IRC logs for #openttd on OFTC at 2019-02-03
⏴ go to previous day
00:04:37 <Samu> must be a client, to test bankruptcy on a server
00:04:54 <Samu> disabling breakpoints :(
00:10:32 <Samu> it's only single player where the bad stuff happens
00:10:48 <DorpsGek_II> [OpenTTD/OpenTTD] nikolas updated pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/fhamZ
00:12:29 *** Thedarkb-T60 has joined #openttd
00:26:44 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhSk4
00:27:00 <Samu> can you take a look peter1138!?
00:28:54 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhS8s
00:38:23 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7109: [OSX] Use high-precision scrolling properties for 2D scrolling https://git.io/fhS8C
00:42:16 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7163: Add: [AzurePipelines] split the CI in two parts: building and commit checking https://git.io/fhS84
01:01:05 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain closed pull request #7163: Add: [AzurePipelines] split the CI in two parts: building and commit checking https://git.io/fhSns
01:01:06 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #7163: Add: [AzurePipelines] split the CI in two parts: building and commit checking https://git.io/fhS82
01:04:08 *** andythenorth has left #openttd
01:05:11 <Samu> /* If Load Scenario / New (Scenario) Game is used, a company does not exist yet.
01:05:11 <Samu> * Create one here whenever start_spectator is disabled.
01:05:11 <Samu> * 1 exception: dedicated network servers. Those can have 0 companies. */
01:05:31 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 commented on pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhS8w
01:21:18 *** Thedarkb-T60 has joined #openttd
01:24:24 <Samu> hmm load game behaviour is buggy
01:24:38 <Samu> if COMPANY_FIRST is an AI
01:24:45 <Samu> I'm put into that company
01:24:56 <Samu> regardless if there's any HU company
01:27:07 <glx> that's the intended behaviour for single player game
01:27:32 <glx> with the setting you give an option to choose your company after loading
01:28:10 <Samu> it's the current 1.8.0 behaviour
01:28:46 <glx> in single player you are company 0 and that's all
01:29:21 <glx> the game can't know what was your previous company in an MP save
01:30:06 <Samu> I could avoid putting me in an AI company
01:30:29 <glx> and I think it also create a new company if the MP save didn't have one
01:30:30 <Samu> would put me in the first hu company it can find, if any exists
01:30:42 <Samu> or if it doesn't, then put me on company 0 then
01:31:16 <Samu> in mp, i could be in an AI company
01:32:13 <Samu> i remember reporting this, let me find
01:45:12 *** andythenorth has joined #openttd
01:45:25 <andythenorth> cargo_age_period (train prop 2B) is pretty ineffective eh? :)
01:45:38 * andythenorth has stayed up too late trying to get results out of it
01:46:19 <andythenorth> increasing it from 185 to max (word) gets about a 5% payment bonus
01:46:21 <peter1138> I stayed up late trying to get results out of RGB CC.
01:47:01 <peter1138> No RGB liveries yet, and, er, there's not enough space in p1 & p2 :/
01:47:16 * andythenorth just wanted to get double payments on some vehicles
01:47:20 <andythenorth> seems that's not a thing
01:47:50 <andythenorth> Eddi|zuHause can explain why I'm wrong in the morning :P
01:48:01 *** andythenorth has left #openttd
02:07:06 *** Thedarkb-T60 has joined #openttd
02:17:41 <Samu> /* Update the local company for a loaded game. It is either always
02:17:42 <Samu> * company #1 (eg 0) or in the case of a dedicated server or starting as spectator a spectator */
02:18:50 <Chrill> well, being a spectator, I see nothing wrong.
02:19:41 <Chrill> hence the way I formulated my comment :P
02:24:17 <dwfreed> that's perfectly fine, but it's less confusing if you swap "a dedicated server" and "starting as spectator"
02:25:57 <Samu> * company #1 (eg 0) or in the case of starting as spectator or a dedicated server a spectator */
02:30:54 <dwfreed> you could put in some commas that can help a bit too; one would go after or and the other would go after server
02:31:18 <dwfreed> eg: company #1 (eg 0) or, in the case of starting as spectator or a dedicated server, a spectator
02:34:41 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhSk4
02:35:32 <Samu> cheat_gui.cpp - what to do
02:36:20 <Samu> "Playing as company: 256"
02:38:21 <Samu> ah crap, forgot the toolbar gui
02:45:26 <Samu> it doesn't show the spectate button anymore, cannot spectate when in a company already
02:49:13 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhSk4
02:53:30 <Samu> the button must list New Company as the only item when the last company bankrupted, i think it does that already
02:53:42 <Samu> else i get an assert about list being empty
02:55:43 *** Wormnest has joined #openttd
04:12:42 *** Thedarkb1-T60 has joined #openttd
04:24:37 <Eddi|zuHause> * andythenorth just wanted to get double payments on some vehicles <-- well, it's not a payment factor...
05:28:11 *** HerzogDeXtEr1 has joined #openttd
05:59:19 *** tokai|noir has joined #openttd
05:59:19 *** ChanServ sets mode: +v tokai|noir
08:32:08 *** andythenorth has joined #openttd
08:32:27 <andythenorth> so....cargo aging period
08:43:36 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7049: Fix #6599: Can still click on buy button in vehicle selection window even if no vehicle is selected https://git.io/fhS0u
08:45:40 <andythenorth> what's it do eh?
08:47:14 <peter1138> Should it be a cargo property not a vehicle property?
08:47:40 <andythenorth> already got some of those
08:47:40 <peter1138> Well, it's our own custom thing, so only us to blame :p
08:48:18 <andythenorth> I'll test it a bit more
08:48:24 <andythenorth> it does affect payment, but eh
08:48:58 <peter1138> Well it is an old one, hmm.
08:49:02 <andythenorth> did I mention how good this ffwd is?
08:49:07 <andythenorth> for testing stuff like this :P
08:49:29 <peter1138> It's kinda too fast.
08:58:26 <peter1138> Should I make an HSV selector?
08:58:51 <peter1138> Scrollbars for R G B look dumb :/
09:06:10 <peter1138> Hmm, actually -5 outside.
09:08:59 <peter1138> Fortunately I accidentally left the central heating on overnight, so it's comfortable inside.
09:11:20 <andythenorth> nml wiki says cargo_age_period can be 0 ... 65535
09:11:51 <andythenorth> but it won't accept more than 65534
09:12:09 <peter1138> So you can have special engines that make cargo pay differently. Hmm.
09:12:33 <andythenorth> well it's the wagons I assume
09:12:51 <peter1138> Um, yeah, sorry, "Engine" in source-code speak.
09:14:03 *** sla_ro|master has joined #openttd
09:15:13 <andythenorth> it does work, I'm testing different values
09:15:20 <andythenorth> I think it's running into a clamp somewhere
09:15:46 <andythenorth> there's nothing intuitive about the resulting profits :)
09:16:14 <peter1138> It'll change the days_in_transit type stuff?
09:16:57 <andythenorth> yes, or something related there
09:18:50 <peter1138> Seems a very odd thing to have :-)
09:19:31 <andythenorth> oops, there's a typo
09:19:37 <peter1138> (svn r22713) -Feature: [NewGRF] Per vehicle custom cargo ageing period.
09:20:52 <andythenorth> I haven't plotted it, but the increase from 185 > 373 > 65534 looks linear to me :P
09:21:14 <andythenorth> I'd expect it to either be orders of magnitude different, or just clamped
09:22:08 <peter1138> There is a limit of 255 for cargo days in transit.
09:22:14 <peter1138> You may be hitting that.
09:22:41 <Eddi|zuHause> <andythenorth> so....cargo aging period <-- it's not a price factor, so you can't make prices above the original price of instant delivery.
09:22:46 <peter1138> Although you'd expect that to hit the lowest values, not the highest.
09:22:59 <peter1138> Yes, it's not a price factor.
09:23:05 <Eddi|zuHause> it only has real effect over large distances
09:23:31 <peter1138> I'm not sure what it's for, though.
09:23:42 <peter1138> Simulating refridgerated wagons?
09:23:50 <Eddi|zuHause> peter1138: original intention was for things like refrigerated cars
09:24:02 <andythenorth> I'm testing over 72 tiles crow distance
09:24:04 <Eddi|zuHause> peter1138: or trains with dining cars
09:24:11 <andythenorth> all of those things Eddi|zuHause
09:24:16 <Eddi|zuHause> andythenorth: that is not long :p
09:24:22 <peter1138> Passengers age less on a dining car?
09:24:46 <Eddi|zuHause> peter1138: more like "if there's a dining car in a train, they make longer journeys"
09:24:52 <andythenorth> also my map is only 256 ^ 2 :P
09:25:18 <peter1138> So you can have a distance of ~ 500 easily.
09:25:45 <andythenorth> let's see what 32 does
09:26:22 <andythenorth> a straight payment multilplier would be more straightforward here :P
09:26:54 <andythenorth> whatever the cargo curve is, you get n * that
09:27:20 <Eddi|zuHause> andythenorth: why would then anyone ever use the less expensive train?
09:28:22 <Eddi|zuHause> andythenorth: i think the real use of cargo aging is shortening the period on tightly packed metro trains or busses where people stand
09:28:38 <andythenorth> my tests are suggesting the same
09:28:44 <andythenorth> it's a malus not a bonus
09:29:37 <Eddi|zuHause> well, the bonus can make a difference if you're like driving 1000km without maglev
09:29:57 <peter1138> Hmm, that reminds me, there's a issue about overcrowding at stations.
09:30:20 <Eddi|zuHause> peter1138: the issue is usually too low capacity (or too high generation)
09:30:52 <Eddi|zuHause> and the transport rating is insufficient for regulating that
09:31:54 <Eddi|zuHause> i mean, we could hide the problem with a flat limit of like 100 passengers/pieces of cargo per station tile
09:32:20 <Eddi|zuHause> but it won't solve the underlying balance issues
09:33:12 <andythenorth> trying to find the inflection points
09:33:47 <Eddi|zuHause> i think your math is off?
09:34:00 <Eddi|zuHause> how is 373=2*185?
09:34:22 <peter1138> It's too high generation
09:34:32 <andythenorth> Eddi|zuHause: typo :)
09:35:57 <peter1138> Hmm, but since we closed a load of issues I can't find it :p
09:35:57 <andythenorth> not sure what I do with this next :)
09:36:14 <andythenorth> generation is in nielsm's PR
09:36:20 <Eddi|zuHause> andythenorth: now run the same test over twice, quad, etc. distance
09:36:35 <andythenorth> yeah I need more time for that :P
09:36:42 <andythenorth> or I need to make a standalone testing grf
09:36:49 <andythenorth> horse compile times are far too long for this
09:37:06 <peter1138> Oh yes, it's not closed :-)
09:37:25 <andythenorth> 30 mins to just get these results :P
09:37:45 <Eddi|zuHause> andythenorth: just duplicate the train with multiple values?
09:37:56 <andythenorth> yes that's what I'd do
09:38:05 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSEn
09:38:44 * andythenorth wonders if a redesign of the grf is looming
09:39:25 <andythenorth> I just assumed that 2 * cargo_age_period would be roughly twice the payment :D
09:39:35 <andythenorth> give or take the cargo curve
09:40:07 <Eddi|zuHause> not how it works :p
09:40:55 <andythenorth> I now assume it just changes the slope of the curve :P
09:40:59 <andythenorth> retrospectively more obvious
09:41:46 <nielsm> ...I should get a shower, put on some clothes, and find some breakfast, in one order or another
09:41:56 <andythenorth> these are all things
09:42:03 <andythenorth> some of those I need to do also
09:42:21 <andythenorth> then a day of childrens football, parties, shopping :P
09:42:25 <nielsm> and look at that PR again after I'm done with all of those
09:42:38 <andythenorth> 16 cargo nml? o_O
09:43:01 <andythenorth> but some more of the commits could use a squash
09:43:31 <andythenorth> I'm finding it hard to figure what's done or not, so I'm hoping we can squash down the working stuff to just a few commits
09:44:09 <andythenorth> Eddi|zuHause: you don't fancy making a test grf, and running it over different distances for a day? :P
09:44:15 <andythenorth> in the name of empiricism?
09:44:27 <nielsm> the prod cb v2 does work
09:44:29 <Eddi|zuHause> i don't think so
09:44:33 <nielsm> so maybe it's actually done?
09:44:42 <nielsm> except for maybe needing the new house props as well, not sure
09:45:32 <andythenorth> I thought there was one more thing we were stuck on
09:45:38 <andythenorth> or maybe it was docs
09:45:51 <andythenorth> logs would know, but eh, breakfast
09:49:49 <peter1138> Oh wait, there is room. p1 isn't full.
09:54:05 *** Progman has joined #openttd
09:56:45 <andythenorth> so cargo age period will have a big impact for slow vehicles
09:57:12 <andythenorth> but if I'm comparing Fast Train 1 to Very Fast Train 2, it's going to be marginal eh :P
09:57:55 <andythenorth> 30 day transit time, the bonus barely touches the sides
10:00:49 <andythenorth> first github result :P
10:01:18 <andythenorth> michi_cc can you remember the back story for train var 2B? :)
10:05:10 <Eddi|zuHause> andythenorth: try comparing transit times of about 1 year?
10:06:14 <Eddi|zuHause> andythenorth: anyway, i think the feature is working as intended :p
10:06:35 <Eddi|zuHause> andythenorth: play bigger maps :p
10:07:32 <andythenorth> I think it does work as intended
10:08:45 <andythenorth> I just wondered what the intent was :P
10:08:59 <andythenorth> my assumption was that it came from TTDP, but it's too new for that
10:13:47 <andythenorth> eh, it's exposed to CB36 :P
10:14:04 <peter1138> Yup, something else to desync :)
10:14:19 <andythenorth> 'vehicle is old, passengers complain'
10:14:27 <peter1138> So we can't fix cargo generation because goal servers / gamescripts rely on it being broken?
10:14:40 <andythenorth> we can't fix it allegedly without a setting
10:14:53 <andythenorth> but the quadratic thing is just a bug no?
10:15:02 <peter1138> Yes, clear and simple, I think.
10:15:09 <andythenorth> 2.0.0 semver major bump, API change
10:15:59 <peter1138> I want it fixed before then, it's not enjoyable having a small town and 1000s of passengers waiting
10:17:14 <peter1138> Either way it fills up massively, it is easier to get rid off without cdist though.
10:17:22 <nielsm> town cargogen, do the game setting I'm implementing in my PR, default old savegames to old algo, new games to new algo
10:17:24 <andythenorth> it's not enjoyable
10:17:29 * andythenorth wonders what vehicle vars could be abused for cargo age cb :P
10:17:46 <andythenorth> probably none, it won't trigger usefully
10:18:06 <peter1138> nielsm, I don't think it should be a setting.
10:18:16 <peter1138> Old savegames don't need it.
10:18:26 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSuI
10:18:29 <peter1138> It's the custom goal servers that may need it, and we have never tried to support them.
10:18:33 <peter1138> They have *custom code*
10:18:38 <peter1138> They can revert the change if they want.
10:19:09 <nielsm> so treat it as a 25 year old bug?
10:19:24 <nielsm> (which has turned into a feature in the meantime)
10:20:11 <peter1138> I guess you already fired up TTD to compare the behaviour ?
10:20:32 <andythenorth> goal servers can ship a fork :P
10:21:01 <andythenorth> or fix it in content
10:21:04 <peter1138> Heh, true, if its server side only, it is easier for them to just change their requirements.
10:21:24 <andythenorth> there's probably a newgrf houses way to get equivalently broken behaviour
10:22:10 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7164: Fix #7108: Group livery command did not check its parameters properly. https://git.io/fhSuq
10:22:18 <nielsm> yes just reimplement the cargo prod cb for houses
10:22:49 <andythenorth> just fake building populations to get the same statistical result
10:23:18 * andythenorth wonders what to do about cargo aging then
10:23:20 <Gabda> will 2.0.0 have backwards compatibility as well, or all the old save game conversion will be deleted, and it will start nice and clean?
10:23:42 <andythenorth> Horse design is based on 'oops, wrong assumption' :P
10:24:03 <nielsm> andythenorth nah, house pops are set in bytes
10:24:19 <nielsm> and default houses already scrape the limit, I think the largest is 240 in the baseset
10:24:46 <andythenorth> hmm, all these extra pax trains and wagons where the only difference is age period :P
10:31:05 <peter1138> So my 555 town in TTD
10:31:14 <peter1138> passengers last month: 23, max: 69
10:31:19 <peter1138> my 555 town in OpenTTD
10:31:25 <peter1138> passengers last month: 2, max: 224
10:31:52 <peter1138> Well, the TTD town is shrinking :p
10:32:01 <peter1138> But that's cos its expanding
10:32:38 <nielsm> this is very inconvenient :P
10:32:45 <nielsm> missing a lot of keyboard shortcuts, and no ffwd
10:34:18 <nielsm> at least it does have 1-2-3-4 for selecting rail build direction
10:34:22 <peter1138> I'm playing it on a dubious website as well, nothing to download.
10:34:39 <nielsm> (I learned that back in 1999 or whenever)
10:37:30 <andythenorth> yair so Horse, for a ~72 tile route, use the 100mph EMU, not the 140mph High Speed Train
10:38:58 <andythenorth> eh it's 87mph EMU even
10:39:49 <andythenorth> ~£364k profit / year vs ~ £346k
10:40:07 <peter1138> EMU should be profitable for shorter routes
10:40:09 <andythenorth> 'but what are the costs andythenorth'
10:40:11 <peter1138> HST for longer routes
10:40:33 <andythenorth> I can nerf the EMU costs, and give the HST more of a bonus
10:40:42 <andythenorth> but scaling costs against route length :P
10:40:54 <andythenorth> there's no factor relating the two
10:41:06 <andythenorth> cost is per year, not per tile travelled or whatever
10:41:34 <peter1138> A faster train travels more tiles in a year, so...
10:41:53 <andythenorth> I bet this would never be noticed in a game
10:42:05 <andythenorth> I only noticed because I'm setting costs
10:42:23 <peter1138> nielsm, well my small test is inconclusive.
10:42:28 <peter1138> Just 2 bus stops :p
10:42:37 <andythenorth> I could give the EMU an aging malus
10:42:42 <peter1138> ottd town is growing, slowly, ttd town is shrinking, slowly
10:42:46 <peter1138> But that could be anything.
10:42:52 <andythenorth> but the tests show that the malus has much more significant effect than the bonus :P
10:42:58 <andythenorth> so I could break EVERYTHING
10:43:34 <nielsm> ahhhh, this version of dosbox has a turbo feature
10:43:43 <nielsm> that makes the passage of time faster :D
10:43:56 <andythenorth> Horse needs a Pacer....cargo_age_period 1
10:44:31 <andythenorth> eh it's on a CB, I can set cargo aging to 1 for metro trains in July and August only
10:44:41 <andythenorth> do we know what hemisphere the map is in?
10:45:07 <nielsm> temperate is presumably england-like
10:45:44 <nielsm> tropical is brazil-ish?
10:45:57 <andythenorth> scale the cargo_age_period according to heat and cold?
10:46:06 <nielsm> subarctic likely northern sweden/finland
10:46:07 <andythenorth> different engines can supply different levels of heat or AC?
10:48:55 <peter1138> If you can't figure out what to do with it, just ignore that property :)
10:50:36 <peter1138> Can we build with that?
10:51:19 <andythenorth> if I give EMUs a malus, to 32 days, I get a 'better' result
10:51:27 <andythenorth> might need to scale it by vehicle speed
10:51:44 <andythenorth> in black and white days, people loved sitting on slow trains all day?
10:52:35 <andythenorth> why isn't there black and white mode before 1950?
10:52:40 <andythenorth> for the whole map?
10:52:45 <andythenorth> also sepia before 1910
10:53:23 <andythenorth> ^ peter1138 RGB easter egg? :P
10:57:51 <Eddi|zuHause> andythenorth: because the world wasn't in black and white back then?
10:57:59 <Eddi|zuHause> andythenorth: there's black and white newspaper
11:02:26 <andythenorth> eh so I need something like 'range' for pax trains
11:02:38 <andythenorth> and if the range is exceeded, there's a cargo payment malus
11:02:57 <andythenorth> probably about 32 tiles for 'normal' pax
11:03:17 <andythenorth> so how many tiles / day for a given speed?
11:03:38 <andythenorth> ignoring acceleration, deceleration, signal stops
11:03:55 <andythenorth> formular must be in source somewhere :P
11:04:10 <peter1138> So if everybody else is happy with an option, keep it an option I suppose.
11:05:59 <andythenorth> 'everybody else'
11:06:12 <peter1138> Maybe not everybody :p
11:08:56 * andythenorth wonders how cargo aging worth with cdist
11:09:03 <andythenorth> if pax stay on the vehicle are they aged?
11:09:07 <andythenorth> seems a bit detailed
11:09:27 <peter1138> As it's a vehicle property, yes.
11:09:57 <peter1138> cargo age refers to the time it took, not how old mr and mrs tables are.
11:13:39 <andythenorth> in cdist case, how does it count time? Just last leg? Or since loading point?
11:14:52 <peter1138> It's how long was it travelling for.
11:14:57 <andythenorth> I will make a better question later :P
11:15:04 <peter1138> Are you Samuing me?
11:15:46 <andythenorth> what do I need in Tesco?
11:15:53 * andythenorth will sort that for self
11:21:24 <Gabda> nielsm: what tool did you use to make the gif for PR #7120?
11:21:59 <Gabda> I would like to have a new gif (of the new function) for that PR
11:22:27 <Gabda> and I thought it would be better to ask for the method than to ask you to make a new one
11:31:40 *** Progman_ has joined #openttd
11:31:53 *** Progman_ is now known as Progman
11:39:05 <nielsm> hmm, I don't think I ever noticed before how much of an improvement UU '37' really is over SH '8P'
11:40:12 <nielsm> 20% cheaper, 10% more power, almost 40% lighter, and around 15% lower running cost
11:41:55 <Gabda> nielsm: thanks, it worked :)
11:46:02 <DorpsGek_II> [OpenTTD/OpenTTD] Gabda87 commented on pull request #7025: Add #6887: Option to show zone inside local authority boundary of towns https://git.io/fhSu5
11:46:22 <Wolf01> <Samu> mirc expired :( <- delete the registry key
11:49:24 *** gelignite has joined #openttd
11:50:48 *** andythenorth_ has joined #openttd
11:51:31 <andythenorth_> eh can’t check src on my phone, but....
11:52:15 <andythenorth_> what’s approxinate ratio of mph to tiles covered per tick?
11:53:45 <Gabda> maybe I shouldn't have put it into brackets :)
11:54:22 <Samu> oh, nice, is that for calc closest town?
11:54:44 <nielsm> peter1138: if only the new/fixed passenger gen algorithm is present, it'd also make trouble for all existing saves, since the networks in those would be dimensioned for the old algo, and would probably begin running at a huge loss
11:54:49 <Samu> AIs could benefit from it
11:55:01 <Samu> CalcClosestTown is one of the heaviest
11:55:05 <nielsm> and I don't think players would appreciate their saves being broken in that way
11:55:21 <Samu> and AIs use it intensively
11:55:26 <Gabda> with this, the calc clsoest town calling will be almost free, even on 4096x4096 maps
11:56:31 <DorpsGek_II> [OpenTTD/OpenTTD] alexanderweiss commented on pull request #7109: [OSX] Use high-precision scrolling properties for 2D scrolling https://git.io/fhSup
11:56:35 <peter1138> This is caching town ID for each tile?
11:56:54 <peter1138> When does it update?
11:57:06 <Gabda> on load and town add/remove
11:57:09 <Samu> what happens when a new town is created?
11:57:21 <peter1138> What happens when a town expands or shrinks?
11:57:42 <Gabda> nothing, closest town doesn't care about that
11:57:49 <peter1138> Or does that not affect closest town.
11:57:54 <Gabda> i mean the calc closes town
11:57:55 <nielsm> that doesn't affect which tile is closest to town sign
11:58:00 <peter1138> Okay, so you still need the town ID stored in the map for house tiles, etc.
11:58:19 <Gabda> but that is already in _m
11:58:25 <peter1138> But it means anything that uses closest town would be considerably faster?
11:58:34 <peter1138> Not just this feature.
11:58:42 <peter1138> TownID is what, 16 bits?
11:58:57 <Gabda> this feature is only a demo, and not part of the PR
11:58:58 <peter1138> 128KB on a normal map. That's ok.
11:59:14 <peter1138> 32MB on a stupid size map.
11:59:19 <peter1138> That's still nothing these days.
11:59:47 <Gabda> if you have a CPU that can handle the search for a stupid size map, you have the RAM as well
12:01:47 <nielsm> and passenger generation for those towns in ottd master is very similar to that in ttd dos
12:01:47 <Gabda> i just have to find the right places to initialize the map after load if I want to substitute the inside of the calcclosesttown with this
12:02:51 <peter1138> Gabda, quite. I have 32GB and that doesn't seem excessive these days.
12:03:15 <peter1138> I suppose closesttown is used in afterload and other stuff.
12:03:38 <andythenorth_> so given a 100mph pax vehicle, what cargo age period do I need to get an aging malus if the route > 32 tiles :p
12:03:43 <peter1138> You could perhaps make calcclosesttown check to see if the cache is valid
12:04:08 <nielsm> how about filling the cache with a sentinel value, then in the regular tile loop calculate the closest town and cache it, if a tile has the sentinel value stored
12:04:40 <nielsm> and if calcclosesttown finds then sentinel value when it checks the cache, it does the full calculation and updates the cache
12:05:05 <Gabda> i calc the closest town with a different method, not comparing all the towns
12:05:13 <Gabda> this is faster, and can fill the whole map
12:05:52 <nielsm> but caching it gradually this way avoids having a separate algorithm and should result much less new code overall
12:06:08 <nielsm> (perhaps 20 line patch total)
12:06:13 <nielsm> (for the caching part)
12:06:52 <Gabda> but you have to drop the whole thing is a new town is added
12:07:20 * andythenorth_ needs a spreadsheet :p
12:07:23 <nielsm> hm well yes, or invalidate the cache around the new or removed town
12:07:35 <nielsm> which could be annoying
12:07:53 <Gabda> but you don't know how far you have to invalidate
12:08:53 <Gabda> of course you don't add a new town that often
12:09:53 <peter1138> Gabda, new town is very rare.
12:10:21 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSzL
12:10:30 <Samu> what about scenario editor?
12:10:31 <peter1138> If it just needs the town coordinates, then you can calculate it early on when loading.
12:10:46 <Samu> you add towns often, there
12:10:46 <andythenorth_> hmmm....maybe I just set aging based on a play test, and never ever ever look at it again :)
12:11:10 <peter1138> Gabda, how long does it take to calculate?
12:12:00 <peter1138> 256x256, and 4096x4096?
12:12:17 <peter1138> Are we talking ms, seconds, or minutes?
12:12:33 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSzq
12:12:43 <Samu> 4096x4096 can generate about 13k towns max
12:13:18 <Gabda> i cannot really load a 4k x 4k map on the virtual machine i am working on :)
12:13:41 <Gabda> for 1024x1024 map with 2260 towns it is 10 ms for me
12:14:05 <nielsm> square of that is 100 ms, would probably not exceed that
12:14:43 <nielsm> might need square of square for 1k to 4k edge size?
12:14:50 <nielsm> in which case it'd be 10 sec
12:14:50 <Gabda> 256x256 and 166 towns is 0,45 ms
12:15:06 <peter1138> Wait, sub-milliseconds?
12:15:30 <peter1138> That's probably faster than an individual call to calcclosest? :P
12:15:41 <Gabda> 456 microsecond according to chrono
12:15:57 <peter1138> 10ms for 1024x1024 is no issue.
12:16:14 <peter1138> So the time it takes to recalculate is no issue for the rare times add town/remove town is done.
12:16:39 <peter1138> Unless... does town generation on generating a map need it?
12:16:57 <peter1138> Updating every time a town is added could get expensive on a large map with lots of towns.
12:17:00 <peter1138> Although it already is.
12:17:04 <peter1138> I dunno if it uses it though.
12:17:19 <Samu> gonna clone it, see what happens
12:17:24 <Gabda> it seems I can generate a 4k map without industries
12:18:58 <Gabda> 4096x4096 map 1739 towns (not much) 48 ms
12:19:47 <Gabda> adding a town is fast, it doesn't recalculate the whole map
12:20:06 <Gabda> only recalculate on removing
12:20:15 <Samu> 1739 towns, that's a bit lowish
12:20:20 <Samu> from what I'm used to se
12:21:38 <Gabda> I don't know why, but town generation can only squeeze in <3000 towns
12:22:57 *** andythenorth_ has joined #openttd
12:24:44 <Samu> gonna see what's happening
12:24:58 <Gabda> 4096x409 map with 37000 town (scenario editor) 135 ms
12:25:47 <Gabda> it think this town count is quite on the high side
12:26:29 <Samu> the absolute max is 64000 i think, or 65535 not too sure
12:27:03 <Samu> where do I start counting?
12:27:34 <andythenorth_> ok so penalty lower bound is cargo specific
12:27:42 <Gabda> it write the time to standard error
12:28:33 <Samu> towns were generated much faster than I used to experience
12:28:52 <Gabda> that has no relation with my PR
12:30:56 <andythenorth_> eh probably I need to control cargos then
12:31:13 <andythenorth_> shall I make an industry set? o_O
12:31:53 <Samu> are you sure? I'm either crazy, or something changed between 1.8.0 and your voronoi
12:32:05 <Gabda> Samu: the calculation only starts when you use a tool with rectangle selection like landinfo tool
12:32:05 <Samu> generating 13k towns felt faster
12:32:53 <Eddi|zuHause> <Gabda> I don't know why, but town generation can only squeeze in <3000 towns <-- running out of town names, usually
12:33:44 <Gabda> Eddi: yes, now I changed it to English, Samu said the same
12:34:03 <Samu> i need to compare with master
12:34:12 <Samu> there's definitely an increase in town generation speed
12:34:28 <Samu> from 1.8.0 to your voronoi, but maybe it's something that is also on master
12:34:59 <Gabda> should be something in master
12:36:23 <Gabda> I will try to put it into calc closest town after lunch, and we will see if town generation will be even better, or not
12:38:21 <Samu> building master, gonna compare with 1.8.0
12:38:48 <DorpsGek_II> [OpenTTD/OpenTTD] James103 commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSzl
12:41:31 <Samu> you're right, it's also faster on master
12:44:21 <DorpsGek_II> [OpenTTD/OpenTTD] M3Henry opened pull request #7165: [core] Implement SmallVector using std::vector https://git.io/fhSz0
12:49:50 <Gabda> yes, if you delete a town, it recalculates the whole thing
12:49:58 <Gabda> to fill the tiles of the deleted town
12:50:02 <Samu> world generator deletes towns
12:50:34 <Samu> really cool, but im not sure how to time this
12:51:59 <Gabda> come to think of it, the current version also calculates the whole voronoi on map generation, as towns are added to it one by one
12:52:00 <Eddi|zuHause> it should be possible to keep the deleting of town less impactful, but that might depend on using euclidean norm
12:53:01 <Eddi|zuHause> (have i mentioned yet that voronoi should use euclidean norm?)
12:53:29 <Gabda> the distance we are used to in the real life
12:53:38 <Gabda> and not Manhattan distance
12:54:15 <Gabda> Eddi: yes, you mentioned it :) but I have different opinion on this
12:55:26 <Samu> distancemaxplusmanhattan?
12:57:19 <Gabda> I think you can define voronoi with almost any of the norms, as it only contains information for every point on what item they are the closest from a set of items
12:57:59 <Gabda> of course, it is possible, that you don't have a dual for it
12:58:09 <Gabda> but that is not important now
12:58:31 <Samu> distancesquare doesn't sqrt
13:00:16 <Gabda> yes, because sqrt is expensive, and comparing Euclidean distances have the same result when the values are squared
13:01:14 <Eddi|zuHause> Gabda: but with euclidean it has a bunch of extended useful properties
13:02:22 <Eddi|zuHause> Gabda: plus, it makes it more intuitive for players what they expect "closest town" to be
13:03:26 <Gabda> Eddi: sure, but the game logic uses manhattan distance, and I only want to make the query faster. Changing the game logic is way above my head :)
13:05:28 <Gabda> and there is even a bug in CalcClosestTownFromTile, but fixing it breaks savefile compatibility
13:10:58 <Samu> duration 1167188 __int64 with 13k towns
13:11:40 <DorpsGek_II> [OpenTTD/OpenTTD] Gabda87 opened issue #7166: Incosistency in closest town calculation with threshold https://git.io/fhSzS
13:12:10 <Gabda> Samu: this is the whole map generation?
13:14:05 <Samu> release mode is maybe faster
13:15:27 <Samu> but i dont know how to measure this in release mode
13:16:35 <Samu> Variable is optimized away and not available. :|
13:17:22 <Gabda> if you start it from a cmd (I assume you use windows) maybe you can see the stderr there
13:19:08 *** frosch123 has joined #openttd
13:21:28 *** andythenorth_ has joined #openttd
13:21:52 * andythenorth_ understands cargo age now
13:22:33 <andythenorth_> the puzzle was why there is so little difference between 185 and 65534
13:24:09 <andythenorth_> but that value is only used above the lower bound
13:24:38 <andythenorth_> and is then multiplied by aging factor, which reduces rapidly
13:26:51 <andythenorth_> at least, that’s what nml docs imply :p
13:26:57 <Samu> Number of towns: 13308 Voroni build. dbg: [misc] [Town fill] 376885199 [avg: 376885199.0] 105921 microseconds. Voronoi initialized.
13:27:37 <Samu> much faster in release mode
13:33:55 <andythenorth_> i think I’d rather have a multiplier to price factor :p
13:34:44 <Gabda> scenario editor / many random towns does not use CalcClosestTownFromTile. I wonder how does it work.
13:39:44 <Samu> oh, start immediately is already mastered, heh cool thx
13:40:54 <Samu> must remove from my ai gui
13:43:29 <Samu> how come it rebased without conflicts?
13:46:03 <nielsm> andythenorth_, how about trains whose running costs increase massively when they run too far between stops at a station?
13:50:55 *** andythenorth_ has joined #openttd
13:51:26 <andythenorth_> nielsm variable running costs? o_O
13:52:04 <michi_cc> andythenorth_: The cargo ageing period affects how fast the payment point moves on the payment curve. The curve isn't linear, as such the ageing prop doesn't affect payment linearly either.
13:53:17 <andythenorth_> afaict, it shortens or lengthens aging period 1 or 2
13:53:22 <michi_cc> As the payment curve is rather flat at the start, for short routes a bonus does almost nothing and only a malus can be seem. On long routes, a bonus has a more visible period.
13:54:32 <michi_cc> andythenorth_: No, it modifies how the age of a cargo packet is counted. The age is what is put into the payment calculations (i.e. the days in the payment graph). The ageing period defines after how many ticks the age counter is increased by one.
13:54:58 *** Thedarkb-T60 has joined #openttd
13:55:13 <andythenorth_> ok so the malus will run age counter up very quicly
13:56:26 <michi_cc> If you double the ageing period from its default (from 185 to 370), cargo that is transported for 20 days will be calculated as if it had only travelled 10 days for the same distance.
13:57:03 <michi_cc> Depending on where on the payment curve that ends up, it can be a small or a big difference in payment.
13:57:11 <andythenorth_> ok that all matches the results I’m seeing
13:57:30 <andythenorth_> it’s hard to explain in docs I think
13:58:54 <michi_cc> Some useful things to do with the prop are for example local and mainline passenger coaches. More capacity and faster ageing for locals, less capacity and slower ageing for mainline. On short routes, it makes the local carriage naturally better (cargo age mostly irrelevant), while long routes prefer the mainline carriage.
13:59:40 <andythenorth_> i tried that :)
13:59:40 <michi_cc> Similar if you have refrigerated carriages and e.g. normal box cars that both can carry the same cargo.
13:59:51 <andythenorth_> it technically works
13:59:52 <DorpsGek_II> [OpenTTD/OpenTTD] PikkaBird commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSgJ
14:01:09 <andythenorth_> the gameplay result is negligible though
14:01:42 <michi_cc> Anything with money is mostly negligible :)
14:10:21 <peter1138> Gabda, so basically, as long as it doesn't need to recalculate every tick, performance is fine.
14:16:12 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSgY
14:20:01 *** synchris has joined #openttd
14:20:21 *** andythenorth_ has joined #openttd
14:22:58 <andythenorth_> michi_cc so what’s a “long” route? factor of vehicle speed and cargo props?
14:24:40 <michi_cc> Anything on the right side of the cargo payment graph?
14:24:41 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSgC
14:25:40 <andythenorth_> I will set up a test map later and see :)
14:26:07 <andythenorth_> the malus (< 185) definitely significant so far
14:26:37 <andythenorth_> values above 185 are barely relevant all the way up to 65534
14:27:22 <michi_cc> I suspect you need a travel time of 120 days or more to see a noticeable result.
14:27:52 <andythenorth_> yeah I’m down at 30 days or so
14:28:20 <Gabda> peter1138: that is what I want to avoid with this cache
14:29:02 <Gabda> as I plan to create some zoning display
14:29:40 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSgW
14:29:53 <peter1138> I'm just wondering what other features we've said we can't do due to performance...
14:29:59 <peter1138> railtype/roadtype varaction variables...
14:31:34 <andythenorth_> splitting cargo to indistries when delivered to station
14:32:04 <andythenorth_> the North tile industry station thing is so janky :)
14:35:27 <andythenorth_> rubi told me that distributing cargo to n industries was not possible for performance reasons
14:35:45 <andythenorth_> because newgrf industry makes it inpossible to cache
14:36:44 <andythenorth_> or at least, far too expensive due to cache churn
14:45:43 <DorpsGek_II> [OpenTTD/OpenTTD] michicc requested changes for pull request #7168: Fix CompanyEconomy documentation https://git.io/fhSgV
14:53:20 <Samu> I forgot spectators also have goals
14:59:42 <peter1138> How can they have goals and story books?
14:59:53 <Samu> oh it's just the goals, apparenlty
15:00:01 <Samu> story book is grayed out
15:00:17 <Samu> it says Global Goals instead
15:01:45 <peter1138> So it means goals for everyone?
15:02:09 <Samu> spectator can also have a story book
15:02:20 <Samu> hmm this is complicating...
15:02:20 <peter1138> I... don't know what a story book is.
15:02:24 <peter1138> I've never heard of it.
15:02:44 <Samu> the button next to general company information
15:03:27 *** Eddi|zuHause2 has joined #openttd
15:05:24 <Samu> spectator story book option is only enabled when a company has a story?
15:06:37 <Samu> anyway... Spectator option must be present, or else the list asserts, because it is empty when i click the button
15:07:21 <Samu> I blame glx, he did not want spectator option
15:09:05 <Pikka> Eddi|zuHause2, I have no problem with dismantling the entire concept of cargodist ;)
15:09:27 *** Eddi|zuHause2 is now known as Eddi|zuHause
15:10:02 <peter1138> I think I prefered cargo*dest* but...
15:10:40 <peter1138> Hmm, some of these new colours are just old colours renamed :p
15:11:24 <Samu> btw, Spectator should be white
15:11:39 <Samu> or else, when selecting it becomes fully black
15:16:32 <Samu> STR_NETWORK_TOOLBAR_LIST_SPECTATOR :{WHITE}Spectator
15:17:08 <Samu> do i have to delete the string from the language files?
15:17:29 <Samu> could I actually rename BLACK to WHITE on the other language files?
15:18:04 *** andythenorth_ has joined #openttd
15:19:37 <michi_cc> Hmm, actually, who knows about network code? There's a "p->Send_uint64(income);" with income being an OverflowSafeInt64. Is this supposed to work with negative income?
15:20:40 <Samu> unless it really is an incom
15:21:55 <Samu> i feel like fixing all language files, dunno if it's okay in this case
15:22:23 <LordAro> Samu: as long as it's a separate commit, changing codes like that is fine
15:23:01 *** Thedarkb-T60 has joined #openttd
15:24:04 <peter1138> Hmm, "Turquoise" is the same as "Light Blue"
15:24:18 <michi_cc> Okay, the C++ standard seems to define how this cast has to be performed. And the definition results in the same bits, so everything is alright.
15:27:39 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #7164: Fix #7108: Group livery command did not check its parameters properly. https://git.io/fhS2s
15:29:42 *** Gustavo6046 has joined #openttd
15:31:57 <michi_cc> nielsm: Probably because MacOS is listed as abandoned, even if I don't know what that means.
15:32:09 <Samu> "Fix: Spectator could be white" ... too racist :(
15:32:25 <Samu> i mean black, even worse
15:32:53 <Samu> Spectator wasn't visible when highlighted?
15:34:31 <Samu> hope it's not sounding offensive that way
15:38:33 <TrueBrain> I put a new one in the queue
15:40:58 <Eddi|zuHause> "abandoned" sounds like "got lost in the mail"
15:41:40 <TrueBrain> I read other stories that the MacOS was being weird
15:41:45 <TrueBrain> Azure had issues this week anyway
15:42:11 <TrueBrain> it is free, what can I say. If we would pay for this, I might be more upset :P
15:43:22 <michi_cc> We're now a proprietary shop: MS GitHub, MS Azure, commerical CDN :p
15:43:52 <TrueBrain> and I am so happy because of it :) Less maintenance .. so much less
15:44:04 <LordAro> just works a bit less :p
15:44:24 <TrueBrain> custom-build is always better in terms of "working" and "workflow", yup :)
15:45:02 <TrueBrain> random quotes are random
15:46:24 <michi_cc> So why's our website not on e.g. zeit.co ? :p
15:46:26 *** andythenorth_ has joined #openttd
15:46:58 <Eddi|zuHause> why is our website not wordpress? :p
15:48:04 <TrueBrain> "100 GB of Bandwidth per month" <- okay, that is not for us :P
15:48:21 <TrueBrain> Eddi|zuHause: that we want to make things more "mainstream", doesn't mean we have to act stupid :P
15:48:52 <michi_cc> There is a pay option.
15:49:02 <TrueBrain> "$0.1 per GB of Bandwidth"
15:49:14 <TrueBrain> would cost us 400 dollar per month ..
15:49:26 <TrueBrain> but yeah, zeit sounds nice for small websites :)
15:52:58 <TrueBrain> okay ... turns out BaNaNaS mirroring is a bit broken :D
15:55:53 <DorpsGek_II> [OpenTTD/OpenTTD] flitzpiepe commented on pull request #7145: TBTR 2.0 (Template-based Train replacement) https://git.io/fhS2V
15:55:59 <TrueBrain> ah .. that was a sillymistake
15:59:14 *** andythenorth_ has joined #openttd
15:59:35 <milek7> why they are using 'serverless' to mean 'runs on some cloud service'?
15:59:38 <TrueBrain> NO andythenorth_, NO :P
15:59:46 <andythenorth_> free wp.com account
15:59:47 <milek7> for me 'serverless' means something like 'clients download content through torrent with dht/ipfs' ;p
15:59:55 <TrueBrain> milek7: dont you just love the word 'serverless' :D
16:00:06 <andythenorth_> i saw a diagram
16:00:07 <TrueBrain> I havent seen that definition, tbh
16:00:20 <andythenorth_> how many servers are needed for serverless
16:00:38 <peter1138> Hmm, anyone know what the original colour maps are? :p
16:00:47 <peter1138> Or do I have to... check?
16:00:53 <TrueBrain> I love how people think 'serverless' means there is no server involved :)
16:01:46 <andythenorth_> haha with 5G it’s all going to ‘run in the network’
16:01:56 <andythenorth_> whatever that means
16:11:15 <DorpsGek_II> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7145: TBTR 2.0 (Template-based Train replacement) https://git.io/fhS29
16:27:23 <Samu> no more crash due to lack of spectator
16:29:04 *** andythenorth has joined #openttd
16:29:15 <peter1138> nielsm, maybe those exceptions are the rule for this? I dunno. Odd.
16:29:19 <peter1138> andythenorth, did you see my WIP?
16:30:10 <peter1138> Although some of them are duplicates.
16:30:13 <andythenorth> is that just 'more'
16:30:23 <andythenorth> grf defined? Or built in?
16:30:25 <peter1138> Because NewCC doesn't define 16 new colours, just 16 replacement colours.
16:30:28 <peter1138> Built in, at the moment.
16:30:48 <peter1138> We should sort the colour list by hue :p
16:31:42 <andythenorth> child #2 tipped them all out this morning
16:31:47 <andythenorth> they are less ordered now :P
16:32:16 <andythenorth> so I should make a test game for cargo age thing
16:32:24 <andythenorth> flat, or with hills? :P
16:36:45 <Gabda> further testing of the Voronoi map: in scenario editor (4096x4096 map) adding the first 1280 random towns takes 1.9 sec, from this the 1280 separate update of the voronoi map took 18 microseconds overall.
16:37:11 <Gabda> i forgot a 0, it took 180 microseconds
16:37:54 <Gabda> still, adding a new town to the Voronoi map is quite cheap
16:38:05 <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick updated pull request #7158: Add: Client setting gui.start_spectator https://git.io/fhSk4
16:38:12 <peter1138> Does it need to update 1280 times, or can it do it once after the towns are placed?
16:38:29 <peter1138> I guess creating a town my need calcclosesttown.
16:39:16 <peter1138> I don't want opinion, just facts:p
16:40:06 <Samu> do it once map is generated
16:40:27 <Samu> then do it once each time a town is added
16:40:57 <Gabda> peter1138 even said why it is necessary to update it after every add
16:41:46 <Gabda> if rerouted the IsCloseToTown function to CalcClosestTownFromTile, which calls the data from the voronoi map, so I need to update it after every new town
16:42:51 <Gabda> but I think if you want to optimize something about the 1.9 seconds, it is not the best strategy to hack from the 180 microsecond part :D
16:43:20 <peter1138> Yeah, 1.9 seconds is fine.
16:43:29 <peter1138> It's nothing when generating a map.
16:43:53 <Gabda> it is just generating 1280 towns to scenario editor
16:44:18 <Gabda> I am kind of interested what is the slow part in it
16:44:36 <Gabda> building up the individual towns?
16:44:39 <glx> 1.9s is nothing if you are adding ECS industries after it ;)
16:45:26 <peter1138> A set of newgrf industries with complex placement requirements.
16:45:33 <peter1138> That requirements make it... slow.
16:46:30 <Gabda> can those requirements be cached beforehand? I am into caching nowadays :D
16:47:12 <Samu> how does cherry-pick works when I make changes to the original?
16:47:15 <nielsm> no, would not be realistic
16:47:24 <Samu> will they get updated automatically?
16:47:28 <nielsm> cherry-pick copies the commit you pick from
16:47:30 <Samu> will need to re-cherry pick?
16:48:25 <Gabda> you even have to make the first cherry-pick disappear
16:48:44 <Samu> that's what i'm gonna do
16:52:22 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7164: Fix #7108: Group livery command did not check its parameters properly. https://git.io/fhSuq
16:56:10 <Gabda> in case there are no windows open, why is there such a big difference in lag between MarkWholeScreenDirty() and scrolling the viewport?
16:56:41 <peter1138> Scrolling the viewport actually copies what was already there, moves it, and then only redraws what's new.
16:56:43 <Gabda> scrolling the viewport should make the whole thing dirty as well
16:58:45 *** Thedarkb-T60 has joined #openttd
16:58:49 <Samu> start date = 0 + spectator slot + fast forward + 15 ais in single player, oh the joy!!!
17:00:41 *** Wormnest has joined #openttd
17:02:35 <Gabda> peter1138: do you also know why does it take longer to redraw the screen on a 4096x4096 map than a 512x512 map while fully zoomed out in both cases?
17:02:46 <Gabda> the number of tiles displayed are the same
17:03:13 <glx> map size to filter the display
17:03:42 <TrueBrain> LordAro: still want to try a beta this weekend? Or are we going to try it again next? :)
17:03:51 <glx> probably a for all tiles, if visible draw it
17:04:53 <LordAro> TrueBrain: not out of the question, what needs doing?
17:05:00 <LordAro> is it just changelog & other docs?
17:05:03 <TrueBrain> master needs to be prepared for it, mostly :)
17:05:09 <TrueBrain> I really dont know what you normally do for it
17:05:12 <TrueBrain> please don't tag just yet
17:05:22 <TrueBrain> means I can test-run stuff before we do that
17:05:46 <TrueBrain> I can already start a test-run ofc; not sure what we expect out of it currently :)
17:05:53 <glx> hmm there was a txt somewhere for releases I think
17:07:12 <glx> at minimum there's something to do in rev.cpp.in
17:07:18 <TrueBrain> currently deb files are called 'openttd_1.8.0-0_i386.deb' for example
17:07:24 <TrueBrain> that surely needs fixing :)
17:07:50 <frosch123> kate changelog.txt known-bugs.txt readme.txt os/windows/installer/install.nsi os/debian/changelog src/script/api/ai_changelog.hpp src/script/api/game_changelog.hpp &
17:08:16 * andythenorth lost in cargo aging tests
17:08:26 <frosch123> maybe files got renamed to .md, or we removed those silly dates :p
17:08:30 <andythenorth> I need a bigger screen
17:12:04 <TrueBrain> LordAro: for this first run, I will not make it automatically trigger on tag; but that is something we can in the future (which might be nice :D)
17:12:08 <Gabda> it seems like MarkWholeScreenDirty() gets worse with the total number of towns
17:12:54 <Gabda> while the same thing is on the screen
17:13:27 <andythenorth> so with default cargo aging, it needs a 256 tile route before the 'high speed' train (140mph) is the most profitabl
17:13:53 <andythenorth> up to 128 tiles, the low-speed EMU is the most profitable by a long measure
17:14:20 <LordAro> TrueBrain: i'll see what i can come up with
17:14:33 * andythenorth runs a few more years to account for variation
17:15:26 <LordAro> interesting, have we not had a beta since 1.5 ?
17:17:16 <peter1138> Ok, 32 company colours mostly works.
17:19:45 <andythenorth> TrueBrain: come up with names for 64 colours :P
17:19:57 <peter1138> Going above 254 would be problematic.
17:20:03 <TrueBrain> sure: blue, light blue, light blueish, lightish blue, lisghtish blueish
17:20:06 <TrueBrain> I can go on all day
17:20:09 <peter1138> And it's 32 only because I only had another list of 16 colours.
17:20:20 <peter1138> Although it turns out many of those were the same as default, so.. meh.
17:20:23 <andythenorth> sometimes 'more' isn't more
17:21:16 <TrueBrain> glx: "nsis - nsis v3.04 already installed." <- seems their images are finally updated :P
17:21:28 <Gabda> the main viewport is an instance of the ExtraViewportWindow class as well?
17:21:45 <TrueBrain> hmm .. there are currently language errors in master btw
17:22:42 <peter1138> Pikka, and still an AI can't pick any of them.
17:23:58 <glx> but I guess it doesn't include the extensions we need
17:24:16 <TrueBrain> "Warning : STR_LIVERY_CAPTION: Param idx #0 '<empty>' doesn't match with template command 'COMPANY' (warning)"
17:24:41 <TrueBrain> well, this is when you compile OpenTTD currently :)
17:25:03 <glx> it readds removed string if present in master
17:25:26 <TrueBrain> meh; this makes a beta a bit difficult :P
17:28:50 <LordAro> i'm busy preparing a release :p
17:28:59 <TrueBrain> a release with language warnings is a bad idea :)
17:29:02 <glx> hmm not exactly the same, but similar
17:29:04 <andythenorth> michi_cc: any suggested values cargo age to give a bonus? o_O
17:29:16 <andythenorth> I am trying 10 * default currently
17:33:10 <glx> is the .pdb stored somewhere too ?
17:35:51 <glx> ah yes seems it is 2019-02-03T16:25:29.9493162Z [BUNDLE] Creating openttd-20190203-master-gcca952d9-windows-win32.pdb.xz
17:36:42 <glx> but naming is not good (I know it's a test run and probably depends on something from the release commit)
17:41:33 <LordAro> oh, i guess #7121 should probably happen before any actual release
17:41:37 *** Thedarkb1-T60 has joined #openttd
17:41:38 <Samu> peter1138: theres something wrong with GSes, they're not updating that well
17:42:09 <peter1138> What does that even mean?
17:42:20 <Samu> taking my Company Value GS
17:42:28 <Samu> in 1.8.0 the update is instant
17:42:34 <Samu> in master, it's ... sluggish, delayed
17:43:00 <Samu> easy to spot when multiple companies are playing (AIs)
17:43:13 <Samu> the scoretable is updated instantly on 1.8.0
17:43:55 <peter1138> I don't know what "the update" refers to.
17:44:15 <Samu> the refreshing of data to the new values
17:44:25 <Samu> the printing to screen in the goal list
17:44:32 <peter1138> Do you mean refreshing the window?
17:44:45 <peter1138> Are you using FFWD?
17:44:53 <peter1138> Then it's expected.
17:45:21 <peter1138> Maybe you need some invalidate calls.
17:46:15 <Samu> uh invalidate calls from the GS?
17:46:36 <Samu> they dont have access to that
17:46:37 <peter1138> Which is the score window?
17:46:53 <nielsm> hmm, how about changing the svn revision in newgrf version number to 0x8000 and then manually increment it by 1 once in a while?
17:47:41 <andythenorth> if you build a train in pause, the clone button doesn't appear on it until unpaused
17:47:52 <Samu> doesnt refresh properly in normal speed either
17:48:17 <glx> increment by 1 after each stable release probably
17:49:14 <LordAro> 190000 , or is that too many bits?
17:49:23 <peter1138> There's no tick handler in the goal list.
17:49:23 <nielsm> there's 19 bits available
17:49:39 <peter1138> So unless the window is invalidated, there's no reason for it to be repainted.
17:49:48 <glx> but the highest bit means stable
17:49:52 <nielsm> decimal 190000 should also fit
17:50:24 <nielsm> glx: that's bit index 19 (the 20th bit) that means stable
17:51:46 <glx> just incrementing current 'svn' version after stable release should be enough I think
17:52:38 <Samu> not sure if it's only my GS that is affected
17:53:17 <Samu> AIs seem to be doing fine, at least I don't spot anything wrong with them
17:54:30 <glx> because next version needs to be higher than 28004
17:56:00 <LordAro> frosch123: where i meant to look up the PR number but forgot :p
17:56:35 <frosch123> we used to put all issue numbers at the end
17:57:18 <LordAro> mm, i'm open to other suggestions
17:58:26 <peter1138> Can we have 7121 before beta?
17:58:46 <LordAro> peter1138: i pressed the approve button, but apparently DorpsGek_II didn't notice
17:59:02 <LordAro> i think all issues with it have been addressed?
17:59:25 <glx> but splitting the available revision bits in major, minor, build could work too
18:00:20 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh merged pull request #7121: Fix #7021: Better revision strings for network and gamelog https://git.io/fh6im
18:00:36 <nielsm> didn't notice before now, but #7121 closes #7021
18:00:45 <nielsm> exactly 100 difference
18:00:59 <Samu> let me test in a multiplayer game
18:01:07 <nielsm> yay no more dumb warnings spamming the console
18:01:22 <glx> anyway the rev part was mostly used to check feature presence in nightlies
18:01:51 <nielsm> yeah, as long as the svn rev doesn't decrease it should be fine
18:02:29 <glx> théorically we could have increased it for the 16 cargos
18:03:19 <glx> maybe we did, not checked
18:03:55 <Samu> gonna try SCP turned off
18:04:24 <Samu> script communication protocol
18:04:31 <Samu> something from krinn,zuu
18:05:12 <Samu> ehm it's somewhere in the AI forum
18:06:04 <andythenorth> after more tests
18:06:15 <andythenorth> cargo age period bonus is useless in the 256 tile range
18:06:26 <andythenorth> it's a low % difference
18:06:37 <andythenorth> cargo age period malus is highly effective
18:06:57 <andythenorth> so much so, that it needs care to set it, otherwise it's easy to get negative profit
18:07:45 <andythenorth> the malus is visible on all routes, but only really kicks in hard for 128 or so
18:07:51 <glx> communication via signs IIRC
18:08:08 <Samu> looks fine with SCP disabled
18:08:13 <andythenorth> not sure what to next, I feel like I'm turning into Samu :)
18:09:11 <peter1138> I'm not sure how a script would make the window not refresh.
18:09:42 <Samu> the window wasn't updating because SCP was getting stuck somewhere
18:10:14 <Samu> some of the recent changes broke SCP support
18:10:20 <glx> post a useful link when you can ;)
18:10:42 <peter1138> So ultimately your gamescript was not updating its goals, rather than the window update being laggy?
18:11:03 <Samu> yes, my script, with SCP library into it, yes, i guess
18:11:43 <Samu> I'm not the author of SCP, I don't know what to say about that
18:12:06 <Samu> but it's what was causing the window not to update, it was being delayed by SCP
18:12:29 <Samu> that delay isn't present in 1.8.0 though, something got broken
18:19:39 *** andythenorth_ has joined #openttd
18:23:31 <peter1138> But the delay with SCP *isn't* causing the window to no update properly.
18:25:33 <glx> yeah the window works as it should
18:25:57 <peter1138> Not window-tick-factor :D
18:26:20 <Samu> well, something broke scp that it now delays at doing its thing
18:26:56 <Samu> which then delays execution of the rest of my code that does the updating
18:26:57 <peter1138> Maybe someone could report an issue about it, probably do the authors.
18:27:27 <glx> or someone could try to find when the problem started
18:27:52 *** Thedarkb2-T60 has joined #openttd
18:28:28 <peter1138> I wonder about this sprite sorting issue. I thought sprite sorting happened after ViewportAddLnadscape()
18:28:35 <peter1138> Hmm, and it still does.
18:29:25 <peter1138> So calls to ViewportAddTile() shouldn't really do anything different regardless of the order they are in.
18:30:06 <peter1138> Hmm, unless ViewportAddLandscape is cropping too early.
18:31:13 <Samu> gonna try an earlier than tick changes master
18:31:43 <peter1138> If it communicates via signs, then tick changes wouldn't alter that.
18:32:24 <peter1138> The tick changes were GUI only, and should not affect how the gameloop runs.
18:36:15 <LordAro> Samu needs to learn the concept of git bisect
18:37:17 <LordAro> well how do you find out things you don't know about, in this day and age?
18:40:07 <nielsm> hm pondering if I should remake the PR for town cargogen with a new branch name, or just stick to the dumb branch name (and same PR) I used initially
18:41:15 <LordAro> nielsm: i'd prefer as few "closed" PRs as possible :)
18:41:26 <LordAro> the branch name is ultimately irrelevant
18:41:43 <TrueBrain> now I want to know what the branchname is :P
18:44:22 <nielsm> TrueBrain, initially I used all of M8 to store some data between tile ticks on houses, hence "cargogen counter" :D
18:45:35 <glx> house-cargogen-counter is the name
18:46:09 <glx> maybe github supports branch renames
18:46:24 <glx> it's a git feature anyway
18:47:12 <nielsm> as far as I can tell the web ui doesn't allow renames
18:49:29 <LordAro> which is why it's a bad idea to use your master branch to make a PR :p
18:49:51 <LordAro> as an example, yes :p
18:50:38 * peter1138 ponders switching to a WM that doesn't mess up moving/resizing OpenTTD.
18:54:35 <Samu> forwarding to Commit 5ff0c249
18:55:14 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh updated pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fpkqa
18:56:23 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fhSwL
18:57:21 <Samu> looks fine, it wasn't the OnTick() stuff
18:57:28 <Samu> well, what could it be now :(
19:01:32 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh opened issue #7171: determineversion.vbs never marks a version "clean" https://git.io/fhSwZ
19:03:31 *** Progman has joined #openttd
19:03:41 <LordAro> Samu: what was the state of 5ff0c249?
19:04:24 <LordAro> Samu: well, go forward a few commits until you find the point at which it is broken
19:04:37 <Samu> forwarding to before those afd88 stuff
19:04:46 <LordAro> (this is the key part of git bisect - it performs the search for you)
19:05:31 <peter1138> Yeah, just 2 choices.
19:05:37 <peter1138> Origin, and whatever the best one is.
19:06:08 <nielsm> I of course prefer my own binomial version because it's more simulation-ish
19:07:06 <DorpsGek_II> [OpenTTD/OpenTTD] michicc approved pull request #7118: Add #5006: Flag to hide rail type from construction. https://git.io/fhSrm
19:07:30 <peter1138> A super-hacky hacked test.
19:07:39 <milek7> make UI accept squirrel expression to calculate.. ;D
19:08:23 <nielsm> would it be problematic if I consumed another random number in the town cargo gen? the original method consumes one 32 bit value, but I'd rather like to consume two
19:08:53 <planetmaker> nielsm, what exactly does the distribution refer to? Cargo generation for different towns?
19:09:33 <peter1138> nielsm, I don't see why it would be a problem.
19:09:53 <nielsm> all town buildings use the same cargo gen algorithm, unless it's a newhouse and the grf has a cargo gen callback
19:09:56 <glx> of course you need to check town newgrf still works as intended ;)
19:10:00 <nielsm> it's not a per town setting
19:11:36 <nielsm> there's two "distribution"s involved, sort-of, one is how the overall amount of cargo generated scales with the population
19:11:45 <planetmaker> I see that it relates cargo generation to inhabitants. I'd not call it 'distribution' then. It's a relation
19:11:59 <nielsm> the other is how the variance in amount between invidual generation events is
19:12:49 <glx> hmm I'm sure generate.vbs used to work, but maybe only for svn
19:13:00 <planetmaker> A distribution implies a frequency (how often what value occurs on the map). A relation hom much a value scales with another
19:13:08 <Samu> still fine on Commit ed325ada
19:13:27 <Samu> forwarding a bit past adf88
19:13:31 <nielsm> I'll fix the terms used
19:13:44 <glx> well determineversion.vbs
19:19:02 <andythenorth> what's the appetite for changing cargo payment algorithm? :P
19:21:13 <Samu> still fine on Commit 36e299fb, that's past adf88 huge stuff
19:21:17 <planetmaker> andythenorth, only when it considers inter-company logistics :)
19:22:12 <milek7> Samu: you know about bisect?
19:22:39 <Samu> it's bad , it's doing bad
19:23:07 <LordAro> Samu: get an actual commit hash the broke it
19:24:20 <nielsm> hmm is (1<<32)-1 well-defined for 32 bit unsigned integers in C?
19:25:49 <nielsm> is that guaranteed or might a processor trap on it? :)
19:27:34 <peter1138> Hmm, might end up with 1?
19:27:37 <milek7> 6.5.7/3 [...] If the value of the right operand is negative or is greater than or equal to the width of the promoted left operand, the behavior is undefined.
19:28:31 <nielsm> okay I'll protect against it
19:30:15 <Samu> don't tell me it's Mail AI spamming signs and SCP reading them all
19:30:29 <Samu> omg, gonna try current master again
19:31:07 <Samu> yeah, reading 4000 signs ought to be slow
19:32:05 <andythenorth> I am concerned that the only way cargo_age_period really works is to give a malus
19:32:19 <andythenorth> so to get a 'bonus' for some vehicles, I need to malus everything else
19:32:59 <planetmaker> andythenorth, no, you can give a bonus as well
19:33:21 <planetmaker> there's a default value after which the cargo looses some value. And you can set it to a lower or a higher value
19:33:27 <planetmaker> via that very property
19:33:47 <glx> nielsm: hmm indeed determineversion is wrong, but wasn't an issue before because only modified state was checked
19:34:40 <planetmaker> Supported by OpenTTD 1.2 (r22713)1.2 This property specifies after how many ticks cargo is aged. Default value is 185. 74 ticks is equal to 1 day.
19:34:46 <Samu> just made 1.8.0 delayed too, it's caused by Mail AI that thought it would be funny to spam the map with 'T' signs
19:35:06 <Samu> testing master again just to confirm
19:35:44 <planetmaker> andythenorth, so any value lower than 185 ticks is a malus, anything higher is a bonus
19:37:11 <andythenorth> bonus is marginal
19:37:32 <andythenorth> for pax cargo with ~default props
19:37:37 <Samu> I thought SCP would only read the signs from one tile
19:37:46 <andythenorth> it's low % difference whether you set 185 or 65534
19:37:48 <Samu> apparently reads all signs in existance
19:37:51 <planetmaker> The effect of the bonus really depends. It is negligible for small maps and short distances. But so is the malus
19:38:02 <planetmaker> The effect is MUCH bigger for large maps, long delivery times
19:38:07 <andythenorth> the malus is pretty effective
19:38:18 <andythenorth> the bonus probably shows up a lot more for, e.g. ships
19:38:29 <planetmaker> for slow vehicles in particular
19:40:25 <andythenorth> I have trains at 140mph over 256 tiles, at that speed for that distance, 185 or 65534, the difference to profit is basically a rounding error
19:40:42 <andythenorth> kinda fine, but I should have tested this before designing horse
19:40:56 <andythenorth> I only noticed it doesn't work because of note Snail made in French grf
19:40:57 <planetmaker> because it probably doesn't take more than 185/74 days to deliver the cargo
19:41:14 <andythenorth> it's working correctly
19:41:22 <andythenorth> I just mis-designed my grfs
19:41:28 <andythenorth> and now have to decide how to fix that
19:42:35 <planetmaker> what is the intended behaviour?
19:42:52 <Samu> sorry everyone, it was a false alarm
19:43:02 <planetmaker> independent of newgrf properties or stuff
19:43:33 <Samu> it was because MailAI had not started in 1.8.0 during my initial comparisons
19:44:00 <Samu> SCP takes time to read signs, and MailAI spams 4000+ 'T' signs everywhere, hence the delay
19:44:07 <andythenorth> intended behaviour is to give a payment bonus to, e.g. refrigerated cars, luxury pax cars etc
19:44:15 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 opened pull request #7172: Fix #7171: incorrect modified status with determineversion.vbs https://git.io/fhSKp
19:45:07 <planetmaker> yes, you probably want to reduce the property for the non-luxury vehicles, I guess. If you target the 256 tiles... which is reasonable
19:45:27 <planetmaker> or you might add some scale depending on map size
19:45:29 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #7172: Fix #7171: incorrect modified status with determineversion.vbs https://git.io/fhS6J
19:45:33 <andythenorth> it requires a little more care to add a malus
19:45:56 <andythenorth> bonus can't do much harm
19:46:06 <andythenorth> the malus I'm testing, to have any impact, can make profits negative
19:46:51 <andythenorth> if I do a safer malus, it has much less effect
19:47:39 <michi_cc> andythenorth: If your food as spoiled due to no refrigeration, a negative profit seems quite alright :)
19:52:56 <andythenorth> green trains are 64 tile, yellow is 256
19:53:25 <andythenorth> the EMU (most square one in each group) has malus 32 here
19:53:39 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 merged pull request #7172: Fix #7171: incorrect modified status with determineversion.vbs https://git.io/fhSKp
19:53:53 <andythenorth> the other two types are default, but the white one is faster
19:54:07 <andythenorth> all have roughly same capacity, but different speeds and costs
19:56:34 <planetmaker> you can basically only compare same route
19:57:58 <planetmaker> hm... bad then :|
19:58:27 <andythenorth> the colours are the same length route
19:58:43 <andythenorth> there are 3 different lengths there though
19:59:09 <andythenorth> I think this _might_ be fine, it just smells wrong
19:59:22 <planetmaker> you see a 20% difference for the longer route
19:59:39 <planetmaker> actually. For *every* route length
20:00:18 <LordAro> so... does anyone want to properly review #7170?
20:00:21 <planetmaker> comparing the white-head train to the one with a normal engine
20:00:35 <planetmaker> you want more than 20%?
20:01:25 <planetmaker> 50% or 100% might be nicer
20:02:09 <andythenorth> the white train is just faster :)
20:02:24 <andythenorth> the cargo age period is default there, so that's just normal expected result
20:02:27 <planetmaker> yes. And that's what primarily affects cargo age stuff
20:03:18 <andythenorth> the EMU with malus 32 is absolutely boss at 64 tiles
20:03:29 <andythenorth> but drops off horribly somewhere after 64
20:03:58 <planetmaker> It absolutely serves its purpose then
20:06:45 <andythenorth> and train capacity is normalised - they're all about equal pax capacity here
20:07:04 <andythenorth> the EMU is still boss in red (16 tile)
20:07:10 <andythenorth> and takes up a lot less space
20:07:19 <andythenorth> at 32 tiles it's all equal
20:07:30 <peter1138> Hmm, working version:
20:07:37 <peter1138> Drawing tile 13ba, 13bb
20:07:41 *** Wormnest_ has joined #openttd
20:07:42 <peter1138> Non-working version:
20:07:45 <peter1138> Drawing tile 13bb, 13ba
20:09:23 * nielsm should really get some food
20:09:45 <planetmaker> 32 tile all equal is really short distance :)
20:10:05 <andythenorth> I do a lot of those kind of distances
20:10:13 <andythenorth> I usually play 256x256
20:10:21 <andythenorth> I hate laying track :P
20:10:28 <peter1138> Ok but which is better, nielsm
20:10:32 <michi_cc> andythenorth: It's not really about the distance though, but the travel time. Early game with sub-100 mph trains and larger maps is the domain of ageing bonuses.
20:10:49 <andythenorth> so that is my other concern about the malus :)
20:10:58 <andythenorth> I have 6 generations of trains, with speed increments
20:11:05 <andythenorth> if I malus the slow ones....might break
20:11:15 <andythenorth> so I need to do it with care
20:11:21 <andythenorth> (slow = early gen)
20:11:29 <planetmaker> my preferred size is like 256x1024 or so. Small. Yet place for some decent building and some long-haul stuff
20:11:43 <andythenorth> ok so I need to swap all these trains for early ones
20:11:46 <andythenorth> autoreplace time :P
20:11:54 <planetmaker> combining the local service with the long-distance one
20:11:57 <andythenorth> did anyone finish train templates yet?
20:12:22 <peter1138> Hmm, so 479f13fc413 is at fault, because it assumes it can draw landscape tiles in any order.
20:12:30 <planetmaker> that's the 207-patch PR (which was redrawn for re-write)
20:12:37 <michi_cc> I guess you might want something along the lines of bonus at the start for mainline and slowly shifting to a malus for the local trains over time.
20:12:46 <planetmaker> well... as GH did it when he removed the first merge in it)
20:13:06 <peter1138> 'Tweak ViewportAddLandscape so it no more relies on "go down as fast as possible"'
20:13:22 <peter1138> But I think this "go down" was necessary to get sprites in the right order.
20:14:57 <andythenorth> so I should test a bonus on same map with the slower train
20:15:04 <andythenorth> I think it will be marginal effect still
20:15:10 <andythenorth> for this default pax cargo
20:15:29 <andythenorth> it won't take long enough to be delivered
20:16:22 <peter1138> So yeah, fine without that commit.
20:16:54 <peter1138> Heh, JGR came to the same conclusion
20:18:32 <andythenorth> this group liveries thing makes these tests so much easier :P
20:19:08 <peter1138> Yeah, reverting that fixes #7136 as well as #7133
20:19:11 <planetmaker> true that :) Easily discernible
20:22:55 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7173: Revert 479f13fc41, Fix #7133, Fix #7136: "Codechange: Tweak ViewportAddLandscape so it no more relies on "go down as fast as possible" tile height model (Patch by adf88, #6583)" https://git.io/fhS6b
20:23:26 <LordAro> peter1138: how badly does it break viewport movement?
20:24:03 <andythenorth> the malus here hurts the EMU more, it's only 60mph
20:24:07 <andythenorth> so it's losing even on short routes
20:24:43 <andythenorth> but the red group are all pretty even
20:24:53 <andythenorth> and you get a lot more EMUs per tiles :P
20:29:55 <DorpsGek_II> [OpenTTD/OpenTTD] michicc approved pull request #7173: Revert 479f13fc41, Fix #7133, Fix #7136: "Codechange: Tweak ViewportAddLandscape so it no more relies on "go down as fast as possible" tile height model (Patch by adf88, #6583)" https://git.io/fhS6j
20:34:25 <andythenorth> so for 'normal' freight wagons, cb 36 on refit, and malus if cargo is perishable? michi_cc :P
20:38:25 <peter1138> LordAro, doesn't affect it.
20:38:38 <peter1138> LordAro, it's another "some unrelated changes" PR :/
20:39:15 <peter1138> If adf88 was around I'd check with him, but alas.
20:39:24 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7173: Revert 479f13fc41, Fix #7133, Fix #7136: "Codechange: Tweak ViewportAddLandscape so it no more relies on "go down as fast as possible" tile height model (Patch by adf88, #6583)" https://git.io/fhS6b
20:42:30 <andythenorth> so for gen 1920s trains, the EMUs need a massive bonus, not a malus :P
20:44:17 <nielsm> planetmaker: for just one new/fixed town cargo gen algorithm, whose main feature is that the amount of cargo generated is linear on house population, what names would you use in Settings?
20:48:00 <michi_cc> andythenorth: IMHO refrigerated and express should get a malus if transported in generic wagons, to increase the spread, the specialised wagons could get a bonus at the same time.
20:48:16 <andythenorth> what could go wrong :)
20:48:25 <andythenorth> apart from me forgetting to add the right cargos :)
20:48:25 <peter1138> nielsm, broken / improved ;p
20:48:41 <nielsm> How much cargo houses in towns produce, relative to their size. The original method has quadratic growth, so a town with 2x the population produces 4x as many passengers. The new method keep the relationship linear, so double size causes double passenger production.
20:48:48 <nielsm> Quadratic (original) / Linear
20:49:04 <michi_cc> No idea if you can test cargo classes in a varact2, but if yes that would be ideal I guess.
20:49:06 <peter1138> Too much / About right
20:49:44 <peter1138> So, er, if I cache the ship's previous position, I can get rid of the glitch.
20:50:15 <andythenorth> maybe I should set up a 4096x4096 map
20:50:22 <andythenorth> and build a diagonal train line across it
20:50:32 <planetmaker> hm, let's see, niels
20:51:43 <peter1138> And... I think it doesn't need to be saved, because if you are just loading a save, you won't see the previous location.
20:52:23 <planetmaker> got the PR number for me?
20:52:41 <nielsm> planetmaker: haven't pushed latest changed yet
20:53:09 * andythenorth requests an additional cargo bonus mechanic :P
20:55:36 <planetmaker> ok. Town cargo generation scaled by population
20:55:42 <planetmaker> but that's lengthy :)
20:56:22 <nielsm> planetmaker: isn't the point of the help text to allow lengthy explanations?
20:56:50 <planetmaker> yes. That's why I was asking for the PR number to actually see the strings
20:57:45 <planetmaker> or can you paste the strings?
21:00:59 <planetmaker> should we reference it really as 'old' and 'new' (even when it is)?
21:02:39 <nielsm> well, it's sufficiently original that it exists in the game chris sawyer had published 25 years ago
21:03:04 <nielsm> so I think it makes sense to point out which option to choose if you want to be closest to that
21:05:15 * andythenorth wonders about just scaling cargo prop 12 (pricefactor), prefereably in low % increments / decrements
21:05:34 <andythenorth> I think cargo_age_period addresses different concerns to mine
21:05:48 <nielsm> and btw 5000 tile ticks is about 47 years
21:06:31 <planetmaker> it's a bit bike shedding: May I suggest that 'linear' is the "upper" value in the UI? From a sorting it feels above quadratic
21:07:27 <planetmaker> the {} adds line breaks
21:09:45 <nielsm> hm is there any good way to swap their position in the gui list without changing their numeric values? >_>
21:10:32 <planetmaker> I don't think so.
21:11:32 <planetmaker> Well, that's more bike shedding. If they are not swapped, swap the order of explanations in my version of the help text compared (like you had it before)
21:13:57 <planetmaker> I'm unsure to use twice and four times over 2x and 4x. But if using 2x and 4x, it should also be used 2x and 2x in the other string. Usually writing it is the way considered to look nicer
21:15:29 <nielsm> I kind of want to add colour to the option names in the help text
21:16:15 <nielsm> but nothing else does that
21:21:47 <planetmaker> what colour would make sense? Light blue as for 'Settings type' and 'Default value'?
21:21:57 <nielsm> yeah that's my thought
21:22:29 <planetmaker> I see why and how that would make sense and the string better readable
21:26:30 <peter1138> How do you handle making a default setting be different for old savegames?
21:27:38 <frosch123> we have some code in afterload for that
21:27:45 <frosch123> dynamic_engines is an example
21:56:20 *** Thedarkb-T60 has joined #openttd
22:01:18 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7174: Fix #7199: When rotating a ship, apply an additional offset to avoid movement glitch. https://git.io/fhSP0
22:02:22 *** Thedarkb-T60 has joined #openttd
22:02:59 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh updated pull request #6965: Add: Four alternative town cargo generation methods https://git.io/fpkqa
22:03:22 <Samu> Text layout in engine preview dialogue windows was
22:04:42 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 commented on pull request #7174: Fix #7199: When rotating a ship, apply an additional offset to avoid movement glitch. https://git.io/fhSPz
22:05:22 <peter1138> nielsm, is it still addin g4? Maybe it wants a title adjustment?
22:06:16 <glx> Samu: click on the + button and tell it there :)
22:06:30 <peter1138> Where did I get that from?
22:07:28 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #7174: Fix #7199: When rotating a ship, apply an additional offset to avoid movement glitch. https://git.io/fhSP0
22:09:04 <peter1138> I'm aware it's some kinda of cheaty hack :/
22:15:32 *** drac_boy has joined #openttd
22:23:06 <drac_boy> hi samu ... and hmm I dunno, never had any shipping problems I guess sorry
22:24:31 <drac_boy> samu just curious but which ships do you like using? (grfwise)
22:25:09 <peter1138> Samu's an AI watching kinda guy
22:27:00 <drac_boy> samu so vanilla .. mm fair enough :)
22:27:00 <planetmaker> nielsm, town_cmd.cpp: is that the correct random call which takes care of not desyncing?
22:27:33 * drac_boy has always used newship.grf anyway (with newshipw.grf copy on hand for some other computers at times too)
22:28:30 <nielsm> planetmaker: it's the same Random used earlier in the function
22:29:36 <planetmaker> ah, yes. Thus should be safe
22:31:35 <drac_boy> anyway have to go sort out the supper food..but I'll try remember to check back at least one more time tho :)
22:34:35 <Samu> I wonder what have you done to fast forward, it's better, even with 15 AIs
22:36:53 <planetmaker> nielsm, looks fine to me. The title / commit message still needs to be changed though to reflect the actual change. Luckily it will be squashed
22:42:27 <Samu> peter1138: are you still interested in improving fps performance? I got a really mean savegame here
22:42:56 <Samu> 4 fps on my system with some stalls of 10 seconds from time to time
22:46:29 <peter1138> I can't do much with a screenshot.
22:47:53 <peter1138> No, just link a file :p
22:48:35 <nielsm> where did you get it from?
22:49:30 <Samu> actually, i think i got it posted there
22:49:43 *** ChanServ sets mode: +v tokai
22:51:23 <Samu> nop, im making a new topic
22:55:51 <Samu> just had a 21 second stall
22:56:18 <andythenorth> meh, costs, aging, blah blah :D
22:56:33 <peter1138> Procrastination, eh?
22:57:14 <andythenorth> well more like, change, wait for compile, reload, run ffwd, decide it's all hopeless, repeat :D
22:57:25 * andythenorth listening to some KLF thing on radio 4
22:57:26 <peter1138> Spend weeks fixing the build time.
22:57:43 <andythenorth> rebuild the grf to make it fast to compile
22:57:48 <peter1138> Hmm, takes ages to load.
22:58:02 <andythenorth> I've rebuilt grfs a few times to optimise compile :P
22:58:05 <andythenorth> seems mad, but eh
22:58:43 <planetmaker> not worse than spending sunday nights on optimizing a 20-year-old computer game :P
23:01:26 *** Lejving has joined #openttd
23:02:23 <Samu> btw, in that savegames there is no more pool space for orders
23:02:33 <Samu> AI's can't insert more orders :|
23:03:59 <Samu> I reported it, but it was closed
23:04:13 <peter1138> Sheer number of vehicles.
23:05:46 <peter1138> Hmm, does 'Find missing content online' not work for AIs?
23:09:25 <Samu> there should be a way to see which AI is causing the most slowdowns
23:17:23 <DorpsGek_II> [OpenTTD/OpenTTD] George-VB commented on issue #6907: Cargo capacity should be recalculated on TRIGGER_VEHICLE_NEW_LOAD https://git.io/fhSXw
23:33:47 <peter1138> andythenorth, tell me what colours to remap and we'll do 3cc? :p
23:34:14 <peter1138> Hmm, game hung on exit :/
23:35:42 <peter1138> Hurm, appears to be SDL waiting for PulseAudio. Huh./
23:36:11 <peter1138> Oh shit, you actually did :/
23:36:14 <andythenorth> although I use them both in sprite generation, so I'd have to change ~everything :P
23:37:29 <peter1138> I was gonna say, what about the water colours, but you might use them on a ship.
23:37:59 <peter1138> Although, all that purple, is that the windows palette instead of dos?
23:38:37 <andythenorth> can't choose blue or green, too hard to draw alongside 1CC and 2CC
23:39:08 <andythenorth> 40-47 is also good contrast, and I also use those for sprite generation :P
23:39:21 <andythenorth> 215 are the magic weird pinks
23:39:35 <peter1138> Yeah, but if they're remapped to another range then they are no longer pink
23:39:42 <andythenorth> there is that yes
23:39:50 <peter1138> You'd need to faff with palettes to edit I suppose.
23:39:56 <andythenorth> yes, but that's possible
23:40:32 <Samu> seems to be some valuator
23:41:06 * andythenorth wonders what the pinks were ever for
23:41:06 <peter1138> nielsm, was it not possible to show performance data for AIs?
23:41:35 <nielsm> peter1138, uh maybe...
23:41:49 <nielsm> it'd need some rejiggering, maybe
23:42:15 <peter1138> Hmm, right, using the purples is awkward cos they move depending on palette :/
23:42:43 <Samu> xtree is causing the huge stalls
23:43:08 <Samu> but that's a visual studio file, not openttd
23:43:32 <Samu> doing some delete and free
23:48:05 <peter1138> Garry finds another bug in more cargos. This one looks ... tricky.
23:48:42 <andythenorth> what about AA-B1 o_O
23:48:52 <andythenorth> actually cargos more important :P
23:49:50 <peter1138> I don't think 256 companies is a thing.
23:56:37 <peter1138> Hmm, CargoMonitorID is saved :/
23:56:44 <peter1138> So this is going to need some munging.
continue to next day ⏵
