IRC logs for #openttd on OFTC at 2017-11-11
⏴ go to previous day
00:10:56 <B_> milek7: Any hint on the directives name?
00:16:18 <B_> distribution_pax = 0 distribution_mail = 0 distribution_armoured = 0 distribution_default = 0
00:30:03 <Samu> (OrderDepotActionFlags)(o->GetDepotActionType() & ~ODATFB_NEAREST_DEPOT)
00:31:07 <Samu> OrderDepotActionFlags is a enum
00:47:14 <Samu> then i got this all wrong :(
00:47:56 <Samu> explain me with example :(
00:52:02 *** gelignite has joined #openttd
01:04:59 <Samu> i dunno what orderbackups are for, but I edited it anyway
01:19:32 *** orudge` has joined #openttd
01:19:33 *** ChanServ sets mode: +o orudge`
04:02:14 *** ToffeeYogurtPots has quit IRC
04:02:41 *** ToffeeYogurtPots has joined #openttd
07:56:39 *** ChanServ sets mode: +v tokai
07:58:52 *** andythenorth has joined #openttd
08:07:47 *** HerzogDeXtEr has joined #openttd
08:51:00 *** PressureLine has joined #openttd
08:51:08 <PressureLine> So what did I miss?
08:53:37 <PressureLine> andythenorth: Road Hod 1.2.1 | Drumbeck Tanker Tram | uses electric sparks even though it is a steam tram
09:00:57 *** gelignite has joined #openttd
09:09:30 *** Alberth has joined #openttd
09:09:30 *** ChanServ sets mode: +o Alberth
09:18:30 *** Breckett has joined #openttd
09:36:41 *** Progman has joined #openttd
09:44:36 *** Progman has joined #openttd
11:00:33 *** synchris has joined #openttd
11:40:33 *** Stimrol has joined #openttd
11:51:59 *** Biolunar has joined #openttd
11:55:53 *** JacobD88 has joined #openttd
12:04:33 <Samu> Alberth: remember that code for airplanes I thought it was useless? I think I understand now what it was supposed to be doing.
12:05:25 <Samu> it should be placed right after leaving a terminal
12:07:38 <Samu> before heading to next destination, check if it needs autoservicing
12:50:23 <Eddi|zuHause> looks a bit like those large model trains where people can sit on it
13:03:12 *** frosch123 has joined #openttd
13:31:53 *** Wormnest has joined #openttd
13:53:03 <Samu> this is harder to fix than i thought
13:54:28 <Samu> when the aircraft enters a terminal, it v->BeginLoading();
13:54:59 <Samu> BeginLoading is doing a lot of stuff... inclusing changing the next station to visit :(
13:55:08 <Samu> i wasn't expecting that :(
14:00:59 <Samu> if I do a check if aircraft needs auto servicing, it will be sending the aircraft to the next airport to visit, and not the airport it's currently at :(
14:31:39 *** Wolf01 is now known as Guest337
14:31:40 *** Wolf03 is now known as Wolf01
14:46:46 <Samu> CheckIfAircraftNeedsService function is becoming too big :(
15:25:49 *** Gustavo6046 has joined #openttd
15:26:08 *** orudge` has joined #openttd
15:26:09 *** ChanServ sets mode: +o orudge`
15:41:00 *** sla_ro|master has joined #openttd
16:13:32 *** Mahjong has joined #openttd
16:38:55 *** Gustavo6046 has joined #openttd
17:26:15 *** andythenorth has joined #openttd
17:30:34 <andythenorth> world of tanks is addictive
17:32:29 <V453000> made a smart rendering script for articulated vehicle consisting of 1 model
17:32:36 <V453000> of course there were errors :)
17:33:09 <V453000> and I was so confused that I even re-checked the blender camera setup because I though it's coming from that, even though the math in there is ultra trivial and everything is ultra consistent with the game
17:35:13 <V453000> /me is back to train set after making PURR and building 700 trains over a few days on our welcome server for "testing purr scientific purposes" :D
17:36:24 <V453000> andythenorth: do you know some way how to extract layers individually from PSD ?
17:40:01 <V453000> currently I was ordering sprites in After Effects but if I can export layers from photoshop automatically, shit's dope
17:40:38 <V453000> being able to remove Ae from the workflow would make it even easier to script ... I could make some exporting script for Ae in javascript but I don't want to touch js if I can avoid it for now
17:42:36 <andythenorth> V453000: never tried layer extraction
17:43:05 <V453000> will try Export a single layer: >>> layer_image = layer.as_PIL() >>> layer_image.save('layer.png')
18:42:26 *** sla_ro|master has joined #openttd
18:42:59 *** Alberth has joined #openttd
18:43:00 *** ChanServ sets mode: +o Alberth
19:01:57 <Snail> question for you… when does the recoloring callback get updated, for trains?
19:02:28 <Snail> it used to be the “callback 32 days” (CB_32DAY), but I believe things got quite messed up at a certain point...
19:03:41 <frosch123> the main misunderstanding is that it is only guaranteed when the cb is called
19:03:47 <frosch123> it is never said when it is not called
19:04:13 <frosch123> some newgrf authors assume that the coloring is stored persistently, but that is just plain wrong, and has always been wrong
19:05:20 <frosch123> the 32day flag means that the color is updated *at least* every 32 days
19:05:34 <frosch123> it does *not* mean that it is updated *at most* every 32 days
19:05:36 <Snail> well, to my understanding, the coloring is always cached, and only occasionally updated
19:05:57 <Snail> so which other cases is the recoloring also updated in?
19:06:17 <Snail> for instance, I experience this happening when reaching a station… or when loading some cargo...
19:06:26 <Snail> it’d help to have a list of these events
19:06:36 <frosch123> it may also happen when the track type changes
19:06:44 <frosch123> the list is endless
19:07:01 <frosch123> if you want to make stuff depend on date, use the date of last service
19:07:17 <Snail> yes, I’m already using it for other purposes
19:08:04 <frosch123> caching is to make stuff faster by not updating them all the time, it is no storage
19:08:28 <frosch123> if we find a way to make ottd multi-core we may not cache it at all and always update it
19:08:53 <Snail> I was wondering if it was possible to add a functionality, that would allow us to update recoloring *only* every 32 days
19:09:11 <Snail> well, I guess it’d require a new callback?
19:09:21 <frosch123> use "days since last service" / 32 or something
19:11:22 *** Gustavo6046 has joined #openttd
19:18:28 *** Func_Vehicle has joined #openttd
19:20:10 <Eddi|zuHause> wasn't there a 30 day callback? or was that 30 ticks?
19:20:20 *** ToffeeYogurtPots has quit IRC
19:20:48 *** ToffeeYogurtPots has joined #openttd
19:21:40 *** ToffeeYogurtPots has joined #openttd
19:24:24 <frosch123> Eddi|zuHause: it exists, but as long there is no persistent storage, it's pretty useless
19:24:45 <Eddi|zuHause> ah, yeah, i can see the issue
19:25:02 <frosch123> it's some kind of indirection, call this callback first before calling the others
19:25:18 <frosch123> but you could achieve the same by just calling the other callbacks
19:34:51 <Snail> frosch123: the aim of CB32 was to get vehicles recolored at random times (as 32 days would be)
19:35:14 <Snail> if recoloring is also updated in countless other situations, what I just said would not work
19:35:39 <frosch123> if it is random, then use random bits?
19:35:50 <Snail> I meant at random times
19:36:18 <Snail> random bits can only be reinitialized at certain well-specified events (loading, building, servicing…)
19:36:43 <frosch123> there is also the 32day trigger for rerandomisation
19:37:58 <frosch123> if it is about updated, iirc supermob changed livery at "current date" > "date for new vehicle generation" + "random delay"
19:38:50 <frosch123> there vehicles get new liveries in certain years, but for individual vehicles the repainting is delayed by a random span
19:39:16 <Snail> yes, I’m also doing something similar
19:39:28 <Snail> I’m changing the liveries in waves
19:39:41 <Snail> but this only happens when a vehicle enters the depot, of course
19:39:51 <Snail> I was talking about wagons changing colors outside of the depots
19:40:01 <Snail> like when the train is moving
19:40:36 <Snail> so far, changing colors almost always occurs when the train reaches a station, because that’s one event trat updates the recoloring cache
19:40:58 <Snail> I would ideally require the recoloring not to be updated when it reaches a station, but only every 32 days
19:41:44 <frosch123> well, do not think about the update moment, but rather about how to get the coloring for every moment
19:42:43 <frosch123> assume the update would happen every tick
19:42:52 <Snail> that would solve my problem
19:43:05 <Snail> but unfortunately it doesn’t happen every tick :)
19:43:22 <frosch123> what does it depend on?
19:43:45 <Snail> difference between last depot visit and current date
19:43:59 <Snail> it’s already different from wagon to wagon, based on random bits
19:44:08 <Snail> that get reinitialized at every depot visit
19:44:58 <Eddi|zuHause> you mean something like it's dirty when it's not serviced after <x> days?
19:45:27 <frosch123> hmm, i guess you can also drop the recoloring callback and use the vehicle stack recoloring capability
19:45:33 <frosch123> that one updates whenever the vehicle moves
19:45:44 <Snail> you mean at every tick?
19:45:54 <frosch123> at every movement step
19:46:14 <frosch123> well, also at every loading step
19:46:20 <Snail> but that would require a majotr rewrite in the code :p
19:46:36 <frosch123> i don't know your code :)
19:46:47 <frosch123> but i would expect only changing like 3 lines
19:47:34 <Eddi|zuHause> and per magic push-pull-whatever
19:47:40 <V453000> frosch123: of course my rendering script had a bug
19:47:43 <Eddi|zuHause> it quickly becomes complicated :p
19:47:44 <frosch123> essentially: instead of linking the decision for the color to the callback, link it to the regular drawing chain and store the result in some register
19:47:55 <frosch123> Eddi|zuHause: no, not per livery
19:48:10 <frosch123> exactly once per callback-id switch
19:48:21 <V453000> I have a 3-part 24/8 vehicle and each of the 8/8 has it's own controller, it's final frame scaling was leaking to the next sequence frame 0 :) now it resets itself properly
19:50:28 <andythenorth> free-to-play tank game
19:50:37 <andythenorth> most MP games are me, with low level tanks
19:50:48 <andythenorth> getting destroyed by people who've spent $$ to have OP tanks
19:51:06 <Eddi|zuHause> that's kind of a point of these games
19:51:13 <frosch123> andythenorth: i thought it's about recognising every tank and knowing the plating structure
19:51:19 <Eddi|zuHause> match the non-paying players with paying players, to get more of them to pay
19:51:56 <Eddi|zuHause> wasn't there recently an outcry about some company patenting such a matching algorithm?
19:53:57 <frosch123> yes, it was on jimquisition
19:54:16 <andythenorth> frosch123: I can't recognise the tanks :P
19:54:20 <andythenorth> I have NFI where to aim
19:55:07 <frosch123> andythenorth: isn't there some wiki?
19:55:34 <frosch123> also, as a brittish person, aren't you supposed to prefer battleships?
19:59:52 <andythenorth> I only have 13" laptop screen, I'd have to have the wiki open on a tablet
20:00:13 <andythenorth> judging by how many tanks don't move at game start, maybe lots of players are reading the wiki first :P
20:04:18 <frosch123> there is nothing weird with using your phone inside of a tank
20:04:26 <frosch123> try to get one with good wifi
20:06:29 <Samu> if (a->FindClosestDepot(NULL, &destination, NULL) && o->GetDestination() != destination)
20:06:42 <Samu> what is the value of the 2nd destination in that block?
20:06:54 <Samu> the one returned from FindClosestDepot?
20:13:44 <Samu> there's the vehicle vehicle, the real thing, and the vehicle engine, the model
20:14:57 <Samu> i'm not entirely satisfied about helicopters servicing at helipads
20:15:17 <Samu> why does this setting even exist :8
20:15:48 <Samu> automatic servicing, that is
20:24:20 <Eddi|zuHause> i imagine it's pretty boring inside of a tank most of the time
20:26:12 <frosch123> all military stuff is pretty boring most of the time
20:41:30 <Samu> i'm gonna change this helipad servicing a bit, instead of "on landing", i'm making it "on departing"
20:42:27 <Samu> avoids OT_LOADING crap always returning false
20:49:37 <Samu> why can't i autoreplace airplanes with helicopters? :(
20:49:49 <Samu> sometimes I don't mind that being possible
21:04:47 <V453000> frosch123: my 20k+ RVs friend got his stuff raised by someone to 50k, but that's not why I write ... he was doing a test and he used roadhog vehicles vs. some longvehicles or something, he's assuming that using articulated parts slows down more than using full non-articulated vehicles. Is this the case, and is there something similar happening for trains?
21:06:14 <frosch123> a single articulated part is obviously less expensive than a full vehicle
21:06:28 <frosch123> it does not do pathfinding or acceleration, and also skips half of the properties
21:06:55 <frosch123> but a 3-part articulated vehicle is obviously more expensive than a 1-part vehicle which does the same job
21:07:08 <V453000> is there any estimation how much?
21:07:11 <frosch123> so, the question is rather, what's the ratio between capacity and articulated parts
21:07:35 <frosch123> i have no idea. usually people test train performance :p
21:08:08 <V453000> :P I'm just wondering if making all vehicles split in 3 parts in a train set for big networks is that much of a good idea
21:08:17 <frosch123> and in train performance the acceleration was noticeable, which is only done by the front
21:08:38 <frosch123> essentially, the front part does stuff every tick
21:08:47 <frosch123> the other parts only do something when the front moves
21:08:56 <V453000> my vehicles move all the time :P
21:09:30 <frosch123> (random number, don't know the limit)
21:15:16 <frosch123> V453000: so, in summary, 1/8 articulated parts are worse than 8/8 :)
21:15:33 <frosch123> 8 times the graphics to resolve for a train of same length
21:16:41 <V453000> even if graphics are only on 1 unit and rest has some 8bpp placeholder only?
21:18:13 <frosch123> traditionally ottd checks for new graphics also when vehicles are currently not visible
21:18:21 <frosch123> noone changed that yet
21:19:50 <V453000> so it's mainly drawing issues then? Since the other tasks seem to be relatively low performance
21:20:01 <V453000> esp with 32bpp/EZ + 4 layers
21:22:36 *** PressureLine has joined #openttd
21:23:46 <PressureLine> time to see if I can get NML to work :D
21:24:14 <frosch123> 4x does not affect performance for many trains
21:26:38 <PressureLine> oh goody. this nml file is commented... in russian.
21:27:46 <Wolf01> Usually comments are for the one who wrote the code, not for others :P
21:28:52 <PressureLine> luckily nml is actually quite readable without the comments anyway
21:29:08 <PressureLine> reminds me a bit of lua
21:29:20 *** qwebirc37680 has joined #openttd
21:30:30 <PressureLine> so. for lazy nml method: "The easy route is to extract all the NML program files in the same directory as your NML project"
21:30:43 <V453000> well theoretically I guess I can make a paramter for using articulation? Although it's something that I guess would completely fuck with grf compatibility with itself
21:31:05 <V453000> probably best to just ignore any needless articulation for no w
21:31:14 <V453000> main reason would be the bending in curves for 24 rotations
21:31:16 <frosch123> V453000: i'll make a profile for psg325
21:31:47 <PressureLine> <V453000> main reason would be the bending in curves for 24 rotations
21:31:53 <PressureLine> sounds like the timecube!
21:32:16 <V453000> frosch123: interested :)
21:32:21 <V453000> waiting for rezults :P
21:32:26 <V453000> making offsets work meanwhile
21:32:29 <frosch123> so we can see how much cpu is eaten by vehicle fronts, vehicle parts and industry animation
21:32:45 *** PressureLine has joined #openttd
21:32:58 <V453000> industry animation 98% :>
21:33:02 <Samu> i hate service at helipad setting :(
21:34:11 *** qwebirc37680 has left #openttd
21:34:26 *** PressureLine has joined #openttd
21:34:36 <PressureLine> marginally embarassing :<
21:35:01 <PressureLine> if im being lazy with NML install
21:35:16 <PressureLine> and i have "F:\NML" folder
21:35:30 <PressureLine> and downloaded source of grf from devzone
21:35:58 *** sla_ro|master has joined #openttd
21:36:23 <PressureLine> do I just wallop the contents of the E:\Downloads\xussr-r6494-source.tar.xz\xussr-r6494-source.tar\xussr-r6494-source\ folder straight into the NML folder?
21:36:58 <PressureLine> or do put the folder inside the .tar into the nml folder
21:42:23 <Alberth> depends on how xussr finds nmlc
21:43:12 <PressureLine> I don't even know how to answer that :D
21:43:20 <Alberth> and how easy it is to customize it to your setup :p
21:44:05 <Alberth> ie there is no fixed place for a project relative to nml
21:44:25 <Alberth> nml is a generic compiler you typically use it in multiple projects
21:55:49 <frosch123> xussr uses make and cpp and stuff
21:56:08 <frosch123> so you will have a hard time to compile it on windows
21:59:06 <PressureLine> would it just be easier to write an override grf for the speed check then
22:01:22 <frosch123> hmm, 63% of cpu time in a game month is used for resolving trains sprites (doing the switches, not doing the drawing)
22:01:30 <frosch123> i did not remember it being that bad
22:01:53 <frosch123> damn, i wanted do to other stuff, not optimising this :/
22:03:24 <V453000> my new train set is supposed to have shitloads of graphical switches :D
22:04:49 <V453000> is that considering pathfinding frosch123 ?
22:05:23 <frosch123> pathfinder is like 3.5%
22:06:28 <V453000> I expected like 90% on that
22:07:46 <frosch123> well, it can be a good thing, it's a big opportunity for optimising ottd
22:09:19 <V453000> well that would explain why some people say NUTS is performance demanding for them
22:09:27 <V453000> NUTS uses a lot of graphics switches for wagons especially
22:10:11 <frosch123> in one game months it resolved 53million vehicles sprites
22:10:24 <V453000> is the number of switches or big amount of switch outcomes worse?
22:10:25 <frosch123> which involved 94million switches
22:10:35 <V453000> ok that's not so bad I guess
22:10:43 <PressureLine> damnit. where did they hide the variable running costs code
22:13:17 <V453000> so theoretically is 1 giant switch for all cargo labels worse or better than many less outputs in 4 individual switches in 4 vehicle layers?
22:16:36 <frosch123> currently 10% is finding the right case within a switch
22:16:53 <frosch123> but that is also something that can be optimised in theory
22:25:22 <V453000> is it like a sequence going one by one and as soon as it finds a right value it returns it?
22:25:30 <V453000> so I could use more likely things at the front?
22:26:09 <V453000> also does the store_temp somehow do bad things like extra performance if I do it for every train multiple times?
22:29:16 <frosch123> putting the more likely ones first is better
22:29:57 <frosch123> store_temp as expensive as say "+"
22:30:02 <V453000> at least something :P
22:30:11 <frosch123> it's nothing special compared to everything else
22:36:28 <frosch123> 3% faster by moving a single line
22:51:12 <V453000> offsets are pain. Looking forward to have this done :D
22:59:48 *** chomwitt has joined #openttd
23:24:14 <Samu> _settings_game.order.serviceathelipad... why do u exist :(
23:24:36 <PressureLine> for helicopters that *never* land at an airport
23:25:59 <Samu> it can get to the point it's serviced too often, and never triggers autoreplace :(
23:27:53 <Samu> i have another issue with this... airports without helipads, if the heli is going between these, my fix will fail
23:28:15 <Samu> i can't assumer the heli goes to every place with helipags
23:28:17 <PressureLine> what does your fix fix?
23:28:55 <Samu> my idea was to ignore the service interval cycle checking, and only check if it needs replacing
23:29:20 <PressureLine> but what does that *fix* ?
23:29:51 <PressureLine> what about it is broken?
23:29:57 <Samu> if it services too often, it would never autoreplace
23:30:13 <Samu> i fix one thing, but break another :8
23:30:35 <PressureLine> isnt autorepace tied to the *age* of the vehicle, not the reliability
23:31:26 <Samu> autoreplace is checked inside the v->needsautomaticservice
23:31:52 <Samu> bool Vehicle::NeedsServicing() const
23:32:34 <PressureLine> "Autorenew when vehicle is [time] after max age." [Y/N]
23:33:18 <Samu> that thing is still part of "NeedsServicing"
23:33:19 <PressureLine> "Autorenew vehicle when it gets old. [Y/N]."
23:33:44 <PressureLine> doesnt it still work when you send it manually to a depot?
23:34:16 <PressureLine> which you can do regardless of whether it needs serviceing or not
23:34:23 <Samu> automatically, it would depend on the servicing interval
23:35:05 <Samu> if the heli travels too fast between helipads, it resets the interval
23:35:24 <Samu> and never gets the chance to see if it needs autoreplacing
23:35:47 <PressureLine> have you actually checked to see if it *does* break autoreplace?
23:37:34 <PressureLine> simpler 'solution'
23:37:48 <PressureLine> change the helptext for the service @ helipad optiion
23:38:01 <PressureLine> ++(breaks autoreplace)
23:39:30 <Samu> ok, that is an idea, in the case I can't come up with a better solution
23:39:38 <Samu> a warning is better than nothing
23:39:44 <PressureLine> easiest to 'code' too
23:40:14 <PressureLine> are there any reports from users of that being an issue?
23:42:11 <PressureLine> if a heli is old
23:42:24 <PressureLine> ie past autorenew threshold
23:42:41 <PressureLine> with helipad servicing
23:43:09 <PressureLine> and it gets either a "goto depot" by clicking the button
23:43:24 <PressureLine> or via a go to depot order in the order queue
23:43:45 <PressureLine> does autorenew work?
23:44:23 <Samu> yes, if you send it to a depot manually, yes
23:44:52 <PressureLine> what about in the order queue?
23:45:16 <Samu> if the order to depot is from it's orders, yes, it will renew
23:45:24 <Samu> the problem is only for automatic servicing
23:45:55 <PressureLine> i think it may be because the 'service' doesnt happen in a depot
23:46:06 <PressureLine> not a proper one anyway
23:46:44 <PressureLine> f-ex you can't buy a new helicopter at the helipad 'station'
23:48:13 <PressureLine> so if you can't buy a new vehicle there, there's no mechanism for the aurorenew to work
23:48:47 <Samu> i somewhat tried to solve that problem of helis on 2 heliports. It will search for a hangar from another source, but this may break aircraft range on the way back
23:49:28 <PressureLine> aircraft range is a broken mechanic anyway imo
23:49:31 <Samu> i tried to limit the range which it can search
23:49:54 <Samu> meh, lots of things to test yet :(
23:53:11 *** Progman has joined #openttd
23:59:15 <Samu> gonna try fracture bool Vehicle::NeedsServicing() const into two functions
23:59:43 <Samu> just because of the ugly serviceathelipad
continue to next day ⏵
