IRC logs for #openttd on OFTC at 2007-08-15
⏴ go to previous day
00:00:07 <glx> hmm may be caused by longjmp
00:18:33 *** Dark_Link^ is now known as Dark_Link^sleep
00:45:10 <Eddi|zuHause> knowing Sacro, that is probably a porn link :p
00:49:14 <Eddi|zuHause> the comments are great :p
00:49:58 <CIA-2> OpenTTD: belugas * r10903 /trunk/src/ (4 files): -Feature: [NewGRF] Add support for variable 44 of Variational Action 2 for Industries
00:54:03 *** ChanServ sets mode: +v tokai
01:31:28 *** Eddi|zuHause2 has joined #openttd
01:50:37 <CIA-2> OpenTTD: belugas * r10904 /trunk/src/newgrf_industrytiles.cpp:
01:50:37 <CIA-2> OpenTTD: -Fix: When using a temporary industry, make sure you do not clear some unreachable tiles upon its destruction.
01:50:37 <CIA-2> OpenTTD: -Codechange: do not perform something that by all means will not change. A zero offset is really meaningless...
02:22:17 <CIA-2> OpenTTD: glx * r10905 /3rdparty/squirrel/sqstdlib/sqstdrex.cpp: [Squirrel] -Fix (r9516): incorrect usage of throw
02:43:28 *** mikk36[EST] has joined #openttd
03:24:57 *** iPandaMojo has joined #openttd
03:40:09 *** Nitehawk has joined #openttd
03:49:32 *** Progman has joined #openttd
04:10:41 *** Hendikins has joined #openttd
04:11:54 * Hendikins pads in, one step closer to being a CityRail slave
04:42:04 *** Diabolic-Angel has quit IRC
05:49:33 *** Deathmaker has joined #openttd
05:52:53 *** goddamnit has joined #openttd
06:17:11 *** drscorp has joined #openttd
06:26:25 *** ThePizzaKing has joined #openttd
06:35:22 *** Desolator has joined #openttd
06:42:05 *** Maedhros has joined #openttd
06:54:42 *** mikk36[EST] is now known as mikk36
07:08:13 *** ProfFrink has joined #openttd
07:10:03 *** ProfFrink is now known as Prof_Frink
08:06:41 *** Farden123 has joined #openttd
08:06:42 *** Farden123 is now known as Farden
08:36:32 *** TinoDidriksen has joined #openttd
09:10:24 *** De_Ghost has joined #openttd
09:17:51 *** Vikthor has joined #openttd
09:34:57 *** XeryusTC has joined #openttd
09:42:38 <CIA-2> OpenTTD: rubidium * r10906 /branches/NewGRF_ports/src/ (airport.cpp airport_movement.h newgrf_engine.cpp): [NewGRF_ports] -Codechange: remove oilrig FTA/FSM stuff from the code as that is already exported to a newgrf.
10:08:13 *** Progman has joined #openttd
10:11:31 *** Gekkko` has joined #openttd
10:14:51 <Noldo> What is FTA in this context?
10:18:40 *** RichK67 has joined #openttd
10:19:41 *** Farden123 has joined #openttd
10:20:02 *** Nameeater has joined #openttd
10:24:59 <CIA-2> OpenTTD: rubidium * r10907 /branches/NewGRF_ports/src/disaster_cmd.cpp: [NewGRF_ports] -Fix: the zeppelin disaster did not lock runway.
10:26:28 *** Farden123 is now known as Farden
10:28:59 *** |Jeroen| has joined #openttd
10:31:02 <CIA-2> OpenTTD: rubidium * r10908 /branches/NewGRF_ports/src/ (4 files): [NewGRF_ports] -Codechange: remove the hardcoded TERMn and HELIPADn and the now unused block definitions.
10:43:05 *** Prof_Frink has joined #openttd
10:44:39 *** NW|Aerandir has joined #openttd
11:05:28 *** Viriatha has joined #openttd
11:06:42 *** ProfFrink has joined #openttd
11:08:28 *** ProfFrink is now known as Prof_Frink
11:08:57 *** Brianetta has joined #openttd
11:11:26 *** johnsonp has joined #openttd
11:16:35 *** Thomas[NL] has joined #openttd
11:17:04 *** De_Ghost has joined #openttd
11:34:31 *** Dark_Link^sleep is now known as Dark_Link^
12:04:33 *** Wezz6400 has joined #openttd
12:17:02 <SmatZ> is it possible to insert command line parameters when running ottd with 'make run'?
12:17:43 <SmatZ> when I want to load a savegame, I have to run 'make' and then 'bin/openttd -g ...'
12:28:18 <Rubidium> you could make it an environment variable too
12:30:25 <peter1138> why does make run exist anyway?
12:30:27 <peter1138> is bin/openttd hard?
12:32:08 <ln-> because someone wanted to mess up the whole build system.
12:46:45 *** Hendikins|OUT is now known as Hendikins
13:05:55 *** Ameecher has joined #openttd
13:06:45 *** Rubidium_ has joined #openttd
13:16:40 *** Digitalfox has joined #openttd
13:17:27 *** Nitehawk has joined #openttd
13:20:44 <Eddi|zuHause2> <peter1138> bin/o<tab> <-- b<pgup> ;)
13:23:23 *** Rubidium has joined #openttd
13:30:41 *** Brianetta has joined #openttd
13:35:49 *** Vikthor has joined #openttd
13:44:37 *** Wezz6400 has joined #openttd
13:55:00 <aneb> How do I set up the game on my server to end at e.g. the year 2100?
13:56:59 <aneb> also, how to make the server announce rules?
13:57:04 <Ammler> restart_game_year in section network
13:57:45 <Ammler> we use Autostart from Brianetta for that
13:59:16 <Ammler> aneb: search for autopilot at tt-forum
13:59:47 <Brianetta> autopilot doesn't restart
13:59:54 <Brianetta> in fact, restarting confuses it greatly
14:00:19 <Ammler> ah, also restart over rcon?
14:00:38 <Brianetta> You can do anything over rcon
14:01:00 <aneb> Brianetta: sorry but can u pls guide me step by step?
14:01:39 <Brianetta> I don't have that kind of energy, and it's all written down.
14:02:14 <Ammler> Brianetta: in fact, if you like to use restart_game_year, you can't use Autopilot?
14:03:46 <Brianetta> Ammler: You can, but the saved game will be repeatedly overwritten, and from time to time it'll be confused about who's on
14:07:43 <aneb> Brianetta: why cant u easily fix that? :-)
14:08:26 <Brianetta> aneb: It was designed on a once-through input, process, output basis.
14:28:55 <RichK67> eek... thats horrid... :( given its part of a tileseq/tiletype, im not sure how i would get the action 6 to poke the values... wouldnt a simpler solution be to declare that bit 30 indicates that the recolour sprite is in the newgrf, and just renumber accordingly
14:29:30 <RichK67> yes, it limits recolour sprites to only 32768 varieties... but hey!
14:29:42 <RichK67> in fact to 16384 varieties
14:30:22 <DaleStan> But then you lose the bit for "Draw this sprite normally even in transparent mode".
14:31:02 <RichK67> ah, ok... next bit then... we really dont need a massive number of remaps
14:31:28 *** ProfFrink has joined #openttd
14:32:35 <RichK67> thus for my example, it would be 45 84 90 24
14:33:14 *** ProfFrink is now known as Prof_Frink
14:33:35 <RichK67> bit 29 indicates upper word & 0x1FFF is color map relative to 042D
14:43:30 *** Brianetta has joined #openttd
14:44:41 <DaleStan> RichK67: It's a possibility. Not necessarily a good one, though, since GRM would have to be futzed not to allocate sprites above 1FFFh (currently, the range is ~1324h-3FFFh), and the wider block is used by a whole mess of things in Patch. Everything except vehicles and houses, in fact, which makes it mighty easy for GRM to fail because it's too far down in Patch's newgrfw.cfg.
14:46:37 <Brianetta> DaleStan: Do you use a handheld calculator? If so, which model?
14:46:59 <Brianetta> You strike me as someone who would love the HP-16c, if one could be prized out of the hands of a collector.
14:47:13 <DaleStan> Brianetta: not often, but when I do it's usually my TI-92. (which is a bit of a stretch for "handheld")
14:47:44 <RichK67> hmm... so could you work my example through, as im totally lost on how to change the pointer to the sprite recolour map (and then I'll have the joys of getting OTTD to do it properly)
14:48:17 <Brianetta> If they weren't so damned rare and desirable, I'd have one
14:57:42 * Belugas only uses the calculator that is on his cell phone
14:58:00 <Belugas> apart from the one supplied by Win XP, of course :)
14:58:34 <Belugas> so i guess my requirements are not as elaborate as yours ^_^
15:00:58 <Brianetta> I required one that could be programmed
15:01:00 <DaleStan> RichK67: "work your example"? I'm not sure seeing an example; just some abstract discussion. (And action 6 can poke any number of bytes from 1 to at least 255, at any offset from 0 to 65535.)
15:01:07 <Brianetta> and which could be operated away from the PC
15:07:01 *** Frostregen_ has joined #openttd
15:07:47 <RichK67> ah.. ok... its a simple prop 09 tile layout: 4A 04 00 00 00 00 00 00 00 00 45 84 90 04 80 where 0490 is a sprite recolor map in an action 1 sprite list, at position 93 (63h = 042D+63 = 0490)... the tile is used in an action 0E....
15:10:19 <RichK67> does action 6 poke into a position within the newgrf file? OTTD reads, interprets and stores the newgrfs, rather than leaving them as pokeable files... so I would need to find a solution to that...
15:12:34 *** Frostregen_ is now known as Frostregen
15:14:30 <peter1138> action 6 is supported...
15:15:47 <DaleStan> Oh, but I lied. Action 6 can only poke up to 127 bytes. The high bit of the byte-count is used to enable the "add" operation, not the high bit of the param-num.
15:15:48 *** skidd13 has joined #openttd
15:16:00 <peter1138> although if it's persistent then we've got it wrong
15:18:01 <DaleStan> Does it apply after GRM? If so, and assuming you don't try to support the "add" operation, there's no real problem with apply-once.
15:20:15 <peter1138> then support would be incomplete
15:20:18 <DaleStan> <RichK67> the tile is used in an action 0E.... <-- Did you mean action 0 feature 0D prop 0E? Action E is deactivate/force-activate.
15:21:16 <DaleStan> s/prop\ 0E/prop\ 09/
15:30:40 <RichK67> its defined in a prop 09, its used in a prop 0E... same as newstations would do it... my definition of prop 0E is different, but not for this aspect of it
15:30:41 <DaleStan> RichK67: Anyway, once you can do something with the result of your GRM reservation (Yes, action 6 does specify byte-in-GRF, and it has to remain pokeable at least until the GRM reservations run), it's just lots of long tedious action 6s. I've been meaning for a while to add a feature to NFORenum so it can create those action 6s automagically; maybe now's the time.
15:32:25 <RichK67> ooo... that would be good... although i could probably just refine the graphics to use the standard blue overlay highlight anyway... but its the principle of the thing... atm i cant use a self-defined recolor map on newgrfs... so any solution would be a good thing
15:33:23 <RichK67> is action 6 a "do it once at file load operation?"
15:33:46 <peter1138> it happens during grf load, certainly
15:34:53 <RichK67> ill have to have a look at how then later... but a cleaner solution would do away with having to poke values altogether
15:35:15 <RichK67> but how is another question entirely
15:35:37 <DaleStan> It was, but it also has to happen after the GRM reservations run, or GRM isn't particularly useful. You can use Action D var 99 to adjust the action A, but not the action 0s.
15:36:19 <aneb> why are real checkpoints better than a dummy one tile station??
15:37:11 <glx> aneb: no need to set "non-stop"
15:37:21 <DaleStan> The best idea I've come up with for generating 6s is to detect and overwrite all instances of a distinctive (user-specified) byte or byte-sequence with the specified parameter
15:38:41 <aneb> should I add a note saying that waypoints are not magical pixies for major routing prooblems?
15:38:51 <skidd13> RichK67: The nfo in the seaplaneport needs some adjustment 108 sprites/seaplaneport.pcx 642 343 09 43 34 3 -2
15:40:39 <DaleStan> Actually, scratch that. Something like \w:p0, to write the word (byte/dword by using \b or \d) in param 0 at the current location. That doesn't support three-byte writing, but that's a pretty rare operation, and the main idea is to generate offsets.
15:47:44 *** Tekky4347 has joined #openttd
15:49:10 <Tekky4347> Hi, I have one little question: Are patches that use floating point numbers (types float and double) multiplayer incompatible, because these are implemented differently on different processors?
15:49:32 <Tekky4347> So that there is a danger of a dysync?
15:49:52 <glx> they may have rounding errors
15:50:05 <Tekky4347> Doesn't the realistic acceleration patch also use floating point numbers?
15:52:17 <Tekky4347> Aha. Damn, that is a severe limitation if floating point numbers are not allowed to be used :(
15:53:54 <Tekky4347> Hmmmm, doesn't ANSI C define specific behavior for floating point numbers? I remember an option in MS Visual C++ 6.0 compiler that allowed you to enforce ANSI floating point behavior.
15:54:06 <RichK67> DaleStan: replacing byte sequences is dangerous unless you only replace after/during parsing, otherwise it is incredibly easy to accidentally replace real data
15:56:52 <Noldo> oh hi Tekky, I was wondering when you would get here
15:57:46 <Noldo> Tekky4347: what kind of operations would you need
15:59:12 <Tekky4347> ah, yes, the /Op option in MS Visual C++ disables certain floating point optimizations so that floating-point inconsistency is improved. This compiler option is implied when using /Za (ANSI compatibility). However, this does not improve compatibility between i386 and non-i386 platforms.
15:59:50 <Tekky4347> OpenTTD does also support non-i386-platforms, doesn't it?
16:00:00 <DaleStan> RichK67: Well, I was thinking distinctive sequences, and at "compile-time", not "run-time". Like typing "BE EF" instead of the real data, and then having NFORenum generate an action 6 that'll replace all instances of "BE EF" with the low word of param 0.
16:02:00 <Tekky4347> Noldo: I sometimes prefer working with floating point numbers in certain situations, but it seems this is not possible in OpenTTD because it would break multiplayer compatibility :-(
16:02:46 <DaleStan> But now I think I like \<width>:p<num> better. Simpler tagging, more powerful, and harder to break, all in one.
16:06:17 <Tekky4347> Currently, all the pathfinding is done on all clients, to keep the game in sync, isn't it? Wouldn't it be more meaningful that every client does the pathfinding for its own trains and the results of the pathfinding operation are then transmitted over the network?
16:06:51 <glx> Tekky4347: all clients moves all trains
16:08:28 <skidd13> RichK67: The SF airport has a deadlock and the plans go down too early.
16:10:50 <RichK67> read the commit that introduced the SFO... ;)
16:11:29 <Tekky4347> glx: Ah, yes, TrainChooseTrack() is called when trains are being moved....
16:12:08 <skidd13> RichK67: Oh I only checked out
16:12:12 <RichK67> Tekky4347: ive had to do some really exotic calculations to avoid using floats... even a simple "round a fraction to nearest int" becomes... er... exotic
16:13:31 <RichK67> since you need to retain precision in the decimal places when using an int... had to crack out my 25year old algebra....
16:14:03 <Tekky4347> Hmmm, when a player issued a command in multiplayer, for example build a piece of track on tile xy, then this command is only executed in the tick when all other clients have acknowledged the command? Or how does this work?
16:15:01 <RichK67> skidd13: Only placement works with the SFO. the state machine is incomplete
16:15:58 <RichK67> basically its first cut from my automated conversion routine, but it needs refining to account for the changes in how newgrf_ports move aircraft & handle blocks
16:16:28 <Tekky4347> Unfortunately, I have never played a multiplayer game, yet I must still make my patches compatible for them :(
16:16:39 <RichK67> eg. it uses an explicit z parameter now to indicate what flight level to use (yes, spiralling holding stacks are possible)
16:17:07 <RichK67> so the default z is zero, hence the a/c hitting the deck while still in flight ;)
16:17:08 <Noldo> The command system is quite interesting
16:17:30 <Tekky4347> interesting = complex, I presume? :)
16:18:26 <Noldo> using the word complex would imply some level of undertanding
16:19:18 *** Diabolic-Angel has joined #openttd
16:24:31 <Tekky4347> Hmmmm, ok, I will use the word "complicated" then :)
16:35:22 <Noldo> Tekky4347: It seems that the server will execute the command on this+1 tick and if the clients can't keep up they will disconnect
16:39:41 <Rubidium> the server "schedules" all commands for its next "execute" tick (there's a variable that reduces the amount of execute ticks, which makes the game a little more laggy). It could be that the client is still lagging a few ticks behind.
16:40:08 <Rubidium> furthermore the command is send to the server and then forgotten about; when the server sends the command back to the clients it is actually executed
16:46:11 <Noldo> Tekky4347: how's the pbs thing btw?
16:50:37 *** Deathmaker has joined #openttd
17:03:59 <Belugas> i really wonder what that means " while FFFFFFFFh can be used instead of the GRFID of the current GRF", for var 67,68 of varaction2 for industries
17:04:04 <Belugas> looks like a wild card
17:10:10 *** Dephenom has joined #openttd
17:13:47 <Rubidium> Belugas: that's simple
17:14:14 <Rubidium> 00000000h == original industries, FFFFFFFFh == "my" industries (i.e. the ones defined in "this" newgrf)
17:14:43 <Rubidium> and the other values are GRF IDs of "other" newgrfs that might or might not be loaded at that time
17:15:26 <Belugas> yeah, that at least i understood.
17:16:10 <Rubidium> FFFFFFFFh is just a "shorthand" for referencing the current GRF, which makes using the same code in several GRFs easier and increasing GRFIDs too
17:16:38 <Belugas> but then, it will leave me with the problem of finding what is the grfid of the current callback
17:16:54 <Belugas> don't see it in spritegroup,
17:17:40 <Rubidium> doesn't it have an industry or so?
17:17:51 <Rubidium> let me look at the code
17:18:13 <Belugas> still rusty as it seems :D
17:18:47 <Rubidium> object->u.industry_location.spec->grf_prop.grffile->grfid
17:19:10 <Rubidium> of object->u.industry.spec->grf_prop.grffile->grfid depending on the circumstances
17:19:57 *** Maedhros has joined #openttd
17:23:47 <Belugas> much appreciated Rubidium :)
17:25:38 <Tekky4347> ah, thx for your explanation to my question, too, Rubidium. I only read it now because I was afk.
17:26:45 <Noldo> I don't like the way CommandCost is used as a substitute for Money
17:29:14 *** ChanServ sets mode: +o Bjarni
17:30:50 <Bjarni> can you agree on what you reply? :P
17:31:12 <Tekky4347> Noldo: About your question about my progress on the PBS system: I had originally wanted for a train to reserve its entire route in advance, so that all conflicts between trains could be resolved well in advance. But I have decided that this would be too complicated for now and too CPU-intensive, so I have decided that trains in my PBS system should only reserve a route for the next 2-3 signals, for now.
17:31:25 * Rubidium doesn't see somebody of about 260 years near him... so Bjarni can't be "here"
17:31:27 <Bjarni> so I wonder if it happens on any other platform as well
17:31:31 <Noldo> Tekky4347: I agree with that
17:31:55 <Rubidium> on the other side, if it would be Bjarni's location he wouldn't have to ask on IRC (or he must be terribly shy IRL)
17:32:05 *** Ammller has joined #openttd
17:32:15 <Noldo> Tekky4347: and also it's not a good idea to reserve the path too early in advance because the situation the a far a way block has time to change many times over
17:33:33 <Bjarni> I once got a mail from a guy, who was sitting like 3 meters away from me, so I turned toward him and replied on his question and he went "Reply on the mail dummy"
17:34:13 <Rubidium> Bjarni: can't reproduce that bug here
17:34:23 <Bjarni> best part was that it was a yes/no question
17:34:52 <Bjarni> Rubidium: it seems to happen when the mouse is moved really slowly. When I move the mouse fast, the issue is gone
17:35:05 <Bjarni> Rubidium: and you are using SDL, right?
17:35:06 <Rubidium> well, I moved the mouse slowly
17:35:17 <Phazorx> Bjarni: in this country it is quiter common to send an email with a question/request and follow it with a visit/phone/something of that kind
17:35:54 <Rubidium> Phazorx: because people have a habit to not replying to that mail?
17:35:56 *** Wezz6400 has joined #openttd
17:36:06 <Phazorx> Rubidium: why sent it then ?
17:36:36 <Rubidium> well, *if* you have to call or visit someone physically to get a reply, what's the use of email?
17:36:52 <Phazorx> Rubidium: my point exactly...
17:37:02 <Tekky4347> Noldo: Well, I still do plan on making such a system, but I have now decided to first try to implement a rather simple system. It should still support things like bi-directional double track, though.
17:37:08 <Phazorx> i was getting like "please explain how X works", at same time as a call "I sent you an email, did you see it"
17:38:10 <Phazorx> also, i hate phones for giving a right to someone, at their conviniece to interupt you with thier problems
17:38:26 <Tekky4347> Noldo: In order to make bi-directional double track work, it should be enough for the lookahead of a train to be for the next 2-3 signals.
17:38:32 <Phazorx> pagers/voice mail are much better for recieving side
17:38:58 <Phazorx> however people can not use these either usualy... cuz mesages usualy lack the point
17:40:17 <Bjarni> dammit. I'm at a loss to why this mouse issue happens to mac only, specially because it's the same with both the cocoa and SDL drivers.... do anybody even have a clue to where to start?
17:40:57 <Tekky4347> I have no experience with Macs, sorry :(
17:41:50 <Rubidium> I do, but the people the macs were from didn't like it: crashing a mac within 2 minutes so thoroughly that you needed to remove the battery and unplug the power cord to restart it...
17:43:20 <Rubidium> Bjarni: try the same with another OSX application in full screen and see whether it happens there too, if so -> Not (our) bug
17:43:55 <ln-> Bjarni: btw, did you try if the bug exists with the big patch?
17:45:06 <ln-> Bjarni: blackis's implementation of graphics.
17:45:28 <Noldo> I updatet the ExpensesType through CommandCost thing if it happens to interest someone who is not on .notice
17:45:28 <Bjarni> did he make anything recently?
17:46:25 <Bjarni> ln-: I'm don't think it's a driver issue as the issue is the same with both cocoa and SDL, so modifying the cocoa driver doesn't appear to be the solution
17:47:03 <Bjarni> if it were the cocoa driver only, then it would be easier to track this
17:47:45 <ln-> Bjarni: but doesn't the current fullscreen implementation handle things quite differently than the version by blackis, since e.g. currently apple-tab and all other key combinations are captured before they get to the OS.
17:48:29 <peter1138> one day, bjarni will learn to use the word 'does'
17:48:45 <Bjarni> however I think it was dead slow and it contains endian issues
17:49:27 <ln-> Bjarni: interesting; since i think it was somewhat slower, but definitely not "dead slow".
17:50:03 <Bjarni> I think I tried to use it in fast forward and the speed didn't change noticeable
17:50:25 <ln-> Bjarni: still, could be worth checking if the mouse bug is reproduceable with that patch.
17:52:33 <Noldo> hylje: is your patch tracking thing ready?
17:56:17 <hylje> however i have little time to spend on it
17:57:34 *** De_Ghost has joined #openttd
17:58:16 <Bjarni> ln-: the diff is outdated and can't be applied :(
17:58:22 <Noldo> I wonder that the other thing I wanted to tinker with was, oh yes ship-pf
18:03:52 <Tekky> Damn, I just had a power failure. Did anyone write me a message in the mean time that I did not respond to? Noldo, did you maybe reply to my messages?
18:04:34 <Noldo> Tekky4347: I said 'yes'
18:05:19 <Noldo> Tekky: can you make it so that the number of blocks reserved is easy to change?
18:05:48 <Tekky> Yes, that should be easily possible.
18:07:07 <Tekky> I don't work with blocks, though, I work with signals.
18:08:00 <Tekky> My pathfinder stores its path as which signals to take. The path to take between two signals is predefined and cannot be changed dynamically.
18:08:16 <Tekky> This makes pathfinding easier.
18:09:17 <Rubidium> that reminds me: Tekky, how can there be less signal blocks than YAPF segments? Long pieces of track with a lot of signals would be one segment, but lots of signal "blocks"
18:09:32 <Rubidium> And why can't there be multiple paths between two signals?
18:11:44 <Rubidium> I can think of some track layouts where you want to be able to have multiple paths between two signals, so you can have multiple trains running over the same junction (in several ways)
18:13:32 <Tekky> I don't know the internals of YAPF that well... a straight piece of track with many one-way signals on it is considered by YAPF to be one single segment, is that correct?
18:13:32 <Sacro> Rubidium: because long lines with no junctions can be classes as a long block
18:14:00 <Rubidium> Sacro: then it is not a signal block
18:14:33 <Rubidium> Tekky: even with two way signals it is. Segments go from junction to junction
18:17:09 <Tekky> Is 8 the correct answer?
18:17:47 <Rubidium> don't know, KUDr should know it
18:18:42 <Tekky> No, 10, I think. The entry and exit line, the two station platform lines and the junction tile containing 6 segments.
18:21:42 <Tekky> My pathfinder doesn't work with segments, it works with "signal hops" or, more precisely, "waiting location hops", where a waiting location can be any possible location where a train can wait, such as a signal, station or depot. I think it is better to work with signal hops than segments when you have many junctions.
18:22:51 <Tekky> If you have a straight piece of single track with many signals on it, YAPF may perform better. But I plan to make bi-directional double track the norm :)
18:24:23 *** Brianetta has joined #openttd
18:24:28 <Tekky> With bi-directional double track, there is a junction after every signal.
18:25:23 <Tekky> I want to only allow one path between two signals to make the pathfinder more efficient.
18:26:43 <Tekky> Then it can all be precalculated which "signal hop" conflicts with which "signal hop". It would cost a lot of CPU resources for this information to be recalculated all the time.
18:29:28 <Rubidium> ln-: on the Avinguda Diagonal looking at the Universitat Politecnica de Catalunya in Barcelona?
18:30:57 *** Farden123 has joined #openttd
18:31:21 <Tekky> However, maybe you are right that I should use individual segments instead of "signal hops" and thereby not predefine the path between two signals, because this would give the pathfinder more options to find a route that doesn't conflict with another train.
18:32:35 <Tekky> I am still undecided about many of these design issues.
18:32:41 <Rubidium> :O my google skill still work ;)
18:32:59 <Rubidium> never been to Barcelona (or even Spain)
18:33:43 <Rubidium> (or even heard of that university)
18:34:13 *** Farden123 is now known as Farden
18:36:58 *** raimar3 has joined #openttd
18:44:45 *** Illegal_Alien has joined #openttd
18:45:30 <Tekky> The reason why I prefer using "signal hops" instead of segments is that the number of segments is higher than the number of "signal hops" and therefore the pathfinder must work with more nodes. If I use segments, I must consider the track before a signal and after a signal as a seperate segment, because when the track before the signal is in use I must lock different segments than when the track after the signal is in use.
18:46:16 <Tekky> Therefore, I cannot work with YAPF segments, because these segments aren't divided by signals.
18:47:29 <Tekky> Rubidium, that is why the number of nodes is smaller with "signal hops" than with segments.
18:47:53 <Tekky> because the segments I would have to use for my pathfinding are different to YAPF segments.
18:51:24 <KUDr> [20:17:02] <Tekky> Is 8 the correct answer? << depends on the origin
18:51:31 <Belugas> Tekky, maybe i'm wrong but i think that tghe number of signal hops depends much more on the density of his signalling system...
18:52:24 <Belugas> while the segment density is always constant, given the same system
18:53:59 <Belugas> therefor, a user who litteraly bombard his railway system with signals will crumble yoyur pathfinding
18:58:17 <Tekky> Well, yes, if someone puts a signal on every single track tile, I will have trouble :)
18:59:35 <Belugas> it is a behaviour i've seen frequently. I do it myself from time to time ;)
18:59:44 <Belugas> well... during my tests, of course
18:59:51 <Belugas> since i do not play anymore !
19:01:49 <Tekky> hehe, I don't play much anymore, either, because I can't stand the game without my new PBS pathfinding system :)
19:02:11 <KUDr> Tekky: 10 is too much. I would tell 3 if train comes from bottom-left corner
19:02:38 <Tekky> how many cached segments will be in memory in your screenshot?
19:02:47 <Tekky> I mean in my screenshot.
19:03:25 <Tekky> Don't you cache segments independent of trains?
19:04:22 <KUDr> 1. incomming from left, outgoing to south, and two platforms
19:05:08 <Tekky> don't you have to add 6 segments for the junction tile?
19:05:09 <KUDr> train position makes difference on whether it is cached or not
19:05:27 <KUDr> only choice generates new segment
19:05:57 <KUDr> and you have 2 choices there
19:06:32 <KUDr> trains going from station are too close to use cached segments
19:06:52 <KUDr> due to signal lokk-ahead
19:07:38 <Tekky> Hmmmmm, what is your definition of a segment? A piece of track with no switches(=choices) on it? Or do you define a segment as a path of a train, i.e. a pathfinder node?
19:08:41 <KUDr> segments close to origin are calculated on demand
19:08:53 <KUDr> as they generate different costs
19:09:42 <Tekky> So a YAPF segment also contains the travel direction?
19:09:56 <Noldo> sleep or talk about pathfinding
19:10:40 <Tekky> so you would have two different segments cached for two different directions on the same piece of track?
19:11:33 <Tekky> aha, now I am beginning to understand :)
19:12:17 <Tekky> Thx for your explanations.
19:13:57 *** Diabolic-Angel has quit IRC
19:14:06 <Tekky> That would mean that there are 20 different segments in the screenshot, if you take all possible train positions into account?
19:16:06 <Noldo> KUDr: I could look in to the regional pathfinding if have the time to give me some instruction about what is unfinished about it
19:16:50 <KUDr> Noldo: you must take instructions and sources from JazzyJaffa
19:17:16 <KUDr> my role in that was only 'consultant'
19:17:18 <Tekky> ah, I counted the entry and exit path twice, despite them being one way.... so I should have said 18 instead of 20.... I don't understand why you say 11, though.... hmmmm, I will make a list of all segments I see. Just a moment.
19:17:38 <Tekky> You can tell me then which ones are wrong...
19:18:22 <KUDr> as it ends on first signal
19:18:44 *** Diabolic-Angel has joined #openttd
19:19:28 <KUDr> the last is not bottom but right
19:20:06 <KUDr> butthey are not all cached
19:21:01 *** [BDS]-Klaus has joined #openttd
19:21:44 <Tekky> hmmmm, you are not counting segments of illegal routes, i.e. ones that end at the reverse side of the one-way signal, are you?
19:22:56 <KUDr> only if it continues behind such signal i ommit it
19:23:35 <KUDr> so from bottom the PF stops at first signal
19:24:57 <KUDr> the theory is very simple: make segment only when one of the following conditions is met:
19:27:29 <Tekky> hmmmm, this is my list of 18 segments:
19:27:36 <Tekky> 3. top station platform entry
19:27:36 <Tekky> 4. top station platform exit
19:27:36 <Tekky> 5. bottom station platform entry
19:27:37 <Tekky> 6. bottom station platform exit
19:27:37 <Tekky> top-left junction tile:
19:27:39 <Tekky> 7. trackdir bottom-left to top-right
19:27:45 <Tekky> bottom-right junction tile:
19:27:49 <Tekky> 17. botton-left to top-right
19:28:05 <Tekky> sorry for the long paste :)
19:28:14 <hylje> try describing some coop stations
19:29:09 <KUDr> 8. reverse of 7 << why that?
19:30:00 <Tekky> you said you store one segment for each direction?
19:30:19 <KUDr> when going from bottom platform
19:33:21 <Tekky> Did you maybe not count 8 because that segment would cause the train to have to reverse at the one-way signal?
19:33:41 <KUDr> i didn't use your method
19:33:59 <KUDr> but i tried to simulate train at 4 corner positions
19:34:17 <KUDr> i.e. from left it generates 5 segments
19:34:47 <Tekky> so are my 18 segments YAPF segments?
19:35:13 <KUDr> it depends on definition
19:35:25 <KUDr> segment is depended on train position
19:35:54 <KUDr> so if you count 3x5 + 1 it is 16
19:37:36 <Tekky> I wanted to build a cache of segments independant of train position, for use by the pathfinder.
19:37:47 <KUDr> YAPF generates segments on the fly
19:38:05 <KUDr> but not all segments are cached
19:38:21 <KUDr> so if you are talking about cache, there are only 4
19:38:22 <Tekky> you only cache frequently used ones?
19:38:42 <KUDr> i can't cache close segments
19:38:59 <Tekky> close segments? You mean short ones?
19:39:06 <KUDr> so the only entry that can generate cached segments if from left
19:39:42 <KUDr> segments close to train == less than 10 signals passed
19:40:24 <Tekky> "from left"? You only cache segments when the train is travelling from left to right? I don't understand.
19:40:44 <KUDr> segments close to train == less than 10 signals passed
19:41:10 <KUDr> so only train coming from left can generate cached segments
19:41:26 <KUDr> all other trains are too close
19:41:40 <KUDr> so they generate 'local' segments
19:42:05 <KUDr> depending on signal states
19:42:15 <KUDr> what is not clear on it?
19:42:18 <Tekky> ah, are cached segments used as an approximation for pathfinding distant segments and you use NPF-style pathfinding for close segments?
19:43:03 <KUDr> YAPF segment is from junc to junc (with some exceptions)
19:43:11 <KUDr> but not all of them are cached
19:44:15 <Tekky> hmmmm, just a moment, I am reading everything you said again, I must have misunderstood something....
19:48:19 <KUDr> do you mean "cached segment cost for rail YAPF"?
19:48:28 <KUDr> this is the only info i see there
19:52:02 <Xintron> Hey guys, should it be "the none working game" or "the not working game"
19:52:59 <Xintron> hrmm... "the non-working game"?
19:53:44 <KUDr> hope you are not talking about openttd ;)
19:54:11 <Xintron> haha, I would never say anything like that about openttd
19:54:22 <Xintron> But is the last one I posted correct?
19:54:55 <KUDr> i am czech so all versions are correct here :)
19:55:15 <Xintron> yeah, I'm swede so I accept all :)
19:56:13 <Belugas> Xintron, it depends of the context, i think
19:56:29 <Xintron> The whole sentence would be "NukeZone the non-working game"
19:57:21 <Rubidium> "NukeZone, the game that doesn't work"
19:58:07 <Belugas> none working game, then, yes
19:58:30 <Xintron> corrupt could work :/
19:58:34 <Rubidium> none working != non working
20:03:27 <Phazorx> non-working is kinda wage... could be corrupt, faulty, erroneous, frozen, etc
20:04:06 *** stillunknown has joined #openttd
20:04:44 <Phazorx> and "none" = not one, while "non-" is a negation
20:05:37 *** De_Ghost has joined #openttd
20:21:17 *** Vikthor has joined #openttd
20:55:02 *** RichK67 has joined #openttd
21:10:01 <CIA-2> OpenTTD: richk * r10909 /branches/NewGRF_ports/ (4 files in 3 dirs):
21:10:01 <CIA-2> OpenTTD: [NewGRF_ports] -Feature: Added Metropolitan airport to airportsextended.grf. Also added remaining minipics to airportsextended.pcx.
21:10:01 <CIA-2> OpenTTD: Metropolitan airport loads from old games.
21:11:48 <Prof_Frink> hylje: Yes. Airport sex. That's how the new designs are going to come about.
21:16:55 <Wolf01> this reminds me two utilities i used to hack the gameboy roms: script inserter and extractor... in short: sinserter and sextractor o_o
21:17:59 <Wolf01> and after 3 years i had a bad surprise when i tried to find pr0n inside their archives
21:19:41 <Wolf01> like penisland.com "we sell the best pens ever"
21:19:45 <Prof_Frink> sex tractor? oo arr.
21:21:00 <hylje> i wonder if penisland is deliberate
21:22:22 <Wolf01> i'm confused, why isn't S.a.c.r.o joining the converstation? (the dots are to avoid recall him)
21:22:30 <Prof_Frink> s/wonder if/suspect/
21:31:12 <Belugas> i was wondering if you were still alive :)
21:31:27 <Prof_Frink> He's abandoned us :(
21:31:51 <Prof_Frink> No, silly, Sir peter of 1138
21:32:46 <Prof_Frink> What up, Sir Sac of ro?
21:36:22 <Bjarni> why do we need a proof of Frink?
21:36:47 <Bjarni> we already believe that it's out there
21:38:30 <Bjarni> I think I detected Belugas present in here :s
21:38:49 <Bjarni> better be safe than sorry
21:39:26 <Bjarni> ohh that reminds me of a great story from the news today
21:40:25 <Bjarni> some woman went to the bank. She put on a mask because she has some sort of allergy to airborn particles. Somebody mistook it as a disguise and called the police
21:41:00 <Dephenom> can never be too careful
21:41:17 <Bjarni> so she was caught by the police, who had to get all sorts of personal info (and verifying it) because of the allergy
21:41:23 <Bjarni> that's some nasty allergy
21:42:32 <Bjarni> Sacro: good one. Did you find it while browsing for boobs?
21:43:15 <DaleStan> peter1138: Can you verify my long-date conversions, please? The date 1/1/1, in a long intro date property, is 381 (7D 01 00 00)? And 31 Dec 65535 is 23936546 (22 3E 6D 01)?
21:43:17 <Dephenom> Sacro, where did you find that last one?
21:43:21 <Bjarni> however I guess it's a US sticker and they will be sued big time if they don't write it
21:44:07 <glx> they probably have been sued already so they write it now
21:44:07 <Bjarni> Dephenom: are you sure you want to know the answer to that question?
21:44:42 <Bjarni> glx: yeah... normally people wouldn't imagine such usage
21:48:50 <peter1138> DaleStan: i don't actually know. i didn't write the conversion stuff
21:49:23 <DaleStan> Do you know who I should ask, then?
21:53:36 <Maedhros> DaleStan: Rubidium is the best person to ask, i think
21:55:49 <Rubidium> DaleStan: 01/01/1: 0000016e, 31/12/65535: 016d3e13 (that's the result if I call ConvertYMDToDate with the give dates, which is correct because ConvertDateToYMD on the result of YMDToDate gives the input given)
21:56:30 *** goddamnit has joined #openttd
21:57:04 <Dephenom> Bjarni, i dont drive so id that the pedestrian underpass :P
21:58:04 <DaleStan> And I think that's off by 15 in both cases. Good. Because I didn't read the wiki quite close enough and used 701280, not 701265, for 1 Jan 1920. That makes me happy.
22:01:38 <peter1138> hemel hempstead's magic roundabout is more regular, but has 6 individual roundabouts
22:03:12 <Bjarni> Sacro: this one is for you too
22:03:31 <Sacro> aren't they made out of smaller bricks?
22:04:19 <Bjarni> this is about the smallest bricks and how they are made
22:05:31 <Prof_Frink> You cannot "split the brick"
22:08:29 <Sacro> Bjarni: they mine them in the darkest forests of daneland
22:10:44 <Prof_Frink> The brick is the basic unit of matter. If you were to "split" it, it would unleash a huge amount of energy.
22:10:58 <Bjarni> don't spoil our online decor designed to fool people
22:13:17 <Bjarni> wanna do something fun?
22:19:01 <Bjarni> some guy downloaded too much and took down the entire internet (or so they claim). The country that was hit the hardest was Nigeria as 94% of their GNP was from internet ventures XD
22:34:01 <CIA-2> OpenTTD: rubidium * r10910 /branches/NewGRF_ports/src/aircraft_cmd.cpp: [NewGRF_ports] -Fix: take foundations into account when determining the height of the 0,0 position of the airport.
22:56:20 <Tobin> Bjarni: What fun thing?
22:56:35 <Tobin> Bjarni: i'm about to go on a short holiday so it had better be _very_ fun.
23:07:19 <CIA-2> OpenTTD: richk * r10911 /branches/NewGRF_ports/ (6 files in 3 dirs):
23:07:19 <CIA-2> OpenTTD: [NewGRF_ports] -Change: restored opntitle.dat by adding basic International airport functionality.
23:07:19 <CIA-2> OpenTTD: Please note: motion on International is not debugged, and it will not work properly. It is sufficient that it loads for now.
23:08:14 <RichK67> and now its time for bed...
23:13:50 *** Dark_Link^ is now known as Dark_Link^sleep
23:18:31 *** Dephenom has joined #openttd
23:22:51 <CIA-2> OpenTTD: richk * r10912 /branches/NewGRF_ports/bin/data/ (airportsbasic.grf sprites/airportsbasic.nfo): [NewGRF_ports] -Feature: Added dummy holding stack to SYST class.
23:32:01 <Bjarni> at least that's what I did all day :s
23:45:19 <CIA-2> OpenTTD: richk * r10913 /branches/NewGRF_ports/bin/data/ (airportsbasic.grf sprites/airportsbasic.nfo): [NewGRF_ports] -Fix: Dummy state machine should be available to all aircraft, rather than none.
23:56:09 <CIA-2> OpenTTD: richk * r10914 /branches/NewGRF_ports/bin/data/ (airportsbasic.grf sprites/airportsbasic.nfo): [NewGRF_ports] -Fix: Dummy state machine positions modified.
23:57:10 <DaleStan> Obviously, when RichK67 says "time for bed", that's not quite what he means.
continue to next day ⏵
