IRC logs for #openttd on OFTC at 2021-01-04
⏴ go to previous day
00:39:57 <snail_UES_> the “veh_num” function in m4nfo, which counts the number of vehicles in the consist (or from a certain vehicle on), displays the number *minus one*, right?
02:02:58 *** DasPoseidon has quit IRC (Quit: DasPoseidon)
03:19:59 *** Wormnest has quit IRC (Quit: Leaving)
03:22:53 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
04:06:23 <DorpsGek> [OpenTTD/OpenTTD] perezdidac opened pull request #8495: FR: Prevent tunnels from being built under extractive industries https://git.io/JLFsM
04:08:55 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFs5
04:08:59 <DorpsGek> [OpenTTD/OpenTTD] perezdidac closed pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFsM
04:09:02 <DorpsGek> [OpenTTD/OpenTTD] perezdidac reopened pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFsM
04:12:39 <DorpsGek> [OpenTTD/OpenTTD] James103 commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFsp
04:15:09 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFGf
05:12:15 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFZH
05:47:04 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
06:01:36 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFcZ
06:03:30 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFcC
06:13:12 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFco
06:28:29 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFCG
07:30:45 *** sla_ro|master has joined #openttd
07:47:33 *** WormnestAndroid has quit IRC (Remote host closed the connection)
07:47:47 *** WormnestAndroid has joined #openttd
08:03:26 *** WormnestAndroid has quit IRC (Remote host closed the connection)
08:03:40 *** WormnestAndroid has joined #openttd
08:06:45 *** snail_UES_ has quit IRC (Quit: snail_UES_)
08:07:18 <DorpsGek> [OpenTTD/OpenTTD] erenes commented on issue #8194: Releases from version 1.10.1 no longer work on mid-2007 iMac https://git.io/JfMJz
08:08:48 *** andythenorth has joined #openttd
08:29:33 *** Gustavo6046 has joined #openttd
08:41:28 *** DasPoseidon has joined #openttd
08:52:43 *** crem has quit IRC (Quit: WeeChat 3.0)
09:07:35 <DorpsGek> [OpenTTD/OpenTTD] LordAro dismissed a review for pull request #8452: Doc: Add labels to landscape grid description. https://git.io/JL555
09:29:41 *** DasPoseidon has quit IRC (Ping timeout: 480 seconds)
09:34:30 *** DasPoseidon has joined #openttd
09:54:35 <andythenorth> lol things that will never get old
09:54:51 <andythenorth> people in forums (not just tt-forums) posting 'is there any news in this thread since last post'
09:55:04 <andythenorth> it's nice for morons to self-identify
10:12:12 *** urdh has quit IRC (Quit: Boom!)
10:18:35 <DorpsGek> [OpenTTD/OpenTTD] michicc commented on issue #8194: Releases from version 1.10.1 no longer work on mid-2007 iMac https://git.io/JfMJz
10:32:25 *** jottyfan has joined #openttd
10:33:09 *** ChanServ sets mode: +v tokai
10:50:24 *** jottyfan has joined #openttd
10:54:05 <Borg> any YAPF experts around? I have like a path w/ 4000 penalty.. and those litle bastards still prefers it.. ;)
10:54:48 <Borg> lets see if 6000 will discourage them...
10:55:51 <Borg> and even if path is blocked... they want to go there...
10:56:17 <LordAro> you are absolutely on your own when fiddling with yapf penalty values
10:59:16 <Borg> I noticed 2 oddities w/ penalites generally..
10:59:36 <Borg> big penalty makes path unusuable..
11:00:15 <Borg> even if everything is blocked.. completly.. w/ is weird. because I couldnt find that in code
11:01:46 <Borg> not fully of course.. still some trains go to aux path.. where pri is free
11:01:59 <Borg> lets enable pathfinder display
11:04:11 <Borg> what was pf.reserve_paths was for?
11:06:02 <Borg> ahh always reserve path..
11:09:50 <Borg> okey.. time to RTFS again.. and debuglevel yapf=3
11:14:04 <michi_cc> You don't happen to have two-way signals somewhere, do you?
11:14:10 <michi_cc> Any maybe two_eol on?
11:16:36 <Borg> michi_cc: nope.. I dont use two-way signals at all.. and two_eol is off
11:16:44 <Borg> I just use PBS + oneway block
11:16:49 <Borg> michi_cc: wanna screenshot?
11:17:44 <Borg> and I use trick w/ PBS back signals.. to influence pathfinder.. to have primary path and auxliary path
11:19:08 <Borg> hmm train going to aux path.. cost is 23568.. still selected
11:20:18 <Borg> its always 0.. not used in YAPF?
11:24:13 <Borg> now.. when I have 9000 penalty over aux path.. pri is finaly selected.. with cost 14272
11:24:28 <Borg> so.. aux path cost is now sth like 26000+
11:24:56 <Borg> so... 14272 vs 24000 yet... 24000 is selected.. why?
11:25:42 <Borg> I was studing that code.. few times.. and seems cost is importand part of path selection.. of course. if path is blocked by train... anything is better..
11:26:02 <Borg> but if 2 paths are empty.. not reserved by other trains... we have 14272 vs 24000 and yet.. 24000 is selected
11:26:06 <Borg> with is completly weird to me..
11:46:24 <Borg> okey. I found the reason.. with makes me even more puzzzled
11:47:16 <Borg> at station, I have cost 13094, but later on that path.. on first PBS signal cost is 23568
11:49:36 <Borg> okey.. got entire dump...
11:49:42 <Borg> for whole path it looks like this
11:50:51 <Borg> cost: 13094 -> 23568 -> 22852 -> (passing 2x PBS back signals, 6000 penalty) -> 9381
12:13:46 <Borg> and yeah.. whole path is <10 signals.. so signal look ahead.. should kick in
12:24:11 *** andythenorth has quit IRC (Quit: andythenorth)
12:25:36 <Borg> what is... segment in pathfinder?
12:47:08 <Borg> when I increase look_ahead_signals = 20
12:47:13 <Borg> correct path is selected
12:47:50 <Borg> if (n.m_num_signals_passed >= m_sig_look_ahead_costs.Size() / 2) return 0;
12:47:59 <Borg> in ReservationCost() function
12:48:04 <Borg> wonder why its half of its size
12:48:09 *** urdh has quit IRC (Quit: Boom!)
12:50:43 *** sla_ro|master has quit IRC ()
12:54:36 <FLHerne> Borg: I think 'segment' is a section of plain track between junctions/signals
12:55:05 <FLHerne> So the pathfinder can treat it as a single unit
12:55:29 <Borg> FLHerne: I doubt it.... in code I see stuff like this:
12:55:41 <Borg> segment_cost += Yapf().SignalCost(n, cur.tile, cur.td);
12:55:51 <Borg> all look ahead signals are added..
12:56:16 <Borg> there is also one weird thing
12:56:32 <Borg> if (segment_cost > s_max_segment_cost) {
12:56:44 <Borg> with is worrying.. because s_max_segment_cost is hardcoded to 10000
12:58:30 <Borg> but still.. its weird.. because increasing signals lookup ahead... make it works fine.. so.. im really puzzled
13:07:09 <Borg> okey lets try something stupid
13:08:58 <Borg> as expected.. stupid thing makes things... bad :)
13:12:06 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8485: Codechange: improve performance for complex vehicle chains by resolving sprites less often https://git.io/JLF6I
13:17:04 <Borg> okey.. I think I found something..
13:17:11 <Borg> with badly influences my costs
13:17:38 <Borg> pbs_cross_penalty <- can anyone explain me.. what it really is?
13:18:10 <Borg> I increased it.. so trains nicely move into platforms when there are some corssing occupied
13:18:19 <Borg> but.. it seems now.. it destroy things in other places
13:21:07 <Borg> time to make test tracks
13:24:44 *** andythenorth has joined #openttd
13:27:58 *** jottyfan has joined #openttd
13:28:08 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
13:28:22 *** WormnestAndroid has joined #openttd
13:30:51 <Borg> changing pbs_crossing to default doesnt change anything
13:33:19 <Borg> now.. doing close inspection on paths..
13:33:33 <Borg> im even more and more puzzled
13:35:25 <Borg> diagnol paths are more expensive?
13:41:26 <Borg> so its best to AVOID them....
13:44:49 <Borg> now is all clear why other patch is still prefered..
13:44:56 <Borg> there is quite long diagonal track in pri path...
13:45:07 <Borg> with gets probably around 1500+ cost
13:45:19 <Borg> what was the reasoning to penalize diagonal tracks?
13:47:07 <Borg> okey. but there is YAPF_TILE_CORNER_LENGTH=71
13:55:06 <TrueBrain> LordAro: nice :D We basically build that ourselves, but not on FUSE level :P
13:55:27 <LordAro> FUSE level is a bit cursed
13:55:41 <TrueBrain> I rather like that we decide when to commit/push, honestly :)
13:55:57 <TrueBrain> but I understand why they did it :D
13:58:15 *** snail_UES_ has joined #openttd
14:01:41 *** sla_ro|master has joined #openttd
14:06:34 *** Con_TheGranny has quit IRC (Remote host closed the connection)
14:10:11 <Borg> guys.. I really need help w/ that
14:10:42 <Borg> Im trying to understand why my short path.. with like 3-4 signals is sooo expensive
14:12:01 <LordAro> Borg: here's what you should do (that i think i've told you before) :
14:12:07 <LordAro> 1) remove all pathfinder settings from your openttd.cfg
14:12:10 <LordAro> 2) don't touch them again
14:12:45 <Borg> because default settings do NOT work
14:13:06 <LordAro> i can almost promise that the number of people who have actually changed any of those settings in a useful way in the last 15 years can be counted on 1 hand
14:13:15 <LordAro> no one knows what those values are, or what they are doing
14:16:32 <Borg> okey.. lets set them to default
14:24:30 <Borg> yeah... MAGIC.... does NOT fucking work..
14:26:12 <Borg> adding 4500 penalty on path..
14:26:40 <Borg> lets see if its persistent
14:27:32 <Borg> difference between paths cost is very small.. weird
14:27:43 <Borg> aux path have +4500 penalty additionally
14:27:57 <Borg> is there something obvious I dont see?
14:30:08 <TrueBrain> I somehow expected to be rickrolled :)
14:32:02 <Borg> again.. path w/ cost 11594 selected
14:32:02 <Eddi|zuHause> do they talk about how on patch-day the government has to turn off the projector that simulates the moon on the sky?
14:32:23 <Borg> where 6062 is empty.. no reservations.. green lights..
14:32:27 <TrueBrain> didn't you hear they are making tinfoilhats mandatory?
14:32:46 <Eddi|zuHause> why aren't there any people wearing tinfoil masks
14:33:33 <andythenorth> "but you hate magic"
14:33:40 <andythenorth> "yes but it has a small carbon footprint"
14:34:52 *** DasPoseidon has quit IRC (Ping timeout: 480 seconds)
14:36:47 <Eddi|zuHause> oh man that's like a conspiracy gold mine :p
14:37:04 <Eddi|zuHause> and this is from 2009?
14:37:15 <Borg> okey.. so noone wanna take a look at my save?
14:40:03 <FLHerne> (if it doesn't have a hundred unobtainable grfs)
14:40:59 <Borg> if (m_pBestDestNode == NULL || n < *m_pBestDestNode) { <- uhm.. n is Node (is it class?) can we do comparision like this?
14:42:14 <Borg> FLHerne: nah, just too small GRFs, they are here:
14:42:17 <Borg> ftp://ds-1.ovh.uu3.net/home/borg/openttd/BSPI.grf
14:42:22 <Borg> ftp://ds-1.ovh.uu3.net/home/borg/openttd/X2025.grf
14:42:31 <Eddi|zuHause> oh, i remember why i made an ignore list... they keep having the obscurest problems and don't take the sane advice
14:44:33 <LordAro> are pathfinder settings even saved?
14:44:45 <Eddi|zuHause> almost certainly
14:44:56 <Borg> they are in save game.. but... now they are all default
14:45:13 <Borg> with in that case have no meaning
14:54:08 *** jottyfan has quit IRC (Quit: jottyfan)
15:11:00 *** DasPoseidon has joined #openttd
15:24:38 <supermop_Home> andythenorth ocean water cycle for pool like temperate hotel?
15:30:03 <Borg> FLHerne: hey, are you going to take a look? or no time atm? I have things setup and waiting
15:30:07 <Eddi|zuHause> because pools are known for meter-high waves?
15:32:26 <supermop_Home> Eddi|zuHause the rooftop pool in the temperate base set never looked all that pleasant to swim in
15:32:55 <supermop_Home> though it was one of those things i noticed as a fascinating detail when i was a kid
15:37:19 <Timberwolf> The one that used to drive me made in temperate base was the passenger carriage.
15:37:36 <Timberwolf> The one in arctic/subtropical is so much nicer.
15:39:37 <andythenorth> supermop_Home blue, with a few ocean dots
15:39:51 <TrueBrain> i'm blue daladilada
15:40:37 <Borg> FLHerne: and segment is probably between junction/branches.. can contain many signals...
15:41:30 <supermop_Home> TrueBrain i think it goes da BA di da ba dai
15:42:12 <Eddi|zuHause> some sentences are worse than rickrolling :p
15:42:16 <TrueBrain> turns out it is daba dee ba da die
15:42:30 <Eddi|zuHause> there's a "da" missing
15:42:44 <TrueBrain> la and ba are very hard to differentiate, but yeah
16:03:40 <supermop_Home> i guess landscape architecture is not really my thing - really lackluster groundsprites so far here
16:08:16 *** _2TallTyler has joined #openttd
16:11:13 <supermop_Home> andythenorth do you put greenery on the sprite or let grass show through from a baseset tile?
16:11:32 <andythenorth> I draw greenery usually
16:11:44 <andythenorth> depends how terrain dependent I wanted it to be
16:14:42 <supermop_Home> i guess a hotel has a grounds crew
16:15:33 <supermop_Home> do you think there is enough for ogfx now?
16:16:30 <FLHerne> Borg: What's the issue supposed to be, then?
16:16:56 <FLHerne> (sorry, I tend to squeeze in IRC between other things...)
16:17:40 <FLHerne> I have the save open, nothing looks obviously stuck
16:18:04 <LordAro> FLHerne: you should polish your crystal ball
16:20:58 <Borg> okey.. some final tought.. with yapf.rail_look_ahead_max_signals = 10 YAPF is completly not deterministic to play.. in that track layout (default settings or mine settings)
16:21:07 <Borg> changing that to 20.. makes thing works right!!
16:21:27 <Borg> FLHerne: please redownload Rand4_tmp1.sav
16:21:35 <Borg> because I went futher with tests..
16:21:46 <Borg> so I copied that snapshot..
16:22:15 <Borg> FLHerne: so if you have time now.. we can go step by step
16:23:03 <Borg> okey, let me open it too
16:23:31 <Borg> also, please set show reserved paths
16:23:56 <Borg> train 37 now selected path is kinda wrong
16:24:15 <Borg> it destination is Pratbourne Halt
16:24:43 <Borg> it should chose that path straight on that junction under bridge
16:24:51 <Borg> its shorter.. it doesnt have 4500 penalty
16:25:05 <Borg> and there is nothing wrong with it.. its also.. EMPTY
16:25:29 <Borg> if you follow the path the train will go now..
16:25:52 <Borg> you will see PBS signal after bridge and junctions... and 3 PBS back signals
16:25:59 <Borg> this is my aux path.. w/ should be selected rary
16:26:15 <Borg> or just unpase the game.. and track the train
16:26:45 <Borg> if you got it.. you can just unpause the game.. and watch most trains go that way
16:29:45 <Borg> you can also play with: debuglevel yapf=3
16:29:55 <Borg> and pause game in right moment and check paths costs..
16:29:59 <FLHerne> I can see it, that is a little odd
16:30:13 <Borg> I still didnt fugured out what is open vs closed
16:30:30 <Borg> yeah.. very odd.. especially. if you cound all signals on path.. they are <10
16:31:56 <Borg> now.. change the setting: yapf.rail_look_ahead_max_signals = 20
16:32:02 <Borg> and they should now select correct path
16:33:19 <FLHerne> LordAro: fwiw, the issue is reproducible with default PF settings and doesn't seem obviously explicable
16:33:58 <Borg> the only place I found that MIGHT influence this:
16:35:08 <Borg> trying to find it.. now..
16:35:16 <LordAro> presumably needs some reducing to a minimal example
16:35:28 <Eddi|zuHause> it goes away if you turn the intermediate signals into path signals
16:35:44 <FLHerne> Eddi|zuHause: hah, that's exactlt what I was just doing
16:36:16 <Borg> got it: if (n.m_num_signals_passed >= m_sig_look_ahead_costs.Size() / 2) return 0;
16:36:23 <Borg> but.. I aint expert... so.. hard to say
16:37:10 <Borg> Eddi|zuHause: I doubt it go away.. because block signals penalty is added only when they are RED
16:37:24 <Borg> if they are GREEN.. no penalty.. and checking them to PBS.. means no penalty either
16:38:20 <Eddi|zuHause> right, turned some more signals, now it seems to work better
16:38:22 <FLHerne> Eddi|zuHause: Which ones? It doesn't work for me
16:38:36 <FLHerne> I tried the pair by the lake, then all of them on both routes
16:38:55 *** gelignite has joined #openttd
16:39:10 <Eddi|zuHause> it's weird, sometimes it works for a moment, and then it stops
16:40:05 <Borg> now its even more weirded
16:40:10 <Borg> I removed all PBS back signals..
16:40:27 <Borg> and it works as it SHOULD.. because upper path.. should have imo sligher cost.. that down path
16:40:47 <Borg> the only difference is.. im still at look_ahead_max_signals = 20
16:41:39 <Eddi|zuHause> now i also turned the ones right after the destination station
16:43:01 <Eddi|zuHause> what makes this more difficult to diagnose is that servicing is enabled
16:44:13 <Borg> but all trains have goto depoint for maintenance? and it kicks not often.. 540 days
16:44:25 <Borg> so it should not influence it at all?
16:45:45 <Borg> also u can turn on: order.no_servicing_if_no_breakdowns 1
16:45:46 <Eddi|zuHause> well, my conclusion: don't mix block and path signals
16:46:06 <Borg> Eddi|zuHause: huh?!! how come.. block signals are essecinal for balancing
16:46:21 <Borg> parameters: yapf.rail_look_ahead_signal_p*
16:46:36 <Borg> without block signals.. penalties never kick in..
16:46:57 <Borg> Eddi|zuHause: did you make my track works only by changing signals to PBS? all of them?
16:47:21 <Eddi|zuHause> anyway, they still take the long detour sometimes, when the routes are blocked. this is usually because the more signals they go through on the alternate path makes it drop the check whether paths are reserved, which makes it APPEAR shorter
16:47:37 <Borg> Eddi|zuHause: thats good!! even more.. excelent.. thats what AUX path is for
16:47:43 <Borg> take it.. when stuff is blocked..
16:47:51 <Borg> but if not... go primary :) thats the goal design here
16:48:06 <Eddi|zuHause> if you remove some signals, that situation should be less likely
16:48:24 <Borg> but.. now.. train do go AUX.. even when PRI is empty.. and free
16:48:56 <Eddi|zuHause> which leads to the other conclusion: don't give the trains too many options
16:49:23 <Borg> Eddi|zuHause: there are just 2 options there..
16:49:30 <Borg> is that too much already? :(
16:50:45 <Borg> okey..I removed some block signals
16:50:49 <Borg> still wrong path selected
16:50:59 <Borg> can you provide me your save game?
16:51:48 <Borg> to sum up... Im kinda happy that yapf.rail_look_ahead_max_signals have big influence on that
16:52:02 <Borg> so... I need to recalc p0,p1,p2 tables again.. and might be nice workaround
16:53:19 <Eddi|zuHause> there's no real reason for this "AUX" path to exist. it doesn't have more capacity for holding trains, and no overflow depot
16:53:28 <Borg> I removed last block signal on path.. and.. it started to work..
16:53:54 <Eddi|zuHause> in the worst case, it locks up, and blocks your other routes
16:54:12 <Borg> just stop 2 trains there
16:54:18 <Borg> you will see how AUX path gets used
16:54:26 <Borg> if its blocked.. get detour
16:54:58 <Borg> AUX path was first also.. I added botton path to unload the AUX because its used for other trains..
16:55:31 <Borg> I remember I had problems w/ block signals before.. that somehow paths w/ block signals were more preffered.. somehow
16:55:39 <Borg> but I never dug up this in source code..
16:57:20 <Borg> I rememoved block signals.. and for few trains it worked
16:57:26 <Borg> now.. they still select AUX path... wtf
16:57:49 <Borg> nows its feel kinda balanced
16:58:00 <Borg> when its a bit blocked... then AUX path is used
16:59:08 <Borg> adding ONE block signal.. into AUX path.. just after that crossing bridge
16:59:13 <Borg> completly change YAPF behavior..
16:59:26 <Borg> now all trains go via AUX path..
16:59:44 <Borg> and I cant find that in source code.. cost calculator..
17:00:53 <FLHerne> Borg: It's kind of bad form to spam IRC with 20 lines of stream-of-consciousness rambling
17:00:53 <Borg> if its RED.. and not PbsSignal.. just add cost
17:03:01 <Eddi|zuHause> that path has 10 signals, and after 10 signals, the "is this track occupied?" check is dropped
17:03:31 <Eddi|zuHause> so if both the platforms and the holding tracks before them are occupied, this longer route is being treated as "shorter"
17:03:55 *** Wormnest has joined #openttd
17:04:19 <Borg> Eddi|zuHause: w000t? . how come? how more signals make it shorter?
17:04:50 <FLHerne> Because it only cares about track occupancy in the first 10 signal sections
17:05:04 <Borg> Eddi|zuHause: also: please removed back signals.. it will make it <10 signals and it will be still prefered
17:05:04 <FLHerne> And track occupancy is very high-cost
17:05:33 <Borg> FLHerne: whats the variable name of it? I want to check sources.. to understand this...
17:09:33 <Borg> for now I only see one place with that in yapf_costrail.hpp, and it add yapf.rail_lastred_exit_penalty = 10000 to cost. But in my settings I have it set to 0, but still dont help. is there any other places?
17:11:14 <Borg> Eddi|zuHause: also, even if.. this is weird because cost is always added.. so more signals.. more possible costs (not less) so this is counter intuitive as well..
17:11:15 <Eddi|zuHause> maybe it's also seeing some loop which will increase signal count
17:11:28 <FLHerne> Borg: I think you want yapf.rail_pbs_cross_penalty and yapf.rail_pbs_station_penalty
17:11:46 *** _2TallTyler has quit IRC (Quit: Page closed)
17:11:59 <FLHerne> (cost per tile of conflicting track reservation, latter case for stations)
17:12:06 <Borg> FLHerne: that is pbs_cross_penalty ? I had to increase it in my setting.. so trains better balance at platform entrances.. but I dont understand code :(
17:12:47 <Borg> 300 -> 1000 make it nice... trains dont wait for platform when some reservation blocks them. they select other uncoopuied platform..
17:12:59 <Borg> with 300.. they were just waiting unnecessary.. blocking flows.
17:13:01 <Eddi|zuHause> the settings are "pbs_cross_penalty" for occupied track, and "pbs_station_penalty" for occupied station. "lastred_exit" doesn't apply because you don't have any exit signals, and PBS signals aren't red or green
17:14:02 <Borg> Eddi|zuHause: okey, I have pbs_station_penalty default... 800 what should I try?
17:14:14 <Eddi|zuHause> i've temporarily set them both to 100 (= normal track), and i've not seen a train using the AUX track
17:15:59 <Borg> Eddi|zuHause: ARGH!! thx :)
17:16:01 <Eddi|zuHause> (those are not good settings, just for testing)
17:16:17 <Borg> setting pbs_station_penalty 800 -> 100 made it alone!
17:16:20 *** jottyfan has joined #openttd
17:16:21 <Borg> all other are default settings
17:16:35 <Eddi|zuHause> i still think you should just remove the AUX track
17:18:27 <Eddi|zuHause> remember, the 800 (=8 tiles) is per station tile, so a 7 tile station will allow for 56 tiles worth of penalties as potential detour
17:19:16 <Borg> but 800 is default with 100...
17:20:01 <Borg> also.. I dont need to touch pbs_cross_penatly then.. because pbs_station_penalty was also breaking my platform balancing.. damn! one little setting.. so high PF influence..
17:20:28 <Eddi|zuHause> you can reduce that, but you should still have it higher than 100, so trains would choose a free track over an occupied track
17:20:30 <Borg> okey.. thanks a LOT guys.. now I will revert back to my own settings and adjust only pbs_station_penalty and I will check what will happen now..
17:21:23 <Eddi|zuHause> setting it to 100 will probably wreak havoc on other places of your network
17:21:24 <Borg> Eddi|zuHause: if it will be problematic.. it will quickly appear on that Nonningwell West station w/ 5 platforms.. when trains will just wait.. instead tacking unocuppied.. :)
17:27:59 <Borg> so far so good.. w/o PBS back signal at AUX.. its being selected.. with is fine... adding one PBS signal.. solves problem.. it because AUX.. selected only when other blocked.. now lets see if other places have issues now
17:30:07 <Borg> there are problem in other places.. DAMN.. :(
17:32:47 <Borg> time to do RTFS more.. what exacly that pbs_station_penalty do
17:33:54 <Borg> AHA! its near the code I pasted earlier about weird look_ahead_max_signals/2 code in ReservationCost()
17:34:48 <Borg> in one fuction.. all those settings are there.. as well pbs_cross_penalty
17:36:35 <Borg> but. luicky.. those 2 are seems to be used only in ReservationCost()
17:36:44 <Borg> so.. it should be easy to tune it up correctly! :)
17:45:15 *** frosch123 has joined #openttd
17:54:43 <Borg> okey.. now I slowly get picture of whats going on..
17:58:59 <frosch123> there is over 6 hours of yapf talk in the log. i guess i was lucky i had to work today?
17:59:28 <Borg> yeah... consider yourself lucky...
18:01:23 <Eddi|zuHause> frosch123: probably better :)
18:02:20 <Borg> is there any reason to cut in half look ahead signal in Reservation Cost?
18:03:11 <Borg> I think best solution would be to set it to larger number like 20? perhaps 30 even.. and carefully recalculate table so past away signals have minor influence on path.
18:03:43 <Borg> touching pbs_station_penalty makes havoc indeed.. like Eddi|zuHause predicted (if you use PBS to balance platforms)
18:04:44 <Borg> as low as I can go there is pbs_station_penalty = 500
18:05:03 <Borg> for for 3 tiles platforms.. it gets 1500. and its enough to balance w/o PBS back signal
18:18:13 <Samu> #8492 - I've been all day trying to make terraforming less restrictive
18:19:06 <Samu> while it diminishes the cases of flooded houses, it doesn't totally eliminate the issue as it currently stands in the PR
18:20:17 <Samu> the PR approach totally eliminates flooding
18:20:46 <Samu> but might be "overkill" in some cases, I'm unsure yet
18:23:20 *** Progman has joined #openttd
18:31:33 <Borg> Eddi|zuHause: I think this is a bug... in ReservationCost() that div 2
18:31:54 <Borg> I just made new p0,p1,p2 params adjusted for look_ahead_max_signals=20
18:32:29 <Borg> BUT of course.. I can be unlucky again.. and create path w/ exaclt 20 signals vs shoter.. and it will break again..
18:32:33 <andythenorth> ships ships ships ships ships
18:32:42 <andythenorth> names of ships are Hard
18:36:33 <Borg> so, any YAPF guru, please take a look at ReservationCost() function and comment why we check only on half of look ahead signals here... what was the reasoning..
18:39:46 <Borg> removing that div 2 fixes everything... and makes things work as it should
18:40:48 <michi_cc> Borg: It fixes everything for your very specific case, nothing more and nothing less.
18:41:12 <Borg> and actually.. im not sure if its wise to have it here at all.. why the check? if we reach Platform Reservation check.. why checking how many signals we passed anyway?
18:46:26 <Eddi|zuHause> Borg: the limit is there because it is a) pretty expensive to check, and b) pretty meaningless if you have a 1000k tile route to the other end of the map, because the situation will be different once you get there
18:50:40 <Eddi|zuHause> the "expensive" part is that it cannot be cached, because it changes all the time
18:52:26 <michi_cc> And yes, the limit to reservation cost is there to prevent to following complaint: "Why does the train go this other long way, the train in front would be long gone until if it just continued straight".
18:52:43 <michi_cc> Strike on until, please.
19:02:50 <Borg> Eddi|zuHause: okey right....
19:02:56 <Borg> michi_cc: yeah.. valid point too
19:04:08 <Borg> okey.. so there is more understanding to this now.. lets try other way around.. shorter look ahead...
19:07:07 <andythenorth> it amuses me when people think we have gurus and experts :)
19:07:18 <andythenorth> instead of people just desperately trying to make it work :)
19:09:17 <Eddi|zuHause> andythenorth: how much difference is there really? :)
19:09:28 <andythenorth> very good question
19:14:14 <Borg> hooly shit.. now it looks even better
19:14:33 <Borg> look ahead max signals = 5 didt splendit work
19:16:39 <Borg> and thanks to that div/2
19:17:36 <Borg> only final PBS signal check for platform availability
19:18:30 <andythenorth> 1926 cargo ship: design speed 11 knots; 2020 hybrid propulsion cargo ship: design speed 11 knots
19:21:44 <LordAro> did we have one of them before?
19:22:40 <LordAro> ooh, that's one of the unfinished langs
19:22:51 <LordAro> can eints even handle those?
19:23:40 <frosch123> though possibly not tested as much :)
19:25:55 <frosch123> the previous persian translator was last active 2016-06-12 :)
19:28:54 <supermop_Home> andythenorth catamaran ferries
19:29:16 <andythenorth> yes but not for logs
19:30:14 <FLHerne> The newer ones have vastly lower operating costs
19:30:24 <FLHerne> But no-one cares about running cost in OTTD
19:31:04 <FLHerne> I think with ships you kind of have to accept that the differences are for visual happiness more than gameplay :-/
19:31:13 <FLHerne> (or reinvent the economy)
19:31:21 <andythenorth> it's just very inconvenient of reality
19:31:31 <andythenorth> it should be arranged for gameplay
19:38:21 <andythenorth> hmm Train Whack is broken
19:38:30 <andythenorth> it doesn't understand dual-headed engines
19:39:45 <FLHerne> I think it's broken by existing, tbh :p
19:41:29 * Wolf01 sits down, "let's do something", runs netflix...
19:45:03 <andythenorth> the shipping industry calls them 'general cargo' but that doesn't mean general cargo
19:45:28 <andythenorth> it means specifically packaged cargo in palettes, bundles, crates or bags
19:45:44 <andythenorth> general cargo goes in a multi-purpose vessel
19:47:18 <andythenorth> and generally if it's mixed breakbulk it goes in a tweendecker for better use of space
19:47:46 <andythenorth> it's also called packet cargo, but packet cargo doesn't go in a packet ship or a ship in the packet trade, that's for mail and valuables
19:48:17 <andythenorth> the cargo class for this is piece goods, but nobody knows what that is?
19:48:28 <Wolf01> I think you are going too much specific
19:48:56 <andythenorth> you can drive trucks and forklifts onto this type of ship with the ramp, but it's not a ro-ro ship
19:49:06 <andythenorth> Wolf01 yes, I just need a name P
19:49:49 <andythenorth> and a working : key
19:50:21 <andythenorth> "Simba Delete This Sprite There is No Valid Name For This Ship Type"? :)
19:51:13 <FLHerne> andythenorth: It's sort of like a big OSV
19:51:58 <andythenorth> gameplay-wise, it's like a railway goods van
19:52:16 <FLHerne> Then what exactly is wrong with 'general cargo'?
19:52:28 <andythenorth> it only refits piece goods
19:52:45 <andythenorth> I could change that
19:53:06 <FLHerne> I don't think that's an issue
19:53:19 <andythenorth> it's all kind of lolz anyway, because the ship names can't be seen in game :) grf.farm/images/ship-buy-menu-width.png
19:53:37 <FLHerne> In principle you *could* ship bulk cargoes in bags on something like that, but no-one would
19:54:13 <andythenorth> maybe you're right
19:54:22 <andythenorth> I don't try and name Horse wagons for what they carry
19:54:27 <andythenorth> I just name them for having a name
19:57:14 <andythenorth> maybe Cargo Liner
19:58:34 <andythenorth> which is quite a cool ship imo
20:06:11 <supermop_Home> andy it sounds like a box truck for boats?
20:06:15 <DorpsGek> [OpenTTD/OpenTTD] perezdidac updated pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFsM
20:07:20 <supermop_Home> also i tried to give all my road vehicles brand names, model numbers, description etc
20:08:26 <supermop_Home> groundsprites grass and snow done btw
20:09:11 <DorpsGek> [OpenTTD/OpenTTD] perezdidac updated pull request #8495: Feature: Prevent tunnels from being built under extractive industries https://git.io/JLFsM
20:10:21 <supermop_Home> just call it a box boat or pallet ship
20:11:53 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUq
20:14:25 <FLHerne> Hm, "box boat" to me always means containers
20:15:06 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUC
20:15:58 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUB
20:16:10 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUE
20:16:21 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUg
20:18:02 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbU6
20:20:44 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbUQ
20:22:06 <supermop_Home> andythenorth ok apparently I PM'd myself the hotel construction sprites last night
20:26:31 <Samu> I was wondering why add a cost check for the terraform
20:27:07 <Samu> with costmods that check is gonna not be effective
20:28:06 <Samu> like when water is made cheaper to clear
20:30:44 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbTZ
20:30:48 *** jottyfan has joined #openttd
20:49:04 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbkj
20:51:29 <andythenorth> how about industries with no-fly zone? o_O
20:55:14 <frosch123> how can we establish "decide whether something is wanted first, before discussing details how it is done"?
20:56:12 <frosch123> it's a pretty as**ole move have random people request changes, and more work from the PR author, when people are still discussion the feature in general
20:57:21 <DorpsGek> [OpenTTD/OpenTTD] mattkimber commented on pull request #8485: Codechange: improve performance for complex vehicle chains by resolving sprites less often https://git.io/JLbIR
20:58:47 <LordAro> frosch123: there's the "candidate: *" tags
20:59:27 <LordAro> can't really stop "random people" (even if they're eddi) requesting changes on "unassessed" PRs
21:00:14 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
21:03:25 <andythenorth> is the feature 'tunnels' or 'industries can request limitations'?
21:03:46 <andythenorth> or I could just stay out of it also, quite happily
21:05:04 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
21:05:31 *** WormnestAndroid has joined #openttd
21:07:05 <frosch123> it's a weird feature. it tries to enforce a certain game-style
21:07:23 <frosch123> but imo players should enforce their own-playstyle by just playing like that
21:07:47 <frosch123> so, i tend to: no religion in ottd
21:08:02 <DorpsGek> [OpenTTD/OpenTTD] perezdidac commented on pull request #8495: Feature: Prevent tunnels from being built under mines https://git.io/JLbIS
21:08:30 <FLHerne> By that argument there should be no costs, no town ratings, no date restrictions, etc.
21:08:44 <andythenorth> FLHerne I swear I didn't draw that XKCD, but I wish I had
21:10:05 <andythenorth> for whatever it's worth, I wouldn't use a restrictions flag in FIRS
21:10:13 <andythenorth> it would just lead to requests for a parameter
21:11:22 <andythenorth> I think there are more interesting things to do than spend extending content API in that direction
21:11:34 <frosch123> FLHerne: what about disallowing to raise/lower corners further than 3 heightlevels from where they were during map generation?
21:11:36 <Borg> I now play for a hour or so.. and no anomalies spotted.... YAPF works fine :)
21:12:17 <FLHerne> frosch123: Oh yes please
21:12:36 <FLHerne> Also, indestructible rivers
21:14:27 <frosch123> ok, second attempt: how annoying is it, when a new industry is constructed over your tunnels, you want to adjust the track layout, remove the tunnel temporary, and notice you cannot put it back?
21:15:00 <frosch123> looks like the town-authority-analogy holds up
21:15:19 <FLHerne> frosch123: Well, they shouldn't be able to mine over your tunnels
21:15:31 <FLHerne> So the industry construction check should fail
21:15:33 <Eddi|zuHause> frosch123: i think the logical conclusion would be, that such industries cannot be founded above tunnels
21:17:22 <andythenorth> ok so it's flag, and it has to be handled in per-tile location checks?
21:17:40 <andythenorth> maybe just industry location checks
21:17:48 <frosch123> also, how often does this scenario even happen?
21:18:00 <frosch123> how often do you tunnel under an industry, and how often is it a mine?
21:18:33 <Eddi|zuHause> frosch123: i vaguely remember this feature existing in the past
21:18:43 <frosch123> people tend to ignore superfluous secondary industries, and tunnel under them if they are in the way
21:18:50 <frosch123> but primary are always intesrting to service
21:18:59 <Eddi|zuHause> but i haven't bothered checking out which patchpack that might have been
21:19:28 <FLHerne> My imaginary proper solution:
21:19:38 <FLHerne> - Industries have classes, like cargo-classes
21:19:50 <FLHerne> (including 'IC_MINE')
21:20:12 <Eddi|zuHause> <frosch123> but imo players should enforce their own-playstyle by just playing like that <-- i'm sceptical about this approach, because often it's easier to follow a rule if it's actually enforced
21:20:22 <FLHerne> - GameScripts or similar can filter all actions like I was proposing before
21:20:23 <andythenorth> FLHerne start a discussion! :)
21:20:51 <FLHerne> - Thus, someone who cares can write a GS that refuses to allow a tunnel if it goes under a mine
21:21:12 <frosch123> please start less discussions about features noone is going to implement :)
21:21:51 <andythenorth> TL;DR we already know this shouldn't be a setting because (1) policy: content > settings (2) newgrf industries won't respect the setting meaningfully
21:21:53 <frosch123> a discussion needs something to be discussed
21:22:23 <andythenorth> if someone turns up with a fully working newgrf patch, a test grf, docs, and an nml PR, then...fine?
21:22:40 <FLHerne> frosch123: Yeah, that's my feeling
21:22:59 <Eddi|zuHause> FLHerne: i can't imagine GS having the proper access to allow/deny constructions
21:23:41 <FLHerne> Eddi|zuHause: My idea was that you could register a Squirrel callback to run on each action
21:23:57 <FLHerne> Not just construction
21:24:00 <Eddi|zuHause> yes, but there's no such thing as squirrel callbacks
21:24:42 <Eddi|zuHause> <FLHerne> - Industries have classes, like cargo-classes <-- what would these classes do differently than the current flags?
21:25:20 <Eddi|zuHause> other than being completely arbitrary and difficult to standardize between different GRF/GS authors
21:25:43 <Eddi|zuHause> i think your suggestion is effectively unimplementable
21:25:58 <FLHerne> Eddi|zuHause: The current flags are even more arbitrary :p
21:26:02 <Eddi|zuHause> completely aside of the fact that nobody would do it
21:26:06 <FLHerne> It seems to mostly work for cargo-classes
21:26:16 <frosch123> Eddi|zuHause: there is a "save" callback
21:26:32 <FLHerne> You get a few odd-looking combinations where vehicle and industry authors didn't quite agree what some flag meant
21:26:57 <FLHerne> But it means that most vehicle grfs work pretty well even with FIRS' 50-whatever cargos
21:27:15 <FLHerne> Which weren't even conceived of when most of those grfs were written
21:28:07 <FLHerne> It may be a bit late for that, of course
21:33:34 *** Borg has quit IRC (Quit: leaving)
21:37:19 *** Progman has quit IRC (Remote host closed the connection)
21:39:11 <Samu> simplified, readable, that kind of stuff
21:41:40 *** frosch123 has quit IRC (Quit: be yourself, except: if you have the opportunity to be a unicorn, then be a unicorn)
21:49:17 *** Wormnest has joined #openttd
21:51:17 <andythenorth> found loads of them
21:51:27 <andythenorth> much easier to draw when you understand the function of each part
21:59:15 *** jottyfan has quit IRC (Quit: jottyfan)
22:04:29 <andythenorth> always a good watch :)
22:04:32 <andythenorth> nobody died, just
22:07:15 <supermop_Home> give objects -z height, like discussed for bridges over stations
22:08:12 <supermop_Home> vanilla mines can have a 'depth' of 0, but if andy wants to make mines with -10 depth he can
22:09:33 <andythenorth> I have zero interest in having industries limit routes
22:09:50 <andythenorth> the nice thing about content APIs is someone else could :P
22:10:02 <supermop_Home> or be crazy and have depth like the meat packing plant
22:18:07 *** Samu has quit IRC (Quit: Leaving)
22:21:18 <supermop_Home> andythenorth do the spritesheets for the ogfx hotel need to arranged same as the current one?
22:22:06 <andythenorth> I don't mind doing that, but I have my head in other things currently :)
22:23:04 <supermop_Home> k i can do it quickly
22:31:10 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
22:31:14 *** WormnestAndroid has joined #openttd
22:33:40 *** sla_ro|master has quit IRC ()
22:37:52 <supermop_Home> which hotel do you want for the base set version?
22:41:22 <supermop_Home> UGH actually my hotel is 5px taller than base set version
22:49:09 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:57:09 <andythenorth> hotel K BR TR for base set?
22:59:21 * andythenorth must to sleep oops
22:59:30 <andythenorth> so failing to do sensible bedtimes :P
22:59:54 *** andythenorth has quit IRC (Quit: andythenorth)
23:06:57 *** gelignite has quit IRC (Quit: Stay safe!)
23:12:06 *** jottyfan has joined #openttd
23:26:11 *** nielsm has quit IRC (Ping timeout: 480 seconds)
23:49:01 *** Wormnest__ has joined #openttd
23:52:49 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
continue to next day ⏵