> perhaps it is possible to make a game which is 3D but has locked isometric camera and possibility to load OpenTTD's sprites with offsets instead of the 3D objects
16:32:00 <V453000> does not sound like a good idea though
16:36:21 *** Alberth has joined #openttd
16:36:22 *** ChanServ sets mode: +o Alberth
16:36:46 <argoneus> add a pvp mode to openttd and fund tournaments
16:37:17 <Eddi|zuHause> is there a "Sankt Nimmerlein" equivalent in other languages?
16:37:26 *** tt_johannes has joined #openttd
16:38:49 <V453000> regardless, if there is something to sprinkle new life and new interest in new features, then I believe changing visuals of logo/stuff, especially since there is a chance to do so - properly - is a good move
16:38:57 <V453000> but yes we can just sit around and declare dead
16:39:14 <V453000> & wait for mr Nimmerlein, whoever the fuck that is :>
16:39:30 <Alberth> Eddi|zuHause: when hell freezes over ?
16:40:06 <Eddi|zuHause> Alberth: well, that means basically the same thing, but it's not really "equivalent"
16:40:37 <argoneus> I don't see how changing a logo would sprinkle new life
16:40:51 <V453000> certainly more than waiting for miracle
16:41:02 <argoneus> you could go the factorio way
16:41:08 <argoneus> and make weekly posts on the front page about what's new
16:41:11 <argoneus> and what you need help with
16:41:35 <argoneus> that makes users feel involved in development
16:41:38 <Eddi|zuHause> argoneus: that kinda only works if there's actually something new happening every week
16:41:58 <V453000> as Eddi says XD also, I don't need users to feel involved, I need people to BE involved XD
16:42:29 <argoneus> I personally don't really have motivation
16:42:40 <V453000> revitalizing visuals can either bring new players, in general get interest, or even bring back old dogs / current dogs can feel like shit is happening and start having ideas again
16:42:41 <argoneus> even if I made a cool AI or patched something (which I can't do either lmao), no one would probably care
16:42:45 <argoneus> might as well use my resources elsewhere
16:43:07 <argoneus> maybe there's people who share my opinion
16:43:31 <V453000> anybody who is lazy shares your opinion, they just have a different excuse/justification :)
16:43:42 <argoneus> I'm programming an AI for brood war right now
16:43:47 <argoneus> that thing isn't, you know, dead
16:44:06 <V453000> brood war has not died yet? :D
16:44:15 <V453000> who plays BW against AI though?
16:44:25 <argoneus> it's AIs competing amongst eachother
16:44:28 <argoneus> it's student competitions
16:44:32 <V453000> :DDDDDDDDDDDDDDDDDDDDDDDDD
16:44:36 <argoneus> there's actually motivation to write the thing
16:44:46 <V453000> well you could do that in any game probably
16:45:30 <argoneus> it's quite prestigious actually
16:46:18 *** sim-al2 has joined #openttd
16:46:22 <argoneus> and when there's things like this
16:46:25 <V453000> I bet the AI would beat me any day XD
16:46:26 <argoneus> why would I write things for openttd
16:46:30 <argoneus> you get my point, right
16:46:40 <argoneus> about the motivation thingy
16:47:23 <V453000> sooo let's make OpenTTD2 with modern engine and 3D that people care about? :P
16:47:42 <argoneus> I think openttd itself could be a great platform
16:48:06 <argoneus> imo better multiplayer/scripting could be nice
16:48:13 <argoneus> then again I don't participate in it too much so I can't tell
16:50:05 <V453000> I kind of want 3D a lot
16:50:15 <argoneus> what was wrong with train fever?
16:50:25 <V453000> idk I haven't played it
16:50:39 <V453000> am guessing no game logic like OpenTTD has (the signals and stuff)
16:51:29 <V453000> realistic shit gets boring relatively quickly due to bad replayability if there aren't interesting mechanics
16:54:37 <Eddi|zuHause> train fever has a bit of annoying UI
16:56:06 *** Wormnest has joined #openttd
16:56:06 <peter1138> clearly the best way to start "openttd 2" is with the logo
16:56:21 <Eddi|zuHause> peter1138: and the name :p
16:56:50 <V453000> revitalizing openttd and starting a new project are two different things
16:59:33 <_dp_> hi, openttd2 idea is in the air lately :)
17:00:16 <Eddi|zuHause> as far as i'm concerned, openttd2 idea always was tucked firmly on the ground, not moving anywhere...
17:00:22 <peter1138> step 2: reverse engineer train fever...
17:00:28 <argoneus> is that a browser game _dp_
17:01:00 <_dp_> I would'n call it a game yet... but, yes, it's browser)
17:01:01 <Eddi|zuHause> _dp_: that looks like a bad version of a model railway planner
17:01:09 <_dp_> and, furthermore multiplayer))
17:01:14 <argoneus> buy the oculus devkit
17:01:24 <argoneus> as one of the first VR games
17:04:53 <Sacro> With my playstation mouse
17:06:16 <_dp_> vr for smth that isn't first-person... sounds weird to me
17:07:18 <V453000> 3D lets you get in the trains :P
17:07:22 <peter1138> _dp_, these people would sit in the drivers seat of a train...
17:09:35 <_dp_> and build rails by driving tracklayer around?)
17:12:01 <_dp_> my main concern with 3d is that, as I understand, it takes enourmous amount of work to get it look nice and work fast.
17:14:03 <argoneus> you just use an engine
17:14:22 <argoneus> it makes no sense to roll your own engine in a very small team
17:14:57 <_dp_> existing engined dont' seem to fit well for open world games
17:15:01 <peter1138> most engines have really nasty constraints on how the rest of it works
17:15:18 <argoneus> unity works fine for cities skylines
17:15:24 <_dp_> minecraft uses it's own for example
17:15:25 <argoneus> im sure you can bend it
17:15:32 <argoneus> minecraft is a great example of fantastic coding
17:15:36 <argoneus> the game runs flawlessly
17:16:12 <peter1138> there's like some huge level of sarcasm here
17:17:32 <argoneus> minecraft is such a great game
17:17:39 <argoneus> people blame java for performance issues ever since it came out
17:17:49 <argoneus> "no wonder minecraft runs slow, it's java!"
17:18:04 <argoneus> "no wonder minecraft runs slow, it's notch!"
17:19:02 <peter1138> it runs a lot better since he left
17:20:43 <argoneus> cities skylines runs on unity
17:20:45 <argoneus> and it seems to work nice
17:20:57 <argoneus> I don't know how many corners they had to cut but all the traffic going around looks nice
17:21:05 <argoneus> and it runs nice too
17:22:05 <peter1138> now mix unity with gpl
17:23:17 <peter1138> what is unity? a 3d engine or a development environment?
17:23:55 <_dp_> it's like flash but 3d as I understand it
17:26:20 <Eddi|zuHause> unity seems to have trouble on linux and amd graphics
17:26:41 <Eddi|zuHause> at least all unity games i've seen so far
17:33:03 <peter1138> distributed processing
17:33:57 <peter1138> most important feature
17:36:50 <Eddi|zuHause> i can totally see the use for an undo knob, but i see no sensible implementation
17:38:53 <Alberth> just a knob would be easy :)
17:39:41 <Alberth> maybe it should reload the last save?
17:40:03 <Eddi|zuHause> Alberth: that would be called "quickload"
17:40:26 <peter1138> i guess my phone's clock is fast
17:40:36 <Eddi|zuHause> 12h clock is weird...
17:40:54 <Alberth> yeah, don't check the time in the morning
17:40:56 <Milek7> there should be a local build mode
17:41:22 <Milek7> and only after commiting changes should be send to server
17:41:32 <Milek7> then it will be easy to add undo
17:41:52 <peter1138> network protocol: git
17:42:03 <Eddi|zuHause> Milek7: that's called a "planning mode"
17:42:37 <Eddi|zuHause> (and is one of Train Fever's biggest missing features)
17:44:12 <Milek7> they are adding new features to openttd or only bugfixes?
17:45:45 <Eddi|zuHause> if you build it, they will come.
18:03:49 *** TheMask96 has joined #openttd
18:36:01 *** Progman has joined #openttd
18:36:49 <Eddi|zuHause> somehow the "Locating Jebediah Kerman" loading text almost sounds like "The little Jeb wants to be picked up from the ball pit" to me ;)
18:38:23 *** Flygon_ has joined #openttd
18:38:47 *** Eddi|zuHause2 has joined #openttd
18:38:58 *** ConductingCat has joined #openttd
18:39:02 *** smoke_fumus|2 has joined #openttd
18:40:19 *** Belugas has joined #openttd
18:40:19 *** ChanServ sets mode: +o Belugas
18:46:37 *** andythenorth has joined #openttd
18:53:45 *** Eddi|zuHause2 is now known as Eddi|zuHause
19:03:09 <andythenorth> Wolf01: pictures + video eh?
19:09:56 *** frosch123 has joined #openttd
19:20:57 *** Clockworker_ has joined #openttd
19:30:28 *** sim-al2 has joined #openttd
19:32:33 * Eddi|zuHause sorting through KSP mods to see which ones impact performance
19:35:03 *** HerzogDeXtEr has joined #openttd
19:44:49 *** HerzogDeXtEr1 has joined #openttd
19:45:43 <DorpsGek> Commit by translators :: r27505 trunk/src/lang/portuguese.txt (2016-01-26 19:45:36 +0100 )
19:45:46 <DorpsGek> portuguese: 2 changes by neuralshock
19:56:11 <dpk> i was wondering what the motive was behind the cargo name change from 'Petrol' to 'Petroleum Fuel' in FIRS
19:56:29 <dpk> and whether that needs to be reflected in the German translation, or if it can stay as 'Treibstoff'
19:57:26 <andythenorth> it’s a question for translators
19:57:47 <andythenorth> I changed, because in UK English, it’s weird to burn petrol in a cement kiln
19:57:52 <andythenorth> or use it in heavy equipment
19:58:14 <dpk> ah, i see, that's what i wondered
19:58:25 <dpk> if it was because of confusion that's specific to English
19:58:45 <dpk> but i think Treibstoff is perfectly okay for that (it just means 'fuel')
19:58:45 <andythenorth> if google translate can be trusted, Treibstoff is highly appropriate (fuel)
19:58:59 <andythenorth> I would have used ‘fuel’ but in the uk that also covers coal, wood etc
19:59:46 <dpk> possible error in the base string: in STR_EXTRA_CEMENT_PLANT_PETROL_SUBSTR should 'petroleum fuels' be 'petroleum fuel'? seems odd to use the plural there, unless there are multiple sorts of fuel
19:59:52 <dpk> (that's just the first one i encountered)
20:01:26 <Eddi|zuHause> dpk: could be "Heizöl"
20:03:36 * andythenorth might have made a boo
20:04:17 <andythenorth> nah, I’ve used the plural deliberately
20:04:21 <andythenorth> quirk of English, sorry
20:04:35 <andythenorth> there are multiple valid fuels
20:07:11 <dpk> so there are multiple kinds of fuel, okay
20:07:28 <andythenorth> petrol, diesel, heavy oil etc :)
20:07:29 <dpk> well, multiple kinds covered by that one cargo type
20:17:20 <dpk> "Edible Oil" is a generalization of olive oil, vegetable oil, sunflower oil, etc?
20:21:18 <dpk> '8 tonnes' should be '8t' in STR_EXTRA_FOOD_PROCESSING_PLANT_BEANS_SUBSTR, no? so it's general for both imperial and metric tons
20:22:07 <dpk> or maybe not, since STR_EXTRA_FOOD_PROCESSING_PLANT_NUTS_SUBSTR is the same. hadn't noticed that
20:23:11 <andythenorth> you make a good point
20:26:12 <andythenorth> pushed, might not show up until tomorrow
20:26:18 <andythenorth> should be ’t’ in all of those cases
20:30:43 <Benny> is there a reason dragging signals on diagonals places signals one tile off?
20:32:27 <dpk> Benny: because each tile of diagonal track is less distance than a tile of straight track
20:33:17 <Eddi|zuHause> Benny: diagonals count as half a tile
20:33:53 <dpk> hmm, how do i get the web translator to accept no change to the string to stop it listing it as outdated?
20:34:21 <dpk> since i already used 't' everywhere
20:34:46 <Eddi|zuHause> Benny: can you be more specific?
20:35:01 <Benny> when i drag signals with a density of 2, which is 1 space between each signal
20:35:05 <Benny> i get 1 tile on straights
20:35:14 <Benny> and 1.5 tiles on diagonals
20:35:55 <Eddi|zuHause> no, since diagonals are half a tile, every 4th diagonal tile means a density of two
20:36:25 <Benny> right so it is a coding issue
20:36:36 <Benny> because it counts every Xth tile and not spacing between them
20:36:40 <Eddi|zuHause> i think it's just a misunderstanding
20:36:52 <Benny> well it seems very wrong
20:36:59 <Benny> with fast accelerating trains it causes slowdowns
20:37:10 <Benny> and trains dont fit between signals properly on diagonals
20:37:33 <Eddi|zuHause> well, yes, it counts distance between the faces of the signal
20:37:41 <andythenorth> dpk: try putting in an arbitrary space?
20:37:45 <andythenorth> might trick eints
20:37:53 <andythenorth> or translate it wrong, then immediately translate it back
20:38:07 <Eddi|zuHause> yeah, trains on diagonals behave a bit odd. but that has nothing to do with the signal spacing
20:38:47 <Benny> i think its the sprites that are weird mostly on diagonals
20:38:49 *** oskari89 has joined #openttd
20:38:59 <Benny> because a 1 tile train fits on 1 tile on a straight and 2 tiles on diagonals
20:39:50 <Eddi|zuHause> yes, some parts of the code try to correct for a more euclidean factor of 0.7, but that is not consistent
20:40:39 <Eddi|zuHause> other parts have a factor of 0.75
20:41:03 <tt_johannes> if you look at an openttd map, it's a square, however, with one corner on top. If you'd have to turn it such that there are two corners on top, i.e. 45 degrees left or right, in which way would you preferably turn it?
20:41:06 <Eddi|zuHause> and the display stays at a factor of 1
20:41:23 <tt_johannes> what sounds more familiar?
20:41:36 <Eddi|zuHause> tt_johannes: (0,0) of the image becomes the north corner of the map
20:42:16 <tt_johannes> Eddi|zuHause: what is 0,0 for you? is it the top left corner of coordinate system?
20:42:27 <Eddi|zuHause> so in preparation you should rotate the heightmap 45° counter clockwise to compensate
20:42:38 <Benny> this might be but i dont know how it connects to signals being 0.5 tile off
20:42:47 <Eddi|zuHause> yes, with images, (0,0) is usually top left
20:43:14 <Eddi|zuHause> Benny: it's not off. your imagination is off :p
20:44:56 <Benny> i think its a matter of not wanting to place signals on "half" tiles
20:45:10 <Benny> by some developer or maybe all of them. or maybe im stupid and missing something very obvious
20:45:47 <Benny> or well, signals taking up 0.5 tile and not 1 like on straight
20:46:08 <Eddi|zuHause> Benny: it's not very obvious, as diagonals are handled very inconsistently in openttd.
20:48:40 *** glx is now known as Guest1062
20:51:15 <dpk> andythenorth: i tried space. will use the "insert junk" approach instead
20:55:08 <dpk> what's the prototype for the 'Liquids Terminal'? i see it takes both oil and edible oil
20:58:03 <andythenorth> there are generic ’bulk terminals’, or subdivided into ‘dry bulk’ and ‘liquid bulk’
20:58:06 <andythenorth> depends on size of port
21:05:00 <Milek7> is there an option to disable to local authorities?
21:05:30 <Eddi|zuHause> Milek7: no, but the magic bulldozer will ignore them
21:06:48 <dpk> Eddi|zuHause: any ideas for a translation for that? 'Hafen für flüssige Massengüter' is quite long, and doesn't match the pattern of industries having simple nouns as names
21:07:14 <dpk> especially since, won't they end up getting names like "Diddlington Hafen für flüssige Massengüter"?
21:07:20 <Milek7> afaik there is no option to enable cheats on multiplayer
21:07:36 <andythenorth> dpk: any literal translation, like ‘liquids port’?
21:07:39 <Milek7> and even if, this would been an overkill ;)
21:08:09 <dpk> i considered Flüssigkeitshafen, googlecount effectively 1 (actually 4 but it's the same article four times over)
21:08:41 <Eddi|zuHause> that would be the closest i would come up with, but it sounds very odd...
21:08:56 <_dp_> hm, having cheats in mp might be good idea
21:09:30 <frosch123> dpk: "Flüssiggüter", not "Flüssigkeiten"
21:10:12 <andythenorth> what does Tanklager-Gesellschaft translate as?
21:10:18 <andythenorth> presumably a company name?
21:10:36 <frosch123> andythenorth: liquid stroage company
21:11:00 <Eddi|zuHause> well, "Tanklager" could work... it wouldn't have "port" in it, though
21:11:20 <Eddi|zuHause> but "Tanklagerhafen" sounds weird as well
21:11:29 *** Wormnest has joined #openttd
21:12:05 <andythenorth> Seehafentanklager
21:14:23 <frosch123> dpk: i vote for "Flüssiggut Verladehafen"
21:14:24 <_dp_> oh, cheat flags are stored in savegame, awesome ^^
21:14:52 <dpk> 'Flüssigguthafen' has a few refults (Wikipedia, some document from a local branch of the SPD (a German political party), a business and economics book)
21:15:01 <frosch123> _dp_: one or two of the cheats work fine in mp, after activating in sp
21:15:03 <dpk> but also suffers from the "sounds weird" probelm
21:15:31 <frosch123> dpk: well, the "bulk terminal" was translated as "Überseehafen"
21:15:49 <frosch123> so, maybe you can find another arbitrary synonym :p
21:16:19 <_dp_> what I'm looking for is a way to let server cheat actually)
21:16:41 <Eddi|zuHause> _dp_: GameScript.
21:16:46 <dpk> the server can do what it likes with game scripts, as i understand it
21:17:08 <frosch123> _dp_: easy, just cheat, and then kick all clients due to desync
21:17:27 <frosch123> then blame it onto someone else
21:18:42 <_dp_> frosch being helpful as usual :p
21:19:30 <Eddi|zuHause> on the "usefulness" scale, frosch123 ranks barely over V453000 :p
21:19:43 <dpk> i think i'll go for 'Flüssigguthafen'
21:20:01 <_dp_> but this whole scheme sounds as it may work...
21:20:10 <dpk> and with that, FIRS de-DE is up-to-date
21:20:10 <frosch123> Eddi|zuHause: when you say that, i am unsure about the orientation of your axes
21:20:25 <_dp_> really miss magic bulldozer for server
21:21:42 <andythenorth> bloody Iron Horse incremental compile
21:21:46 <andythenorth> it saves loads of time
21:21:49 <andythenorth> until it goes wrong
21:22:08 <Eddi|zuHause> frosch123: well, V453000 and i are usually on diametrically opposite opinions :p
21:22:14 <andythenorth> then I have to deal with multiprocessing being unable to pickle errors
21:22:20 <Eddi|zuHause> frosch123: so anybody else must be closer :p
21:22:23 <andythenorth> and then my json file for tracking deps breaks
21:22:31 <dpk> andythenorth: considered using make?
21:22:43 <andythenorth> it does use make :)
21:22:50 <andythenorth> for some value of ‘use'
21:24:58 <frosch123> all:\n\tpython3 mybuildscript.py
21:25:23 <dpk> well, maybe replace multiprocessing with -j 8 and your deps json file with a deps graph in your Makefile, i mean
21:26:09 <andythenorth> yeah, I could learn how to do that
21:26:12 <dpk> (replace 8 with your preferred number of processes)
21:26:48 <andythenorth> I think I’d need to change a few other things
21:27:06 <frosch123> dpk: problem is, compilation time takes "start up time" + "source size"*"compilation speed"
21:27:32 * andythenorth wonders why Iron Horse is _so_ slow for nmlc
21:27:33 <dpk> you need a process script that works on one target file at once
21:27:57 <andythenorth> FIRS is fast now, or fast enough
21:27:57 <frosch123> dpk: there was a version of iron horse, where the source was split in 100 parts, and "start up time" dominated over everything
21:28:05 <andythenorth> still that way now
21:28:09 <Eddi|zuHause> andythenorth: same reason why CETS is so slow? sheer size of the code?
21:28:24 <andythenorth> the incremental compile is cost tradeoff
21:28:31 <andythenorth> it’s much slow when everything needs compiled
21:28:45 <andythenorth> usually only sprites or such changed
21:28:56 <andythenorth> but the engineering of it is daft
21:29:02 <andythenorth> result is good, but engineering not
21:29:27 <dpk> i assume by "incremental compile" you mean you produce a load of subsidiary files then combine them into one at the end or something? so you're not just producing "1 GRF file", you're producing n intermediate files which are the dependencies for producing the 1 GRF file at the very end
21:30:20 <andythenorth> the multiprocessing use is incidental
21:30:35 <andythenorth> because they’re split to 1 file per vehicle, I can also run them in parallel
21:30:46 <andythenorth> same as make would with -j
21:32:15 <frosch123> when you change 20 files, you can try to compile them in two sets of 10
21:32:24 <frosch123> and hope that the enxt time you only need to recompile 10 of them
21:32:39 <frosch123> i.e. don't split stuff into 100 compiles, but only 10
21:32:50 <frosch123> with sometimes changing the composition
21:33:02 <andythenorth> I wonder about doing that by splitting apart the rosters in Iron Horse :P
21:33:06 <andythenorth> maybe to separate grfs
21:33:15 <andythenorth> but I hate the separate grfs idea :P
21:33:18 <andythenorth> we need grfmerge :P
21:34:08 <frosch123> and you do not want to tuse that name
21:34:20 <frosch123> though maybe it was just grfmrg
21:34:27 <frosch123> because 8 letters are too much
21:34:29 <andythenorth> someone made a grf-smusher?
21:34:49 <frosch123> andythenorth: before newgrf, people modified the baseset
21:35:06 <frosch123> but since there is only one baseset
21:35:19 <frosch123> you needed grfdiff and grfmrg to extract diffs between basesets
21:35:22 <frosch123> and merge them into another
21:35:50 <dpk> frosch123: that's interesting. if you set it up with make to build in parallel, the startup time of any individual build script (assuming they're written in Python and C) should be negligible
21:36:09 <andythenorth> you’d think, but frosch123 timed it
21:36:12 <frosch123> it's basically an actionA, but offline
21:36:39 <frosch123> dpk: nml has a start time of 0.9 seconds (on my machine back then)
21:36:58 <andythenorth> and I start it maybe 70 times or so
21:37:02 <frosch123> when the sources are split so much, that each file then takes 0.1 seconds to compile, you have 1 second per file
21:37:07 <dpk> starting processes is fast, especially on Linux, and Python uses memory-mapped files for its bytecode so modules will be shared between processes
21:37:28 <frosch123> a compete compile then costs n seconds cpu time
21:37:56 <frosch123> while as a single file it only costs (0.9+n*0.1) seconds cpu time
21:38:12 <Eddi|zuHause> maybe we should use CETS as a performance test, with one huge file, individual files, and clustered files
21:38:29 <frosch123> if you split often enough it will takes ages on any amount of cores
21:38:33 <andythenorth> 85s to compile the set with a multiprocessing pool of 16 workers
21:38:51 <dpk> sure, 85s build time from cold seems reasonable
21:38:54 <Eddi|zuHause> frosch123: i don't think n*0.1 is the right formula
21:39:22 <frosch123> dpk: same in c++, processing of header files takes so long, that a complete compilation of many cpp files takes longer, than pasting them all together
21:39:27 <dpk> if you're doing incremental compilation, then, it should only take 1s or 2s or so to rebuild after one file changes
21:39:39 <andythenorth> depends how many vehicles are in the file, but yes
21:39:41 <andythenorth> that’s about righ
21:39:57 <frosch123> dpk: yes, andy's incremental compiles were fast, but complete compiles were slow
21:40:00 <Eddi|zuHause> well, it wasn't in the case of CETS
21:40:10 <andythenorth> I just recompiled 8 vehicles in 7s
21:40:19 <Eddi|zuHause> CETS with 1000 engines didn't take 1000*0.1s
21:40:22 <dpk> well, how often do you need to do a complete compile?
21:40:30 <andythenorth> only if I change the classes
21:40:33 <andythenorth> or when I screw up :P
21:40:34 <Eddi|zuHause> dpk: often enough that it gets annoying
21:40:48 <andythenorth> when I screw up, the JSON breaks, then the incremental compile fails
21:41:02 <andythenorth> I should put guards on the thing I screw up, it’s always the same places
21:41:10 <andythenorth> or I should stop the JSON breaking
21:41:13 <dpk> right, so rebuilds should also be atomic
21:41:31 <dpk> so that if they break, you still have the old build to work from as a base for the next incremental rebuild
21:41:52 <frosch123> dpk: at some point the conclusion was to change to compiler, so it could compile multiple files independently, but with only one start-up time
21:42:04 <frosch123> but noone finished that :)
21:42:36 <andythenorth> it would be very nice to have :)
21:43:05 <andythenorth> sometimes I think it would be nice to just pass it the python classes I have :P
21:43:11 <andythenorth> but I’m not smart enough
21:43:52 <dpk> what happened to CETS? i'd really like a train set with lots of historical German trains. DBSet is nice but only starts in 1920
21:44:13 <Eddi|zuHause> lack of motivation happened
21:45:01 <andythenorth> also lack of graphics
21:45:54 <frosch123> eddi had no danmack, v or pikka
21:46:00 * andythenorth has made very ugly vehicle cargo graphics for phosphate
21:46:29 <andythenorth> everyone needs a danmack or pikka
21:46:31 <Alberth> rebuilding nml is worth the trouble?
21:46:33 <Eddi|zuHause> i had a oberhümer...
21:46:39 <andythenorth> Alberth: honestly, not
21:47:11 <andythenorth> being able to link files somehow would be nice :P
21:47:20 <andythenorth> but not essential
21:47:44 <andythenorth> Eddi|zuHause had a redneck patch for including comments, so that reliable splits/joins could be done :P
21:48:09 <andythenorth> then use grfcodec for speed
21:50:10 <Eddi|zuHause> that is basically a giant hack for circumventing an nmlc problem with large source files
21:50:46 <Eddi|zuHause> (where "large" is around 100kloc)
21:51:39 <andythenorth> if I was picking things, a station GUI that worked on 2x UI zoom is much preferable to faster nmlc :P
21:53:45 <frosch123> eints got a lot better due to python 3.2 -> 3.4
21:53:50 <frosch123> did anyone try nml since then? :p
21:54:37 <Eddi|zuHause> i certainly didn't
21:54:50 <Eddi|zuHause> my "full CETS" tests were still with python2
21:55:00 <andythenorth> I can’t upgrade to 3.4, my python build is borked because [setuptools again]
21:55:05 <Milek7> current is Python 3.5.1 :>
21:55:20 <Alberth> python2 is probably still faster
21:55:28 <Milek7> i don't understand why people are using old software
21:55:37 <Eddi|zuHause> i wanted to set up the CETS makefile so you could easily switch between full and split compiles, but i don't think i ever did that
21:56:07 <Alberth> Milek7: you want to be very compatible with others, or you use a computer but you cannot install new software on it
21:56:10 <Eddi|zuHause> Alberth: i don't think it's a speed issue, it was more an "exploding memory usage" issue
21:56:17 <andythenorth> Alberth: both valid here
21:57:02 <Eddi|zuHause> Milek7: sometimes updating to something not within your distribution's repository ecosystem may be too much of a hassle
21:57:05 <andythenorth> currently I have no supported way to get a python build on this mac
21:57:17 <Alberth> Eddi|zuHause: likely, I looked into that, but my attempt exploded in some way, don't remember how exactly
21:58:28 <Milek7> Eddi|zuHause: so i always use rolling distros, because freezing branch model have always ancient packages
21:59:06 <Alberth> Milek7: and as a result, any software that you make cannot be run at all major distributions
21:59:14 <andythenorth> nor in production
21:59:20 <Eddi|zuHause> Alberth: my suspicion is that python's internal string handling was never meant to handle taking apart huge strings, and it tries to store each partial string in memory, in case it comes up again
21:59:25 <andythenorth> nor for development against fixed production KGS
21:59:59 <andythenorth> also “VersionConflict: (setuptools 0.6c12dev-r88846 (/System/Library/Frameworks/Python.framework/Versions/2.6/Extras/lib/python), Requirement.parse('setuptools==1.4.2’))”
22:00:05 <Alberth> Eddi|zuHause: true I think, the problem there is RE, which is not designed to read its output from a stream while processing
22:00:06 <andythenorth> so no python 3.4 for me
22:00:21 <Eddi|zuHause> Milek7: i hate rolling distros, because it's almost impossible to only update one single program. unrolling dependencies will almost always update almost everything
22:00:55 <andythenorth> “I wanted to update libpng, but now my GCC is being rebuilt” :(
22:00:56 <Alberth> Eddi|zuHause: also, nml seems to make zillions of the same object copies, but I cannot see if these can be shared without breaking something
22:01:18 <frosch123> comparing firs: parsing 31s -> 27s, preprocessing 23s -> 12s, writing 7s -> 5s
22:01:30 <andythenorth> frosch123: that’s fast :O
22:01:44 <andythenorth> anyone tried it on pypy?
22:01:56 <Alberth> people are working on speeding up python computations, thus
22:01:57 <frosch123> for firs python 3.4 seems to be a lot faster for the "preprocessing" phase
22:02:15 <frosch123> i have no recorded numbers for iron horse
22:03:00 <frosch123> Eddi|zuHause: eints memory consumption was halfed by python3.2->3.4
22:03:21 <andythenorth> “sudo easy_install setuptools==1.4.2”
22:03:28 <andythenorth> "The 'setuptools==0.6c9' distribution was not found"
22:03:34 <andythenorth> I really really hate python packaging
22:03:47 <andythenorth> it is not done well, and suffers from multiple brown bag releases
22:04:26 <Milek7> some distros have own packages for python modules
22:05:49 <Eddi|zuHause> uhm... so "make" telling me "everything is up to date" is taking 10s :p
22:08:39 <frosch123> ironhorse takes 0.8 seconds for most files
22:08:44 <frosch123> no idea what it was before
22:08:52 <Eddi|zuHause> hm. i don't think i've exhausted my 6 cores in quite a while
22:09:00 <frosch123> some take up to 1.3
22:09:08 <andythenorth> frosch123: my numbers won’t match yours so eh :)
22:09:10 <frosch123> there are some "bigger files"
22:10:06 <frosch123> andythenorth: did i tell you, that i made "make" an alias for "nice make"
22:10:13 <frosch123> specifically due to your grfs :p
22:10:16 <Eddi|zuHause> grfcodec isn't exactly instant either...
22:10:35 <andythenorth> frosch123: what you don’t want your fans to run? It’s good for them
22:10:50 <andythenorth> I thought I fixed the defaults to be more polite
22:11:39 <frosch123> sadly i did not manage to teach the admin at work, to run the virus scanner with nice
22:12:34 <Alberth> Eddi|zuHause: your unix2dos spits out more than just a version number
22:13:02 *** Clockworker has joined #openttd
22:13:29 <Alberth> I put a fix in the template project iirc, but euhm, the old makefile is copied to everything :p
22:13:31 <Eddi|zuHause> anyway, CETS compile with all files split is: real 4m5.529s user 16m55.766s sys 0m52.229s
22:14:20 <Alberth> it has been a while :)
22:14:27 <Alberth> let me see if I can find it again
22:15:05 <Eddi|zuHause> so... where did i put the info what file to merge into an nml file...
22:17:51 <Eddi|zuHause> i'm fairly sure i already have a function that combines multiple vehicles into one file
22:20:39 <Alberth> sorry, can't find it :(
22:29:59 * andythenorth living dangerously with pythons :(
22:40:14 *** Wormnest_ has joined #openttd
22:46:13 <Eddi|zuHause> so, i tried combining files, but something in there broke nforenum, maybe...
22:47:00 <Eddi|zuHause> i combined each "company" into an individual file, and so far it used: real 4m48.448s user 6m11.252s sys 0m17.191s
22:48:32 <Eddi|zuHause> [GRFCODEC] cets.grf
22:48:33 <Eddi|zuHause> Error: Encountered invalid character looking for literal byte.
22:48:35 <Eddi|zuHause> While reading sprite:105733
22:52:52 <Eddi|zuHause> oh, there's a bogus "(" in the line "\2| 7D 81 20(\dxFFFFFFFF "
22:52:57 <Eddi|zuHause> how did that get there?
22:54:51 <Eddi|zuHause> it's not in the original ".onfo" file... how did NFORenum mess that up?
22:55:44 <Eddi|zuHause> (".onfo" is the output of nmlc)
22:58:17 <Eddi|zuHause> uhm... i think i ran into a "disk full" with cets.nfo being around 1GB...
22:58:29 <Eddi|zuHause> how did that happen?
23:02:08 <Eddi|zuHause> hm, i think i messed up the combination of the combined files...
23:05:53 <Eddi|zuHause> needs filtering out duplicates...
23:06:34 <planetmaker> andythenorth, did you try to speak to orudge about the wiki credentials?
23:06:46 <andythenorth> no I should I guess
23:07:40 <Eddi|zuHause> now cets.nfo is back to 10MB :p
23:12:09 <Wolf01> andythenorth, I just returned, the news are really promising, and a bad realisation for my wallet ;)
23:12:22 <andythenorth> you like the pictures?
23:15:06 <Wolf01> yes, I would have preferred a frame with cables and pulleys for the excavator, but everything else seems very nice
23:16:38 <Wolf01> I think I'll purchase at least the volvo, the BWE and the claas, maybe 2 of the claas...
23:17:21 <andythenorth> stacker / reclaimer?
23:17:41 <Wolf01> eh that will be too much :D
23:18:30 <Wolf01> the only thing I didn't understand is that if the BWE has only one L motor
23:30:49 <Wolf01> I was looking on the phone while I was at the pub, now I'm going to look better at the pictures
23:31:35 <Wolf01> yes, XL, still only one... /me plots a full RC conversion
23:32:12 <Eddi|zuHause> sooo... i think that now worked
23:32:26 <Wolf01> also, it's wonderful how it automatically converts dirt to lego pieces in the box image
23:32:43 <Eddi|zuHause> with every company in a file for its own: real 1m51.827s user 4m28.259s sys 0m9.781s
23:38:43 <Eddi|zuHause> @calc (4*60+28)/(1*60+51)
23:38:43 <DorpsGek> Eddi|zuHause: 2.41441441441
23:39:07 <Eddi|zuHause> @calc (16*60+55)/(4*60+5)
23:39:07 <DorpsGek> Eddi|zuHause: 4.14285714286
23:40:29 <Eddi|zuHause> so... assuming that all the multiprocessing goes into nmlc, and all other things are single process (which is not entirely true), one should be able to extrapolate how much of the time is spent in nmlc :p
23:40:58 <Eddi|zuHause> (i have 6 cores)
23:47:42 <Eddi|zuHause> also, the "revision" bit at the start looks wrong...
continue to next day ⏵
