IRC logs for #openttd on OFTC at 2011-02-16
⏴ go to previous day
00:14:19 *** valhallasw has joined #openttd
00:45:14 *** supermop has joined #openttd
00:55:24 *** amkoroew1 has joined #openttd
02:51:54 <supermop> anyone up for a game?
03:22:50 *** Maarten has joined #openttd
03:39:39 *** DanMacK has joined #openttd
03:52:15 <supermop> i have some drawing related questions
04:02:31 <supermop> so i have few non-fully realized ideas floating around, one of which is my shed grf
04:02:55 <supermop> i think i have worked out most of the glitches, but it needs more content to really be complete
04:03:10 <supermop> mainly, actual depots that match the sheds
04:03:28 <DanMacK> I thought you did that for some reason
04:04:05 <supermop> so i drew those, but as far as i can tell, the only way to get the depots in game is via a railtype grf, or a lot of rail grf aware code in the main grf
04:04:56 <supermop> so i am faced with putting together a track grff, and I don't know how to code it to just use my depots with the tracks of other newgrfs already loaded
04:05:19 <supermop> so do you advise drawing up a basic set of rail types,
04:05:34 <supermop> or just making some place holder graphics as proof of concept?
04:05:57 <DanMacK> TBH, this would be better directed at the coders... I'm not familiar with the newgrf structure of newrails
04:06:13 * DanMacK isn't familiar with newgrf period...
04:06:23 <supermop> well, its a struggle for me
04:06:48 <supermop> I had kind of hped that if i drew something I could attract someone to code it
04:06:55 <supermop> but that has yet to happen
04:07:03 <DanMacK> I draw, and I find coders :P
04:07:28 <DanMacK> You seem to be doing a prettty good job on your own
04:07:34 <supermop> at any rate, do you have any tips for drawing rails?
04:09:37 <DanMacK> They can be a real pain to align properly
04:10:24 <DanMacK> Best thing is start with the horizontals. Decode the main grf and go by those rails to get the slopes correct
04:12:35 <supermop> how do you usually draw glass?
04:14:09 <supermop> they feel odd to me on an industrial building
04:14:26 <supermop> where the glass is not meant to be transparaent
04:14:46 <supermop> do you shade the various rail slopes?
04:14:59 <supermop> can't remember if ogfx does
04:15:59 <DanMacK> Well, the dark bluish purples
04:16:09 <DanMacK> re: Shading, I have...
04:18:02 <DanMacK> depends on the set really
04:18:27 <supermop> well, this one is pretty basic, can alway darken it later
04:35:45 <supermop> hmmyeah, the tracks on this slope do not look centered at all
04:41:17 <DanMacK> Takes awhile to do. Keep trying, it'll eventually look right :P
04:41:26 * DanMacK is off for the evening
05:48:38 *** roboboy has joined #openttd
05:56:18 *** Eddi|zuHause has joined #openttd
06:38:59 *** DayDreamer has joined #openttd
07:06:49 *** Br33z4hSlut5 has joined #openttd
07:10:18 *** amkoroew has joined #openttd
07:10:45 *** Kurimus has joined #openttd
07:34:53 *** Cybertinus has joined #openttd
07:55:47 *** perk111 has joined #openttd
08:09:54 *** JVassie has joined #openttd
08:11:20 *** valhallasw has joined #openttd
08:45:14 <James_> I've made a patch that reverts to previous autosaves by guessing the filename, would it be useful to anyone?
08:49:21 <SpComb> did you call it "undo"?
08:50:12 <James_> Is there already an "undo"?
08:52:04 <planetmaker> why do you guess the autosave and not keep it in memory?
08:52:59 <James_> Keeping it in memorhy would be a better idea
08:53:17 <planetmaker> or rather last save instead of last autosave
08:55:14 <James_> The point is to stop people vandalising multiplayer games, so last autosave would be best
08:56:52 <planetmaker> well. Or the last save. Whatever is newer
08:57:25 <planetmaker> autosave is just the convenient way to make sure there is an actual savegame
08:57:34 <Noldo> so a kind of snapshot-rollback thing+
08:58:18 <James_> "revert n" reverts n autosaves, when n=1 it will revert to the start of the current month
08:59:34 <planetmaker> imho better keep a list of filenames of file list entries.
08:59:55 <planetmaker> or people will complain that they saved and that they can't revert to it
09:01:16 <planetmaker> and 'current month' is very relative as it depends on the selected autosave frequency. So be very carful with specifying a game time anywhere near there
09:01:32 <planetmaker> the file stamp also is no indicator as the game could have been paused
09:02:53 <planetmaker> maybe it might be - for this purpose - be sensible to keep a list of "ingame time - real time - filename"
09:04:31 <James_> Then a "listsavegames" command would help the server administrator choose which one to revert to?
09:05:58 <planetmaker> hm, well. Possibly it will be most useful if I - like you probably did - just specify a number which gives me the amount of savegames to go back in time
09:10:20 *** HerzogDeXtEr1 has joined #openttd
09:11:31 <James_> Should the list of savegames be kept in a double-ended queue, so that old savegames can be removed when overwritten?
09:13:05 *** DayDreamer has joined #openttd
09:19:30 *** Devroush has joined #openttd
09:21:15 <planetmaker> well, I've not properly thought about the problem. It probably would suffice to keep a list of similar length as the number of autosaves which shall be kept
09:27:12 <planetmaker> the only thing I was really concerned about when you wrote "guessing the last autosave" is the "guessing" part.
09:28:00 <planetmaker> Guessing sounds like it can fail. And that has to be avoided or caught and appropriately communicated to the user. Otherwise it's a 100% certain bug report as soon as it would hit trunk
09:28:41 <planetmaker> but a "rollback 5" console command for servers - would be awesome :-)
09:28:53 <planetmaker> where 5 could be either real time or number of savegames
09:29:00 <planetmaker> dunno what would be better :-)
09:29:21 <planetmaker> possibly savegames backward would be far easier.
09:31:20 <Terkhen> what if you rollback too much? could you then do rollback -2 for example?
09:31:50 <planetmaker> ^ good question. Should work IMHO
09:32:17 <planetmaker> that'd make it quite useful for those griever-cases
09:32:28 <Terkhen> it could store the number of rollbacks somewhere, and as soon as a new autosave is done reset the counter
09:32:57 <Terkhen> I don't know what happens to connected clients when you load a game, though
09:33:33 <planetmaker> IIRC. But I'd need to test again to be sure
09:33:48 <planetmaker> otherwise they are just drop
09:35:19 <Terkhen> what could a griefer do that cannot be solved by removing his company or that can be limited by the terraform settings?
09:36:17 <planetmaker> well. It can now be solved somewhat by the TF settings.
09:36:30 <planetmaker> But e.g. it cannot be solved on a coop server, Terkhen
09:36:30 <perk111> but sometimes people forget to set passwords
09:36:57 <planetmaker> the TF settings only make it (much) harder to do (big) damage
09:37:20 <perk111> and coop server too, yeah
09:37:22 <planetmaker> but on our coop server a person could easily just crash lots of trains etc
09:37:31 <Terkhen> hmm... right, if it is a passworded company with a publicly available password then he can do much damage
09:38:07 <planetmaker> you're right, the issue is much less severe :-)
09:38:17 <planetmaker> but this rollback feature might also be nice in single player
09:38:28 <planetmaker> As the much asked-for "undo knob" :-P
09:38:36 <Terkhen> I just load older autosaves
09:38:47 <planetmaker> well, that's what it'd do, too
09:39:01 <planetmaker> but with more comfort as I gather
09:39:20 <Terkhen> if you focus this to single player, people are going to ask for a nice GUI for it and wonder why it does not work for smaller intervals of time :)
09:42:00 <perk111> I don't think people use console in single much
09:42:14 <Rubidium> doesn't OpenTTD already keep the "ID" of the last autosave?
09:42:26 *** perk111 is now known as perk11
09:42:31 <Rubidium> so it'd be incredibly simply to "guess" the last autosave, as it just knows it
09:44:31 <planetmaker> that I'd not call 'guess' then :-)
09:44:53 <planetmaker> maybe that's what he does - I've not seen a patch. If that's what is done... should probably be fine
09:45:02 <planetmaker> though it ignores manual saves
09:45:47 <planetmaker> Dutch, Spanish and German people are most eager to vote ;-)
09:45:53 <planetmaker> at least going by the names :-P
09:46:06 <Rubidium> hmm, OpenTTD restarts from 0 each time it starts
09:46:32 <Rubidium> unless you're using "keep all autosaves", in which case the guessing is a bit harder
09:46:46 <Rubidium> especially when someone renames company #1
09:48:40 <perk11> why don't just take last modified file with .sav extension?
09:48:58 <planetmaker> perk11, because it could be from another instance of OpenTTD ;-)
09:49:32 <planetmaker> Not sure though how likely *that* is.
09:49:32 <perk11> but sharing save folder beetwen 2 instances is not very good idea
09:50:03 <planetmaker> for autosaves it might be a bad idea, yes
09:59:57 *** Vikthor has joined #openttd
10:48:06 <James_> Sorry, Internet connection stopped working
10:48:51 <James_> My "guessing" code uses the same printf() line as the doAutosave function
11:17:24 <xiong> Why is there a limitation on autorenew? That is, from -12 to +12 months.
11:17:47 <xiong> I mean, why can't I autorenew an engine that's only, say, 10 years old?
11:19:20 <SmatZ> there always has to be a limit
11:20:00 <SmatZ> maybe it would take too long to change -120 to +120 using 241 mouse clicks
11:20:22 <xiong> SmatZ, Well, that's a UI issue.
11:21:20 <xiong> Another way to get the same effect -- one I'd prefer -- would be to allow autoreplace to replace to the same engine. That would be much more tunable and workable for the user.
11:21:49 <peter1138> every one thinks that's a good idea
11:21:55 <peter1138> autoreplace has no time limits
11:22:10 <peter1138> you would have constant replacing if you allowed that
11:22:51 <xiong> This is not some weird edge case or crazy desire to run only new trains; not for me. NARS has several different models of the same engine -- for example, the American goes through two wood-burning models and two "Heavy", coal-burning models, each with stats significantly improved over the last.
11:23:17 <peter1138> the stat changes are "soft"
11:23:22 <xiong> I can think of several ways to get the effect.
11:23:25 <SmatZ> [12:19:54] <peter1138> http://www.boxcar2d.com/ <== only the first "car" seems to be placed on a road, others just fall down the screen - maybe it's not a bug, but it looks like that
11:23:26 <peter1138> ottd doesn't know they've changed
11:23:40 <peter1138> SmatZ, odd, worked for me
11:24:10 <xiong> No, the game doesn't "know". I think all models of the same engine share the same ID.
11:25:26 <peter1138> autoreplace just checks if a type of vehicle needs to be changed
11:25:30 <xiong> A clue is that in the autoreplace dialog, only the latest model of engine is shown in the left-hand pane, with the current stats; even though you can then replace existing, older models with some other engine. The count shown in the left pane is for all models of the same engine.
11:25:49 <peter1138> if you allow it to change to itself, it will need to be changed instantly again
11:25:57 <xiong> Well, I hate to impute much sentience to a computer program.
11:26:04 <planetmaker> and it'd be autorenew ;-)
11:26:21 <xiong> Well, I think you'd want to set a flag. That would be a new thing.
11:26:27 <peter1138> i do think autorenew should be more flexible
11:26:45 <planetmaker> possibly. What way do you mean?
11:26:52 <peter1138> xiong, then it would only replace once
11:27:10 <xiong> I already spoke about other choices besides 1-to-1. I'd like to be able to replace a single-unit train with an MU setup, for example; or the inverse.
11:27:40 <peter1138> i think the reasoning is: autorenew after 3 years would give you an unfair advantage
11:27:53 <SmatZ> there is still only 10.0.45.2 as "stable" version of flash player for 64bit linux :-/ with all its security problems...
11:27:54 <xiong> peter1138, Um, that would be the idea: Replace once. When "Stop Replacing" is pressed, the flag is cleared on each such train.
11:28:42 <SmatZ> xiong: why would you want to autorenew all your engines at once? no matter how old they are?
11:28:55 *** KenjiE20 has joined #openttd
11:28:57 <SmatZ> not to say, with many trains, the replacing round can take several years
11:29:11 <xiong> I wouldn't talk about automatic upgrading if it could be done by hand. But if I, say, pull a train into a depot, buy a new engine and throw it on the train, then remove the old engine, strange things happen. I have had empty groups with maxint-1 members.
11:29:38 <peter1138> thanks for the bug report
11:29:42 <SmatZ> xiong: then it's a bug (unless you have a modified version)
11:30:00 <xiong> SmatZ, Er, I thought I'd explained myself, why I want to do this. How have I been unclear?
11:30:24 <planetmaker> SmatZ: your sentence needs shortening to "there's only flash player with security problems". All other words in that sentence are unnecessary.
11:30:36 <SmatZ> please open a bugreport at bugs.openttd.org, attach a savegame and describe steps how to reproduce that
11:31:45 <SmatZ> xiong: sorry, perhaps I missed the line where you explained that
11:32:43 <peter1138> there are open source flash plugins...
11:32:52 <peter1138> not that they work very well...
11:33:14 <xiong> This is not some weird edge case or crazy desire to run only new trains; not for me. NARS has several different models of the same engine -- for example, the American goes through two wood-burning models and two "Heavy", coal-burning models, each with stats significantly improved over the last.
11:34:36 <xiong> Obviously, the user has a choice whether to continue to extract the last bit of useful life from the older model or change to a newer. Currently, the workaround I use is to autoreplace to a different engine entirely, then replace back to the first.
11:35:39 <xiong> This works only fairly, since there is not usually going to be an equivalent engine available -- all other engines will be inferior to the one being cycled, since that was chosen as the best one available.
11:36:53 <xiong> If I'm fortunate, the alternate engine will merely be less reliable; if less fortunate, it will be slower. If unfortunate, it will be less powerful and the autoreplace will fail, because (I assume) it sees that the proposed engine will not be able to pull the cars.
11:37:33 <xiong> Note that the last case holds even if the new engine *is* able to pull, but is less powerful by some amount. I haven't figured out what the threshold is.
11:41:44 <xiong> I feel that's perhaps a complete statement. More?
12:20:21 <xiong> SmatZ, I don't know how to take your lack of response.
12:22:54 *** Phoenix_the_II has joined #openttd
12:38:34 *** DanMacK has joined #openttd
12:40:34 *** fjb is now known as Guest1574
12:47:57 <SmatZ> xiong: sorry, I was away
12:52:00 <xiong> SmatZ, That's fine. I wondered if I'd answered your question adequately.
12:54:09 <SmatZ> xiong: so the problem is with newgrfs, where engine built in 1970 has different properties than engine built in 1960?
13:00:40 <xiong> At least in NARS, it's not year-by-year; there are sharp distinctions among, say, four introduction dates or models of the same engine. The same applies to passenger cars or mail cars.
13:04:42 <DanMacK> NARS is designed so that some locomotives evolve over time. the F-Unit is a great example
13:05:01 <DanMacK> It goes from the 1250HP FT in 1939 to the 1500HP F7
13:07:18 <xiong> DanMacK, Do you philosophically believe that I should not be able to exchange an older model for a newer one?
13:07:42 <planetmaker> like you can buy the same car every year. But every 2nd year it'll be a new generation :-)
13:07:46 <xiong> That is, that I should just wait until it ages out and autorenew gives me the newer model?
13:08:32 <DanMacK> You should, although most FT's were replaced by roadswitchers like the GP9 and GP35
13:08:38 <xiong> I like to start in 1850. The improvement in the American in 1870 really matters.
13:08:56 * DanMacK never plays with autorenew
13:09:54 <xiong> I suspect that answers my philosophy question.
13:10:16 *** Adambean has joined #openttd
13:10:25 <xiong> DanMacK, Do you play with breakdowns off, then? Otherwise, I fear to see the smoke on your layout.
13:12:01 <DanMacK> oh yeah, breakdowns are never on either, too much hassle :P
13:12:25 <planetmaker> :-) Still autoreplace makes sense then. Only autorenew is kinda... useless then.
13:12:37 <planetmaker> or do you never upgrade engines to a newer model?
13:12:56 <xiong> Oh. Well, with no breakdowns, you can have steam in 2000, I guess.
13:13:40 <DanMacK> oh I upgrade. It just makes it easier to have those steam excursions :P
13:14:13 <xiong> Sorry; I don't mean to be a jerk. I'm not one of those guys who starts up a convo in the hobby shop and sneers, after 15 minutes, "Transition!"
13:15:02 <DanMacK> So you're a modeler then too?
13:15:46 <xiong> Not a fascist but yes, I do like a feel for period.
13:15:58 * DanMacK modeled in HO for years
13:16:30 <DanMacK> I've also done N and Z, and I'm a part of our local O scale club ;)
13:16:48 <xiong> I don't understand a guy who blows a few hundred bucks on a gorgeous, ancient steamer and then throws on a package of RTR containers.
13:17:29 <DanMacK> yeah... that makes no sense
13:18:11 <xiong> The hardest thing to learn about the hobby is not to buy everything in the shop that looks nice.
13:20:09 <xiong> I'm living in San Francisco now, where people live in the garages and basements, so the real stuff is all in storage; which is why I'm into OTTD in the first place.
13:20:57 <xiong> There's a well-established Free-mo club in Marin but that's kinda far.
13:21:25 <DanMacK> I have a T-gauge loop on my desk ATM, that's it :P
13:22:21 <DanMacK> Off to drive a friend to work, laters :D
13:22:26 <xiong> I'm thinking of a Free-mo module that can reasonably double as a shelf layout. A yard.
13:54:33 *** DayDreamer has joined #openttd
14:05:39 *** Devroush has joined #openttd
14:11:06 *** Chris_Booth has joined #openttd
14:16:08 *** Brianetta has joined #openttd
14:28:34 <JVassie> Can anyone reccomend a font set which works with departure boards please?
14:30:28 <Terkhen> I don't understand, why does a patch require specific fonts?
14:31:04 <JVassie> Well there are some display problem with the default fonts it looks like
14:31:09 <JVassie> sometimes it shows the full time
14:31:24 <JVassie> others, it only shows 02:.. for example
14:32:11 <Terkhen> that sounds like a bug in the patch to me; if done correctly the window should be resized to fit all strings on it
14:33:02 <JVassie> Ill post in the topic i guess
14:33:10 <Terkhen> if that's the issue then you should pick a smaller font size (and report the problem to the patch coder)
14:33:10 <JVassie> just wonderign if anyone knew of a temporary fix
14:33:20 <JVassie> how do I pick a smaller fotn size?
14:34:26 <Terkhen> JVassie: edit small_size, medium_size and large_size at openttd.cfg (it only works if you set up a font different than the default one in the *_font settings)
14:34:51 <JVassie> so i need to use a fotn.grf of some description for it to work?
14:35:29 <Terkhen> no, just use one of your system fonts (Arial, etc)
14:36:20 <JVassie> small_font, medium_font and large_font are all empty, change them to Arial for example?
14:38:06 <JVassie> didnt need to change sizes
14:38:10 <JVassie> just setting the fonts worked
15:05:20 *** DayDreamer1 has joined #openttd
15:18:01 <xiong> Ever notice how, when trains crash into each other, they always do so (a) at the most congested point in the network and also (b) the point where you are currently working to fix the congestion?
15:18:43 <SmatZ> modifying the track is a pre-requrement for trains to crash
15:19:22 <xiong> Yes, indeed. So it is fiendishly precise that any crash will be at the most inconvenient possible place.
15:26:38 <xiong> I now have a terrible snarl as the jam propagates up all the lines that feed into the blocked junction. Impatient trains have reversed, compounding the confusion.
15:26:59 <SmatZ> you can disable train reversing at signals
15:27:21 <peter1138> ahh, the days of hackykid's pbs
15:27:29 <peter1138> when trains *would* 'randomly' crash
15:27:42 <xiong> I saw that setting and wondered if I should. Or if it would be better to build a more robust network.
15:28:41 <xiong> In prototype, it's not unusual to run trains wrong-way on the main. Not done constantly but certainly, if there's an issue up ahead, backing a train at least to a siding is quite reasonable.
15:30:01 <xiong> In fact, it is this particular issue that makes me rethink my reliance on path signals. Trains are quite content to run wrong-way on a path-signaled main; but they see no signals going "their", wrong way; so they feel they need to wait for an immense length of track to clear before proceeding.
15:30:44 <xiong> Maybe it would be enough to scatter a few wrong-way standard path signals.
15:31:39 <zydeco> or you could use one-way path signals
15:31:56 <xiong> But it would probably be better to identify sections of main that can plausibly run both ways, install crossovers, and block signals.
15:32:06 <xiong> zydeco, I don't see how one-ways would help.
15:32:55 <zydeco> or stop trains that are going into the junction you're modifying
15:33:01 <zydeco> that would avoid the crash :p
15:33:05 <xiong> Currently, I use a very small number of one-ways; just enough to keep trains coming out of depot the wrong way, routinely.
15:34:12 <xiong> Oh well, the crash wasn't caused by bad signaling. I think that's very, very hard to do, if not impossible. The crash was caused by my removal of a signal while in the process of clearing a jam at a nonstandard junction.
15:35:35 <xiong> Now, I have a much worse jam. I need to clear that up; I now have multiple deadlocks.
15:36:07 *** Wolfsherz has joined #openttd
15:36:13 *** tokai|mdlx has joined #openttd
15:36:37 <xiong> It would help if I could urge trains that have swapped ends (impatiently) to *use* the track they're trying to claim.
15:37:04 <Mazur> In my experience, crashed happen when I remove a signal, and at the same moment another train has just cleared the signal before the one I removed. The lesson I learned was: never remove a signal unless the downtrack part of it is free and cannot be entered within 2 seconds of my removing.
15:37:27 <xiong> At this point, I'm afraid, and with the network I have; the best thing to do probably is send All Trains to depot and sort it out.
15:37:52 <Mazur> Still goes wrong on occasion, though, when I'm getting tired and lose focus.
15:39:03 <xiong> Mazur, You're right. But I had about 8 trains all fighting over the same tile. I backed 3 out of there, one at a time, wrong way and ordering them to blow through signals, then (carefully!) stopping them before they crashed. I fiddled and diddled; with more trains coming into the back of the jam, I had less and less room to work in.
15:40:01 <xiong> I so loathe to All Stop, let alone send all to depot. It clears the ways but messes up every possible delivery or agreement between anybody, anywhere.
15:40:36 <xiong> Heaven forfend you should have trains on timetable when you fix a jam that way.
15:44:46 <xiong> You know, this might just be a fine time to go to bed. Night, guys.
15:45:51 <Mazur> Yeah. Another safety measure is to temp stop all trains upward from the signal with CTRL-<right-button>, remove the signal and start thrm up again one by one.
15:46:10 <Mazur> But who has the patience for that?
15:46:29 <Mazur> all the nearest ones, that is.
15:48:03 <Mazur> Or left, I get tha mixed up at times. The most used one,.
15:49:05 <Mazur> And I use my mouse left-handed, from when I had a job and needed to rest my right arm in my spare time to avoid a painful shoulder.
15:53:34 *** Chris_Booth has joined #openttd
16:40:18 *** Prof_Frink has joined #openttd
16:45:00 *** Progman has joined #openttd
17:10:10 *** snorre_ has joined #openttd
17:13:42 *** snorre_ is now known as snorre
17:19:36 *** supermop_ has joined #openttd
17:19:47 *** supermop_ is now known as supermop
17:20:30 <supermop> I started drawing a rail set last night
17:20:47 <planetmaker> :-) Rail or train?
17:22:20 <supermop> nothing so fun as actual trains
17:22:26 <supermop> that still has to wait
17:22:45 <planetmaker> some people claim that OpenGFX needs better tracks.
17:22:51 * peter1138 considers the unconsiderable
17:22:57 * peter1138 also ponders the unponderable
17:23:08 <planetmaker> up to the challange, supermop ? :-)
17:23:44 <supermop> mine are too ugly yet, but it could happen
17:24:14 <supermop> i need to just make some basic rails so that i can make a grf that gets depots matching my sheds into the game
17:24:40 <planetmaker> supermop, just... grab an exiting track set
17:24:45 <planetmaker> And replace the depot
17:24:56 <planetmaker> Or... just replace the default depot
17:25:03 <planetmaker> If that's your concern... no need for tracks
17:25:43 <peter1138> Hurr, r22088 is r22087!
17:27:35 <planetmaker> and that's actually doing it the complicated way already
17:31:18 <supermop> i need to add several different depots
17:32:19 <planetmaker> :-) if it's all rail depots... yeah, then you got no choice.
17:34:47 <planetmaker> but if the railtypes exist - then you only need to supply the depots anyway :-)
17:37:06 <supermop> anyway i am going to try to make a simple set, with monorail, maglev (maybe JR and Transrapid versions), and three regular rail speeds
17:43:35 *** |Jeroen| has joined #openttd
17:43:45 <supermop> but i want provide a lot, then maybe make it detect if another rail grf is loaded, and if yes, it just overwrites those depots, but doesn;t use my rails
17:44:22 *** frosch123 has joined #openttd
17:44:23 <supermop> so i want to get close to 16 depots in, in case someone is using a grf that uses all slots
17:45:42 <planetmaker> supermop, but... then I'd start with the grf detection and the depots :-)
17:45:47 *** Wolf01 is now known as Guest1602
17:45:48 *** Wolf03 is now known as Wolf01
17:46:17 <supermop> that involves code i can't understand yet
17:46:19 <planetmaker> or rather with the railtype detection
17:46:37 <supermop> i was hoping if i made nice enough tracks, someone might want to help me
17:50:13 <Mazur> Actually, none of them is my winner.
17:52:38 <planetmaker> well. And the grf definition. But that's a simple grf block
17:52:40 * Mazur aims for: includes all 3 (4) types of rail with trains, RVs, planes, ships, a few eyecandied stations, like an industry specific station next to its industry, and yet is overall not too crowded a picture, so the transport elements all come out well.
17:55:31 <Mazur> Oh, there are a few more.
17:58:42 *** Vikthor has joined #openttd
18:00:48 <Mazur> And I;m realising eyecandied industry stations are out because they aren;t in the standard set.
18:01:34 <Belugas> pleasure of discovering :)
18:04:43 <supermop> was there talk of ogfx platforms?
18:06:53 <planetmaker> there was. Long ago by FooBar and obviously is free for the taking ;-) Will be nice to see
18:08:16 <Mazur> Well, pm, I made a first selection. Later, when I'm in the mood again, I'll refine my choices further.
18:08:42 <Mazur> 10 (9) out of 30 pass my "tests".
18:08:58 <supermop> wait, what does that code do?
18:09:25 <planetmaker> supermop, it changes the depot of the RAIL railtype when it is present
18:09:30 <planetmaker> with snow awareness
18:10:04 <planetmaker> supermop, I'll make you a short ready-to-compile newgrf
18:10:23 <planetmaker> where you then can copy it for other railtypes. You'll need to install NML in order to compile it
18:10:24 <supermop> what if there are many RAIL railtypes?
18:10:38 <planetmaker> supermop, there can be only one railtype with the label RAIL
18:10:49 <supermop> what do i need to do for nml?
18:19:47 <supermop> i will have to look into that when i get home from work tonight
18:21:41 *** Chillosophy has joined #openttd
18:22:23 <planetmaker> ^ @ supermop for that conditional depot
18:22:34 <planetmaker> I haven't yet tested whether it actually works, though
18:24:13 <supermop> i will have to email the link to myself
18:25:15 <Ammler> I have 4 bananas newgrfs wich use nfo version 1
18:25:17 <supermop> now just need to take another shot at it tonight
18:25:29 <planetmaker> Ammler, yes, we know ;-)
18:25:45 *** Fenris3 has joined #openttd
18:25:51 <planetmaker> they don't actually use it. They just claim to use it
18:26:02 <Rubidium> Ammler: yes, we intentionally broke them!
18:26:15 <Ammler> Rubidium: not the first time :-P
18:26:54 <Ammler> where are all the nice devs gone?
18:28:30 <peter1138> tron left years ago
18:32:54 <Belugas> indeed, he was such a nice dev :)
18:38:11 *** Chillosophy has joined #openttd
18:45:47 <CIA-11> OpenTTD: translators * r22089 /trunk/src/lang/ (6 files in 2 dirs): (log message trimmed)
18:45:47 <CIA-11> OpenTTD: -Update from WebTranslator v3.0:
18:45:47 <CIA-11> OpenTTD: afrikaans - 96 changes by Maccie123
18:45:47 <CIA-11> OpenTTD: arabic_egypt - 29 changes by kasakg
18:45:47 <CIA-11> OpenTTD: danish - 69 changes by krak
18:45:49 <CIA-11> OpenTTD: indonesian - 17 changes by adjayanto
18:45:49 <CIA-11> OpenTTD: brazilian_portuguese - 25 changes by Luis_Mizuchiro
18:45:55 *** JOHN-SHEPARD has joined #openttd
18:50:49 *** Fenris3 is now known as Fenris
18:50:55 <Eddi|zuHause> [16.02.2011 19:26] <Ammler> where are all the nice devs gone? <-- maybe you can join the "goal" team on the "we hate Rubidium" website :p
18:51:59 <Ammler> Eddi|zuHause: the issue with your feature requests is that you could simply patch your SP game, we have to use official releases :-)
18:52:39 <Eddi|zuHause> Ammler: again. i made a BUG REPORT. not a FEATURE REQUEST.
18:53:33 <planetmaker> which resulted in a mis-implementation being removed and replaced by a proper one
18:54:05 <Ammler> Eddi|zuHause: it was a bug for you, it was a nice feature for us :-)
18:54:35 <SmatZ> Belugas: I don't see any Rubidium there, but a lot of girls :)
18:54:57 *** Wolfsherz has joined #openttd
18:55:13 <Belugas> Rubidium, female, 17, philipines
18:55:20 <Belugas> thus"Not that site" ;)
18:55:29 <Ammler> Eddi|zuHause: also you needed around 5 years to find such a "bad" engine
18:56:21 <Eddi|zuHause> Ammler: actually, i did find a similar issue before, but i couldn't place it in relation to the turning feature
19:00:25 <Ammler> Eddi|zuHause: you seriously prefer the current situation?
19:00:44 <Hirundo> Users tend to be quite bad at reporting bugs, especially non-crashing ones... 5 years says very little IMO
19:01:08 <Eddi|zuHause> Ammler: which exact situation do you mean?
19:01:29 <Ammler> disabling flip to fix your rare glitch
19:02:11 <Eddi|zuHause> Ammler: yes, i prefer the "proper" handling. ask your favourite newgrf author to provide a fix.
19:02:46 *** Chillosophy has joined #openttd
19:03:02 <Ammler> Eddi|zuHause: for you as SP gamer, it is easy to say :-)
19:03:37 <Eddi|zuHause> Ammler: technically, i don't use the turning feature at all. i stumbled over this by accident
19:05:26 <planetmaker> luckily there are users who report bugs and glitches.
19:05:33 <Ammler> so you reported something which disabled a feature we use but you don't, how sarcastic :-)
19:05:50 <Hirundo> again, Eddi did not disable anything
19:05:56 *** andythenorth has joined #openttd
19:06:03 <Ammler> Hirundo: of course not, never said that
19:06:14 <planetmaker> can we please stop this pointless whining. Now?
19:06:26 <Ammler> but devs wouldn't have "found" it else :-)
19:07:36 <planetmaker> It was to my knowledge discussed already in extensive length
19:07:40 *** andythenorth has left #openttd
19:07:44 <planetmaker> hello andy... :-(
19:09:55 <Eddi|zuHause> "byet" sounds russian...
19:10:37 *** andythenorth has joined #openttd
19:10:40 *** andythenorth has left #openttd
19:10:47 *** andythenorth has joined #openttd
19:26:44 *** Biolunar has joined #openttd
19:27:30 *** andythenorth has joined #openttd
19:41:10 *** Progman_ has joined #openttd
19:42:35 *** Brianetta has joined #openttd
19:46:59 *** Progman_ is now known as Progman
19:57:45 *** andythenorth has left #openttd
20:01:11 *** ABCRic is now known as Guest1623
20:21:05 <DorpsGek> planetmaker: OpenTTD uses TCP and UDP port 3979 for server <-> client communication, UDP port 3978 for masterserver (advertise) communication (outbound), and TCP port 3978 for content service, a.k.a. BaNaNaS (outbound)
20:26:48 *** ABCRic is now known as Guest1628
20:37:55 *** fmauneko has joined #openttd
20:50:58 <zydeco> openttd is showing me non-existing days
20:51:23 <Rubidium> yes, 18-02-2011 doesn't exist
20:51:41 <zydeco> no between 5 and 14 october 1582
20:52:35 <zydeco> it's when the gregorian calendar was implemented
20:52:57 <avdg> well, I think its more complicated then that
20:53:12 <zydeco> yes, some countries switched on different dates
20:53:27 <Rubidium> e.g. Turkey switches in 1926
20:53:31 <planetmaker> zydeco: you're welcome to supply a fix ;-)
20:53:42 <avdg> yeah, its more a case by case for each country/zone
20:53:42 <planetmaker> make sure it's a user interface setting only ;-)
20:53:51 <planetmaker> as it should then be locale-dependent
20:54:09 <Rubidium> yes, make sure you support the julian calendar as well....
20:54:34 <Rubidium> oh, and before ~1200 there would be counting in a different way
20:55:07 <Terkhen> the chinese calendar should be implemented too then :)
20:55:16 <planetmaker> I want the juish calender
20:55:17 <Rubidium> though, to please the English... the new year thingy should be at March 25th until 1751
20:59:08 <Rubidium> anyhow, in the OpenTTD simulation of the Gregorian Calendar was introduced in the year 0
21:02:39 <zydeco> maybe I'll just show 1582-10-04 for a few days in my admin client
21:02:48 <zydeco> or not use a date class
21:06:19 *** supermop has joined #openttd
21:07:33 <dihedral> i just hope nobody tries joan / grapes at those dates
21:07:39 <dihedral> just in case something goes wrong
21:07:48 <zydeco> it probably just shows the wrong date
21:08:23 <planetmaker> dihedral: good point. I'll start a middle age game with 1.1 ;-)
21:09:18 <dihedral> planetmaker, you are way too mean :-P
21:10:01 <planetmaker> sorry, I just had an evilgasm
21:11:48 <dihedral> spare me the details pm :-D
21:12:27 <CIA-11> OpenTTD: planetmaker * r22090 /trunk/docs/multiplayer.txt: -Doc: Explain OpenTTD's ports and their role in multiplayer and especially for servers
21:15:52 <planetmaker> (one of my favourite comics ;-) )
21:16:14 <krinn> planetmaker, can trouble you with noai here so ?
21:16:31 <planetmaker> you can. But _I_ no not much about AI programming
21:18:18 <planetmaker> krinn: the whole point of this channel is: don't ask specific people. Just ask your question
21:18:37 <planetmaker> IRC is not like a telephone conference. Rather like speaker's corner in hyde park
21:18:39 <krinn> :) ok, there's no function to get the vehicle last servicing date if i'm right, is this plan to be add ?
21:19:23 <krinn> i know enough irc to know the rules are not define by common sense, but by the dictatorship op
21:19:56 <dihedral> planetmaker, in which case i'd like some virtaul (rotten) tomatos to throw :-D
21:20:32 <planetmaker> krinn: so it depends on the ops ;-)
21:20:47 <dihedral> krinn, the "dictatorship op" in this channel has all the common sense you might lack in other channels
21:21:01 <planetmaker> and as long as it's somewhat remoteley related to openttd we go here by...
21:21:05 <DorpsGek> planetmaker: Don't ask to ask, just ask
21:21:45 <planetmaker> those who know and have time or are online (or think to know :-P ) will answer
21:23:57 <planetmaker> hm... so you ask whether we will add some function? Maybe... maybe not :-)
21:24:02 <dihedral> for the feature or the answer?
21:24:17 <krinn> the answer, for the feature i might die here no ?
21:24:17 <planetmaker> The question is: can a player get that info, krinn ?
21:24:30 <krinn> no, can an ai get that info :)
21:24:48 <planetmaker> krinn: and my question is: can you as player get that info?
21:24:58 <dihedral> the ai may only get the info a human player can get also
21:25:00 <planetmaker> if no: then that will not be planned. If yes: then provide a patch :-P
21:25:09 <krinn> you can also set the interval for servicing / vehicle
21:26:06 <krinn> any vehicle -> details -> last line of the window: interval service + last date it was made
21:26:20 <planetmaker> yes, that I know :-) But I so rarely play with servicing... I forget those details :-)
21:26:33 <planetmaker> yes, you're right, of course
21:27:08 <krinn> well, it's not the servicing that is really good to get (for an ai pov, well, in my mind)
21:27:26 <krinn> i could just setup the ai to query a servicing itself
21:27:38 <krinn> but the info that a servicing was made at date X is interresting
21:27:56 <krinn> it's a proof the vehicle can service (depot exist, can be access...)
21:28:07 <dihedral> to send the train 'manually' to a depot?
21:28:50 <krinn> dihedral, still think about an ai: sending a train to a depot is easy to do, seeing it has done it, is not
21:28:56 <dihedral> it's not - i.e. in cases where breakdowns are disabled and 'services if breakdones disabled' is disabled
21:29:10 <krinn> except if you wish your ai to wait & follow it until it really reach it
21:29:19 <planetmaker> krinn: you could get that info also via the reliability
21:29:40 <dihedral> can the ai get hold of the info if a train cannot find a depot?
21:30:09 <dihedral> and should an ai not know how it built stuff? i.e. in a way that a train can access the depot?
21:30:15 <krinn> no, you can just know if the goto depot order wasn't accept (if train is too far for example)
21:30:47 <planetmaker> dihedral: it doesn't know that when a savegame is loaded
21:31:07 <planetmaker> unless it saved all that...
21:31:12 <krinn> planetmaker, reliability is better for checking that, but still, sometimes not, a too crowd town will get down your reliability even you have a depot
21:31:22 <dihedral> but an ai is only loaded if the ai functioned as that company
21:31:49 <krinn> also my ai can destroy depot (lol)
21:32:06 <dihedral> but it should know which depot it destroys
21:32:12 <dihedral> rather than just destroying it
21:32:16 <krinn> or in fact, my ai accept to run with a dead depot (for example when i'm low on money to build one)
21:32:41 <dihedral> wow - which ai are you talking about? ...
21:32:44 <planetmaker> well... it's a function which makes sense
21:32:59 <krinn> and also, my ai can destroy (virtually) your depot
21:33:02 <dihedral> planetmaker, yes, but the use of the function does not always make sense
21:33:09 <planetmaker> I seem to recall there was in the NoAI section a thread where such functions could be requested...
21:33:14 <krinn> suppose my ai build something on your road line: your depot is dead, while still present
21:33:49 <krinn> you have a bus+depot from a->b
21:34:03 <krinn> i my ai build between a & b your depot might be unusable
21:34:42 <dihedral> if: there is only one route from a to b; you build rail tracks and place a train and do not move the train
21:34:54 <dihedral> in which case i'd be happy to not even load your ai ever again
21:35:18 <krinn> ? i'm not speaking about a war there
21:35:28 *** chmstsg has joined #openttd
21:35:28 <krinn> and don't think about train tricks...
21:35:58 <dihedral> then my route from a to b is not affected at all times
21:35:58 <krinn> it's just if i need to build myself a road depot, you don't really my ai to also check it won't bug your own road route right ?
21:36:39 <dihedral> why would your road depot bug mine?
21:36:51 <dihedral> unless you create traffic jam
21:36:58 <krinn> not only road depot, anything that destroy your route
21:37:07 <krinn> suppose you have a route in a town
21:37:20 <dihedral> i do not see how you destroy a route without removing road tiles, or blocking them
21:37:20 <krinn> everyone will be (except too poor rating) to destroy road there
21:38:56 <dihedral> and that is why you wanted the date check for? in order to see you did not remove a road tile that was required for the route to your own depot?
21:39:23 <krinn> it was part of the idea yes
21:39:32 <dihedral> then we got to the root of it :-P
21:39:38 <dihedral> planetmaker, see what i mean :-D
21:40:06 <krinn> dunno for him, i'm still don't see what you mean there
21:40:11 <dihedral> + you cannot always remove road tiles at will
21:40:22 <krinn> there's plenty events, not even related to a player that can do that too
21:40:25 <planetmaker> well, technically also another player could do that, dihedral
21:40:40 <dihedral> yes - and those players can get kicked
21:40:51 <dihedral> krinn, those are disasters
21:40:53 <krinn> dead ai (player?) going bankrupt
21:41:03 <dihedral> roads are not sold (iirc)
21:41:06 <krinn> and so its bridge/tunnel are removed...
21:41:15 <planetmaker> so it's a valid desire to check whether a vehicle can reach a depot. But... you could do the same as the player, krinn: check the route and find a depot - or fail to find one
21:41:18 <dihedral> iirc those are kept also
21:41:24 <planetmaker> it needs a somewhat modified road path finder
21:41:30 <planetmaker> those are available as lib for AIs
21:41:57 <dihedral> the requrest for sure is valid - it's just the reason for the request is dubious ;-) that's all i am pointing out
21:42:13 <planetmaker> however... for today I wish all a good night :-)
21:42:23 <krinn> i have myself done a checking (walking the road and check it's valid)
21:43:11 <dihedral> i will do the same - night :-)
21:43:48 *** Chruker has joined #openttd
21:44:53 <Zuu> krinn: In order to detect broken road links, I keep a long term and a short term mean of the total income of all vehicles on a route. If the short term income drops too much, it activates the pathfinder to check if the road path is borken.
21:45:44 <Zuu> This also fixes the case when someone tricks my AI when I use someone else road and they decide to rebuild it to have a long detour.
21:45:46 <krinn> ah thank you zuu, looks a good idea and less time consuming
21:46:30 <krinn> i'm not sure if dihedral has done an ai itself, there's plenty case when your ai can use another road and something can break that road
21:46:46 <Eddi|zuHause> zydeco: you can't just leave out days in one year. you have to add them in other years. the days didn't just "disappear", they were renumbered backwards, which is what openttd calculates
21:46:59 <krinn> again, if someone goes bankrupt, everyone then could destroy its road
21:47:09 <Zuu> Whenever I receive a message about crashed trucks I trigger some code that tries to replace the level crossing with a bridge. Its not fool prof, but works in many cases.
21:47:40 <Eddi|zuHause> zydeco: the dates DO exist, just the people at the time did not know them
21:47:41 *** ccfreak2k has joined #openttd
21:48:24 <Eddi|zuHause> zydeco: openttd uses a normalised gregorian calendar with year 0
21:48:43 <krinn> zuu, i have more a build costs reduction idea in my mind, specially for early and hard settings game
21:48:47 <Zuu> The biggest problem I have is that I don't think I handle rail crossings on road peices adjacent to road crossings/turns and slops in the wrong direction (so that you can't just replace the road with a bridge)
21:50:21 <krinn> i had think too about checking income for the vehicle, but this fail if the vehicle has done an income already most of the time
21:50:41 <Zuu> I have an idea about modelling your connections as lines and trying to give bonus to new connections that have high match against existing lines (high chance of reusing existing road). If you manage to use some of the leach ideas from Rondle to detect other companies connections that wolud probably highly benefital.
21:51:46 <Yexo> just modify the pathfinder not to try and build new road but only reuse existing road
21:51:59 <krinn> well, leaching is good (not waiting for it), but you drop a depot+station and pathfind, i don't see anything bad if pathfind find a route to use it
21:52:08 <Yexo> because the number of possible routes is extremely limited (compared to a pathfinder free to build new road) it's also very fast
21:53:52 <krinn> i will look at that yexo
21:54:49 <krinn> look like i'm the only one to think servicing date isn't a good info
21:55:28 <Zuu> Most AIs reuse exsiting road when the pathfinder happen to come by existing road between A and B. However that happen on a microscopic level. What would be interesting is to at a macroscopic level choose A and B as to increase the probability that the pathfinder will find existing road to reuse.
21:56:10 <Yexo> Zuu: admiralai first tries a bit to find existing routes before it'll try to build a new one
21:56:36 <krinn> remember i have still in mind early days
21:56:50 <krinn> it's always good to reuse a road than spending hard money on it
21:57:20 <krinn> and all ai i have see reuse roads so far
21:57:37 <Zuu> Also pathfinding to build new road can be quite time consuming if the land is rough and have many lakes.
21:57:39 <krinn> (i'm not speaking about waiting to leech them like rondje)
21:59:24 <Zuu> Also it would from a company perspective (even if you don't reuse other companies roads) be benefital if you can find connections that re-use the road that you have built.
22:00:17 <krinn> it's just my ai can destroy road itself
22:00:41 <krinn> i can handle that, i really don't have time to handle how others will react to that :)
22:01:46 <krinn> still, don't take it wrong like he did, it's not for war purpose, just my roads & depot are good spot to build
22:01:57 <krinn> because i'm sure no town/player will refuse me to build on my own road
22:04:24 <Zuu> A good AI should handle that the road it has been using that is owned by someone else is removed/reconstructed.
22:05:54 <krinn> that's what i think, i don't see anything bad at doing that, as long it's not for wars
22:06:03 <Zuu> It may be considered a bit hostile on some multiplayer servers if you try to cause problem to the other players by reconstructing a bit of road that they use a lot if you don't have clear reasons to do that.
22:06:29 <krinn> well, it's when expanding conditions
22:06:38 <krinn> adding more stations per example
22:07:02 <krinn> i generally remove my depot and pickup the place (if the depot was made next to the station)
22:07:08 <krinn> but that's not always the case
22:07:24 <Zuu> Bt then some players have interesting demands on how the AI should play. Eg - that the AI is not allowed to decide how to name their own stations.
22:08:58 <krinn> well, i wonder how they will react if they themselves not allow to name their stations?
22:08:58 <Zuu> Btw, krinn what's the name of your AI?
22:09:23 <krinn> you can find it at openttdcoop
22:09:34 <Zuu> I think I've read about it somewhere on the NoAI forums.
22:09:39 <krinn> it's not yet usable (in my mind)
22:10:21 <krinn> never spoke about it yet, i might release a preview soon
22:10:55 <krinn> i need players to check it (for debug/crash purpose), but it's not yet to a point i will put it on banana
22:11:15 <Yexo> that reminds me, Zuu, would you be ok with moving your AIs from noai.openttd.org to dev.openttdcoop.org in the future?
22:11:37 <Zuu> I think that would be ok.
22:11:55 <Yexo> most projects there use mercurial instead of svn, but svn might be possible if you really want it
22:11:57 <Zuu> All I would need is to learn a bit more about HG but that's probably good anyways.
22:12:27 <Zuu> krinn: Actually, you have writtern about yoru AI in the short-names thread :-)
22:13:33 <krinn> the dictator name is just i need a name and i name the president as a dictator, not as evil
22:14:15 <Zuu> I think it is fine, it just give you a lot to live up to :-)
22:14:16 <krinn> i'll try to make it competitive, but also kind to human (limit vehicle to not crowd road too much...)
22:14:46 <Zuu> "Clueless" is imho a better name if you don't want high expectations :-p
22:17:07 <ABCRic> Zuu: but users DO have high expectations for an AI with such a clever name :)
22:18:22 <krinn> and i was kind, even i had evil ideas that came to mind (like slaving another instance of myai to boost main instance...), wasn't really fair :p
22:19:52 <Zuu> hehe, I've had some wierd ideas about how to make two AI instances communicate within the means of available API. :-)
22:20:17 <Zuu> It could actually be quite interesting to try to write a data transport layer :-)
22:20:22 <krinn> hihi you could, i have think about it
22:20:52 <krinn> you can check tile for a road, then check the road owner
22:21:26 <krinn> was too evil, but i had think about electing a master/slave doing that
22:21:45 <krinn> then the slave ai will bankrupt while building structure for the master ai
22:21:50 <krinn> told you it was too evil
22:22:41 <Zuu> well, from a programming/AI point of view I would find it interesting if someone picked up that route.
22:22:47 <ABCRic> meh, I see no API function to purchase a tile :(
22:22:59 <krinn> drop a road, tile is now yours
22:23:27 <ABCRic> krinn: but I like the little flag thingy
22:24:01 <krinn> oh it's easy, just add an algo to pickup a position where to work on
22:24:15 <krinn> and you can then set roads on tiles to speak with the other ai
22:25:16 <krinn> it's how the SOS message work when you're alone on an island :)
22:25:29 <ABCRic> what if a mean player draws some road in there? communication intercepted!
22:26:03 <krinn> eheh, as the tile position could be invalid (water...) your algo need to find some others places
22:26:24 <krinn> but remember, your 2nd ai instance have the same algo, and will also pickup another place as the first place isn't valid
22:26:59 <krinn> some players do that i saw, in coop they use a space to speak with others
22:27:06 <krinn> drawing the plan to build on...
22:27:28 <ABCRic> I guess it's easier than chat :D
22:27:49 <krinn> i think it's because you need drop a message to someone not there
22:28:25 <krinn> that could read it later directly on the map (ai can't use sign for that, human could, dunno why they have done it like that)
22:29:20 <Terkhen> AIs cannot check signs?
22:29:25 <Rubidium> as using other AIs to build your road is cheating, and they didn't want to get AIs to be working in such a way
22:29:31 <Zuu> AIs can only read their own signs
22:29:52 <Zuu> it was made that way to make it harder for AIs to cooperate in these ways.
22:30:17 <Zuu> Or for players ot use an AI as a slave.
22:30:24 <krinn> well, i suppose i should have keep that for me so
22:30:52 *** _Chruker has joined #openttd
22:31:00 <krinn> it's even easier for human speaking to an ai
22:31:46 <Zuu> Yes, now there is in-game changable settings.
22:31:54 <krinn> but seriously who would dev an ai to cheat on the game ?
22:32:09 <ABCRic> i read something about a framework for communicating with AIs or something like that...
22:32:14 <Zuu> There are plenty of forum requests to have an AI to help building rail etc.
22:32:30 <Zuu> manage your vehicles etc.
22:32:54 <ABCRic> krinn: ask the AI to pathfind for you, makes it easier to build routes
22:33:01 <krinn> well, you have good idea to learn how to do things already, just looking at them running should be enough
22:33:24 <ABCRic> of course pathfinders aren't very helpful sometimes... getting from A to B and stuff like that
22:34:42 <krinn> i did think about an ai that could show how to build stuff, but not for cheating, it was more like a tuto
22:35:12 <krinn> start a map and the traintutoai and look the ai building a train line from a to b while putting sign that say to player how to do things...
22:35:20 <Zuu> There are some management/statistics tricks that really only an AI can do, that players would have use for if it was available to them.
22:36:17 <krinn> Zuu, only case i could think about is replacing trains on different rail type incompatible with others
22:36:36 <krinn> but i found a grf that does that already
22:36:59 <Zuu> You can loop through all your stations quite often and make vehicles skip orders if too many wait at the same station for example.
22:37:22 <Zuu> You could implement ~50% load.
22:37:40 <Zuu> That would be a micro management hell for a player to do.
22:37:49 <ABCRic> of course, some AIs need better vehicle management. I remember once when I removed road that was being used by an AI's route, so its vehicles were jamming everything. the AI, unaware of that, detected the stations had a lot of passengers and built more vehicles, jamming even more
22:38:23 <krinn> lol ABCRic it was what i was speaking about earlier
22:38:41 <krinn> a simple road remove can bankrupt an ai
22:39:11 <ABCRic> yup, so remember: check for jams before checking for cargo...
22:39:43 <Zuu> And if you don't want to actually check for jam, check for income dips.
22:40:14 <krinn> i had think about income checks, not enough it seems
22:40:57 <krinn> checking yearly income fail if income done already > vehicle running coast when the road is crash
22:41:05 <ABCRic> the player will always find a way to disrupt the AI. *evil laughter*
22:41:36 <Zuu> That's why I said income dips.
22:41:54 <Zuu> Monitor the income regularly and see if there is a dip in the income.
22:42:23 <krinn> how you then find high breakdown or crowd road that lower the income?
22:42:27 <Zuu> Do it at an aggregated level as of the sum of all vehicles running on the same route.
22:43:40 <Zuu> IIRC I keep a long-term sliding mean income. I also have a short term mean income. If the gap between the short-term mean is too large (and with the short term being < long term), I detect that as a potential broken road.
22:44:32 <Zuu> I don't remember all details exactly, for that I would have to look at the code.
22:44:45 <krinn> looks like the reliability could do the same no ?
22:45:47 <Zuu> Might even be a bit more robust against noise from production changes.
22:46:49 <krinn> would be hell to handle boat like that no ?
22:47:27 <krinn> as low income could < 0 with boats running cost and extermly high when reaching the station
22:48:35 <Zuu> If you use a NewGRF that have ships with so high runnig costs that the income is < 0, then I don't see why you would use those ships.
22:48:53 <krinn> reliability could be better in my mind, but still if i could check last time the vehicle was servicing at depot
22:48:55 <ABCRic> add something that checks if one of the vehicles on a route is broken down
22:49:37 <krinn> actually i have only one grf (the rail one)
22:50:04 <krinn> and my boats (as human, i don't handle boat with the ai) do that for oil
22:50:45 <Zuu> I'm sure it is possible to figure out a way to handle ships.
22:50:55 <ABCRic> or check the speeds of vehicles on a route, if there are several vehicles on 0 it's because a vehicle is broken down
22:51:10 <ABCRic> or they are all stuck at the station...
22:51:18 <ABCRic> but usually they keep moving
22:51:20 <krinn> i do that for station balancing
22:51:36 <Zuu> My AI vehicle manangement is at the moment however focused on low-capacity vehicles (eg RVs) and have sometimes problems when it serve a secondary industry served by in-frequent large volume deliveries.
22:52:01 <Zuu> You can get the current vehicle state, which I would think includes broken down.
22:52:15 <ABCRic> I'm talking RVs of course
22:52:17 <krinn> yes, i check AT_STATION && VS_RUNNING
22:53:00 <Zuu> Eg: local veh_list = GetRouteVehicles(); veh_list.Valuate(AIVehicle.GetVehicleState); veh_list.RemoveValue(AIVehicle.VS_RUNNING); veh_list.RemoveValue(AIVehicle.AT_STATION);
22:53:22 <Zuu> There you have all vehicles at your route that are not on a station / running.
22:53:50 <krinn> :) that's not useful like that
22:54:39 <krinn> and check vehcile speed at 0 + running = stuck at station if location < 3 station distance
22:55:18 <ABCRic> well, I'm gonna catch some Zz's, many AIs for you all. :) 'night
22:55:56 <Zuu> krinn: Yea, many ways to do things. :-)
22:56:50 <krinn> i will keep up the route checking code, i'm just afraid it will be too time consuming with lots of route
22:57:35 <Zuu> That's a good reason to have some higher level checks before running the lower level route check code.
22:58:30 <krinn> that's the reason i think servicing date + reliabilty on vehicle might be a real fast check
23:02:21 <Zuu> You will have to watch out for if old vehicles drop in reliability faster than new ones etc. (I don't know if that is true)
23:03:02 <krinn> oh i'll try to remember that too
23:03:05 <Zuu> Eg, a fast check like that would need a good treshold value to compare with.
23:03:43 <krinn> for now i only check reliability < 50%
23:05:20 <Zuu> The power of collecting statistics and detecting quick changes on the other hand is that you don't need a pre-defined treshold value. You can probably find a few parameters that you can base your treshold on and it might be good enough for most cases.
23:06:06 <Zuu> Some NewGRF can probbaly break some of your assumptions but that's true on many areas for AIs.
23:08:35 <krinn> anything except following the vehicle would do it :)
23:11:51 *** Maarten has joined #openttd
23:18:08 *** Markavian has joined #openttd
23:20:44 <krinn> thank you for suggestions, i will think about your low/high income check
23:22:15 <Zuu> going to bed is a good idea - especially since my bike is at work and I need to use public transport or walk tomorrow.
23:26:42 <__ln__> i wish my bike could do my work for me
23:46:46 *** Devroush has joined #openttd
continue to next day ⏵