IRC logs for #openttd on OFTC at 2019-09-30
⏴ go to previous day
01:24:01 <Eddi|zuHause> i seem to have developed this habit of building stations with too few platforms...
01:25:50 <glx> and the town built around the station when you notice it
01:55:27 *** Thedarkb-X40 has joined #openttd
01:58:56 *** arron_sh[m] has joined #openttd
02:03:36 *** arron_sh[m] is now known as arron[m]
03:46:09 *** Wormnest has joined #openttd
04:40:54 *** reldred has joined #openttd
05:01:39 *** tokai|noir has joined #openttd
05:01:39 *** ChanServ sets mode: +v tokai|noir
09:35:53 *** Smedles has joined #openttd
09:53:54 *** WormnestAndroid has quit IRC
10:06:45 *** WormnestAndroid has joined #openttd
10:17:59 *** blackli0nxx_ has joined #openttd
10:21:12 *** andythenorth has joined #openttd
10:21:50 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on pull request #7752: Prevent sounds being produced by inactive industries https://git.io/JentZ
10:22:19 <andythenorth> trying not to write that without being a dick
10:22:33 * andythenorth should go to work
10:31:20 *** blackli0nxx_ has joined #openttd
10:45:29 <andythenorth> oof I had a good idea, but it FAIL
10:51:31 *** WormnestAndroid has quit IRC
10:54:30 *** andythenorth has left #openttd
11:13:09 *** WormnestAndroid has joined #openttd
11:55:17 <DorpsGek_III> [OpenTTD/OpenTTD] abmyii commented on pull request #7752: Prevent sounds being produced by inactive industries https://git.io/JenmJ
14:29:26 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on pull request #7752: Prevent sounds being produced by inactive industries https://git.io/JenO5
14:37:14 <_dp_> btw, why is sound using game state random?
14:43:00 <_dp_> It's kinda fair that everyone will be equally annoyed by those sounds ofc but then you can just disable them altogether :p
14:43:57 <nielsm> yes that is kinda weird
14:45:57 <LordAro> if in doubt, probably hysterical raisins
14:48:34 <nielsm> now checking the end year thing for myself
14:48:51 <nielsm> is it really 2049-2050 year change in original ttd
14:49:04 <nielsm> or 2050-2051 as implemented in ottd
14:49:46 <nielsm> having to run ttd in dosbox and then run dosbox in fastforward mode
14:50:34 <nielsm> yeah 2049-2050 year change is the correct
14:50:43 <nielsm> for how long has ottd been wrong and had a 101 year game?
14:51:41 <nielsm> it should also, at least in singleplayer, switch music to the title music
14:55:46 <peter1138> The sounds are synchronised, see.
14:56:22 <LordAro> nielsm: surprising, given all those sprites for the endscreen would've been hardcoded
14:59:06 <nielsm> the sprite says december 31, 2050
14:59:15 <nielsm> the game ends at december 31, 2049
14:59:31 <LordAro> obiwan in TTD itself?
15:00:27 <nielsm> otoh this is tested with ttdpatch for the Year cheat
15:00:45 <LordAro> closing 3 PRs at once is always nice
15:00:53 <nielsm> so should maybe see if I can generate a save in dec 2049 and load it without patch
15:01:12 <peter1138> Grr, USB port on my Garmin is fucked :(
15:01:22 <peter1138> The GND pin is bent back.
15:04:34 *** andythenorth has joined #openttd
15:10:37 * andythenorth wondering if town ID can be used as a hack to force industry co-location
15:10:41 <andythenorth> in absence of a regions function
15:11:25 <andythenorth> I can check distance to other industry types, when building the industry, but it's at risk of deadlocks, and gets hard to maintain
15:12:55 * andythenorth draws ideal world, where industries return some UID token, and openttd only builds industries with that UID in certain regions of the map
15:15:09 <Eddi|zuHause> LordAro: i don't see where the obiwan would come from there, but why was it checking against "ending_year - MAX_YEAR" before, shouldn't that have been BASE_YEAR?
15:17:57 <LordAro> honestly i've no idea
15:19:08 <LordAro> it's possible it's been wrong the entire time
15:19:35 <DorpsGek_III> [OpenTTD/OpenTTD] abmyii commented on pull request #7752: Prevent sounds being produced by inactive industries https://git.io/Jen31
15:22:37 <Eddi|zuHause> tbh, i've never reached the year 2050
15:25:53 <DorpsGek_III> [OpenTTD/OpenTTD] abmyii commented on pull request #7752: Prevent sounds being produced by inactive industries https://git.io/JensU
15:37:14 <andythenorth> can towns be iterated arbitrarily from a newgrf :P
15:37:22 <andythenorth> towns have permanent storage
15:38:09 <andythenorth> so at game start, if I could walk say 25% of towns and set a storage bit, for example 'chemicals industry town'
15:38:18 <andythenorth> then industry location check could probably read that storage
15:38:36 <andythenorth> it could even be done when the first n industries of that type are build
15:39:28 <andythenorth> it would limit to placement within the town catchment though, rather than within n tiles
15:39:34 <andythenorth> might be very crowded :P
15:40:40 <peter1138> Does OpenTTD have RTX support yet?
15:41:12 <andythenorth> gamescript is blind to newgrf
15:41:22 <andythenorth> except by reading the produced / accepted cargos
15:41:32 <andythenorth> also ew gamescript :)
15:53:56 <nielsm> andythenorth or you could do some kind of check on map coordinates, maybe that works best for mining industries
15:54:04 <nielsm> (areas of the map rich in various natural resources)
15:54:16 <andythenorth> yes, could divide the map up into blocks
15:54:34 <andythenorth> that's already used for clustering similar industries somewhat, but not using x,y coords
15:55:18 <andythenorth> so, for example, chemicals industries might always locate in top corner of map
15:55:32 <andythenorth> in a 128*128 block
15:56:23 <andythenorth> interesting idea
15:57:24 <nielsm> you can maybe also emulate less rectangular areas by making diamond shaped areas instead, or the sum of a square and a diamon for a kind of 8 pointed star
15:57:41 <andythenorth> need some clever algorithm for checking x,y?
15:57:52 <andythenorth> like a parametric function or something?
15:58:01 <andythenorth> I am not very good with that maths
15:58:21 <andythenorth> but in schoold I worked out how to use parametrics to draw a smiley face on a graphical calculator
15:58:34 <andythenorth> unfortunately that was not an exam question :P
15:59:01 <nielsm> decide on a center point of the industrial zone x0,y0
15:59:06 <andythenorth> I think there probably is a case for regions in OpenTTD
15:59:15 <andythenorth> but we'll never discover it without trying some hacks first
15:59:16 <nielsm> take the candidate position of the industry x,y
15:59:28 <nielsm> subtract from the center and take the absolute value of each coordinate
15:59:43 <nielsm> dx,dy = abs(x0-x),abs(y0-y)
15:59:54 <nielsm> then check the range of dx and dy
16:00:03 <andythenorth> "decide on a center point of the industrial zone x0,y0" requires fixed rules that can run in a callback
16:00:20 <andythenorth> and fallbacks, in case, e.g. it's all sea
16:00:33 <nielsm> yeah it wouldn't work well in the general case
16:00:50 <andythenorth> running this cb every time an industry is built would not be ideal
16:01:03 <nielsm> otoh, if all the coal on the map is located under the sea... yeah that's a problem you can solve by filling in the sea :D
16:01:35 <nielsm> yep it's better suited for a GS situation, where the GS picks the location of the industry
16:01:52 <nielsm> instead of the game trying a bunch of random locations until one works
16:02:53 <andythenorth> the GS route remains puzzling
16:03:00 <andythenorth> GS isn't event driven afaik
16:03:18 <nielsm> it would be the other way around
16:03:27 <andythenorth> makes player funding hard
16:03:34 <andythenorth> and game start would be hard / slow
16:03:43 <nielsm> the GS just decides it wants to build an industry, then uses its knowledge of the map and any rules decided by itself to pick out a position
16:03:52 * andythenorth ignoring the other problems with GS
16:04:13 <andythenorth> GS is a problem looking for a solution in most newgrf cases
16:04:34 <nielsm> I'm out, going to visit copenhagen and check the newly opened metro line
16:04:46 <andythenorth> taking the parrot?
16:05:06 <nielsm> they are too difficult to take on trips :P
16:05:14 <andythenorth> my uncle has one, it has a lead
16:05:18 <andythenorth> but sometimes it flies away
16:05:23 * andythenorth back to GDPR and EU data transfers after Oct 31st :P
16:05:52 <DorpsGek_III> [OpenTTD/OpenTTD] planetmaker commented on pull request #7747: Feature: Configurable ending year https://git.io/JenGf
16:06:33 <planetmaker> oh, that sounds like an aweful lot of fun @ andythenorth :|
16:13:18 <andythenorth> it was but I am not happy with what it wsa
16:13:22 <andythenorth> marmite on toast
16:13:29 <andythenorth> not enough salad
16:17:17 <planetmaker> nielsm, in order to not spam the github issue: what is the idea of adding both, ending_year and game_length? It seems homologuous in terms of what it achieves, thus duplicating a setting to two variables
16:18:32 <planetmaker> (it's ok to have the choice to set either ingame for start config, but it only needs one internal variable)
16:29:22 <_dp_> kinda agree, it's hard to imagine usecases for those two settings
16:29:42 <_dp_> removing scoreboard completely is nice, that I can use
16:30:49 <_dp_> but no ideas for any other case
16:32:17 <_dp_> citymania has game length setting but it does a lot of extra stuff on game end, not sure how that's relevant to vanilla
16:33:18 <peter1138> andythenorth, oh, mine was salad and... er... cake :/
16:34:40 <peter1138> Fruit cake, leftovers from the MacMillan cake day thing.
17:08:33 *** Wormnest has joined #openttd
18:15:03 *** sla_ro|master has joined #openttd
18:21:16 <Wormnest> Samu: The main difference with you in what I usually do for testing my ai´s is that I set vehicle limits a lot lower than the maximum 5000
18:22:03 <Wormnest> Besides that in the tested game it took a long time before WormAI was first, it was 3/4 for a long time
18:24:50 *** Progman has joined #openttd
18:46:12 *** arikover has joined #openttd
18:57:04 *** gelignite has joined #openttd
19:23:19 *** HerzogDeXtEr has joined #openttd
19:29:52 *** andythenorth has joined #openttd
19:30:19 *** andythenorth has left #openttd
19:30:23 *** andythenorth has joined #openttd
19:38:32 *** supermop_pdx has joined #openttd
19:40:48 <andythenorth> planetmaker: any interest in a devzone refresh? :)
19:45:48 <DorpsGek_III> - Update: Translations from eints (by translators)
19:47:09 <planetmaker> andythenorth, in what way are you talking?
19:47:35 <andythenorth> frosch has a repo with some dockerisation of devzone
19:49:10 <planetmaker> interesting. That does look nice
19:49:44 <andythenorth> it would be nice to know the future of devzone
19:51:51 <andythenorth> most of my grfs will no longer build
19:52:23 <andythenorth> I started moving them to Github + Azure + Amazon, but it's problematic for eints I believe
19:53:06 <andythenorth> I don't like to bother anyone too often about devzone, and not sure who to bother either :)
19:53:32 <planetmaker> well, I started moving repos to github. But the build part... is quite unique. Anyhow, I don't plan to retire the machine ...
19:53:48 <planetmaker> the upgrade part is the ... "interesting" part. I've no spare to test on :)
19:54:00 <andythenorth> do we own the hardware?
19:54:08 <planetmaker> no, I pay monthly 45€ for it
19:54:31 <planetmaker> I planned to use my old machine... but I fried it
19:54:43 <andythenorth> get a 2nd rental, move A->B ?
19:55:23 <planetmaker> well, yesish... that would be my choice
19:56:24 <planetmaker> I was hoping actually s/o had a machine to use transitionally :)
19:57:14 <planetmaker> i.e. move current devzone onto it, and setup everything anew on the existing machine
19:57:31 <planetmaker> and migrating stuff to it as it becomes available
19:57:35 <andythenorth> rent a commodity VM?
19:59:18 <andythenorth> TrueBrain knows about servers? :)
19:59:26 <planetmaker> hm, yeah. A cheap monthy VM contract, moving everything essential there... then an upgrade of the HV is easy
19:59:27 <andythenorth> or is Spike still around in coop?
19:59:54 <planetmaker> well... I can contact both... but both are about as available as you know it :)
20:00:13 <planetmaker> spike just took up a new job...so dunno :)
20:00:49 <planetmaker> but I guess some help from that side would be available
20:00:54 <planetmaker> with some planning
20:08:38 *** Etua is now known as Guest3775
20:28:02 *** firewire1394 has joined #openttd
21:14:30 <TrueBrain> Get a droplet on Digital Ocean .. pay as long as you use it
21:14:43 <TrueBrain> Pay-as-you-go is awesome
21:15:06 <TrueBrain> (@planetmaker, @andythenorth)
21:15:57 <planetmaker> hm, let's see. thx
21:16:08 <andythenorth> I have FIRS publishing from Azure to AWS
21:16:16 <andythenorth> but hmmm, it was too much clicking
21:16:47 * andythenorth doesn't like clickety click
21:16:51 <planetmaker> well, hm... memory... is expensive there
21:16:59 <planetmaker> game servers use loads of memory :)
21:17:04 <glx> I'm sure you can publish without clicking
21:18:03 <TrueBrain> Memory optimized ones too?
21:18:18 <TrueBrain> Andythenorth: GitHub actions make your life easier
21:18:43 <glx> yeah with github actions you can compile from github
21:18:44 <TrueBrain> No clicks .. lot of text :p
21:19:07 <andythenorth> there are these crazy AWS configs to paste around
21:19:08 <TrueBrain> Just ... you need to sign up for the beta for I believe two more months
21:19:13 <andythenorth> and a config generator
21:19:27 <andythenorth> the lolz is that the generator makes invalid syntax due to version mismatch :)
21:19:41 <planetmaker> github is promoting these actions as much as they can do... for all I always see
21:23:00 <TrueBrain> can't blame them; it is something to be proud of
21:23:17 <andythenorth> can we abandon coop jenkins? :)
21:25:09 <glx> the only issue I see is it's hard to test changes in PRs
21:26:00 <TrueBrain> you need to test it in a fork, I nocited :D
21:27:23 <planetmaker> do you want to abandon it, andythenorth ?
21:28:27 <planetmaker> I'm not quite sure whether the actions work for all we need to build newgrfs... maybe...
21:29:43 <TrueBrain> actions == docker == linux machine
21:29:52 <TrueBrain> so you -could- do everything .. just the question is if you want to :D
21:29:57 <TrueBrain> which is a fair thing to consider first ;)
21:30:08 <TrueBrain> burning everything first, thinking about that later, is the andythenorth approach ;)
21:30:13 <glx> it's azure pipeline integrated in github
21:32:55 <andythenorth> I just distrust jenkins :)
21:33:59 <andythenorth> actually what I liked about Azure Pipelines (not much) was that it's cattle not pets
21:34:09 *** WormnestAndroid has quit IRC
21:34:12 <andythenorth> jenkins historically tends towards hand-curated worker nodes
21:34:22 <andythenorth> or hand-managed images for worker nodes
21:34:34 <andythenorth> AP I just list python deps in a text file, and that's all
21:34:38 *** WormnestAndroid has joined #openttd
21:35:04 <andythenorth> no maintenance, no deps, no package manager, no conflicts, no CVEs,
21:35:19 <planetmaker> right. So I will setup a machine which can run docker containers. So then from github they can be triggered to build NewGRFs
21:35:24 <planetmaker> is that understanding correct?
21:35:32 <andythenorth> sounds pretty good
21:35:41 <TrueBrain> GitHub Actions run docker images
21:35:55 <TrueBrain> so you don't need to setup any machine :P
21:36:14 <andythenorth> I should give you access to my AP and AWS account :P
21:36:19 <planetmaker> so devzone is obsolete
21:36:25 <andythenorth> if we want it to be
21:36:34 <andythenorth> it's a choice :)
21:36:42 <TrueBrain> its a lot of work ;)
21:36:45 <andythenorth> bundles is *not* obsolete, there's no commodity solution for that
21:36:47 <planetmaker> iff github actions run docker images on github infrastructure
21:36:50 <TrueBrain> you need to make a docker image that can build NewGRFs :D
21:37:08 <TrueBrain> but yeah, GitHub runs those images free of charge for Open Source projects
21:37:26 <TrueBrain> basically, they give you access to a Windows / Linux / MacOS machine
21:37:48 <andythenorth> if the goal is repo -> published artefacts on bundles, then I had it working for FIRS, but there are 5 significant problems
21:37:56 <andythenorth> 1. devzone repos don't import to github
21:38:05 <andythenorth> 2. Azure Pipelines is quite horrible
21:38:11 <andythenorth> 3. AWS is quite horrible
21:38:18 <andythenorth> 4. no way for eints to add translations
21:38:25 <planetmaker> well, bundles is there. And can remain there. Same for eints
21:38:27 <andythenorth> 5. have to re-write bundles for AWS
21:38:41 <planetmaker> And... I have a new machine now where we can make everything tidy and retire the old
21:39:09 <planetmaker> (yes, the machine is 15 minutes old :P )
21:39:14 <andythenorth> frosch thinks that we could just scp or sftp to bundles, but that means keeping auth creds in Azure
21:39:20 <andythenorth> and probably it's not so simple anyway
21:39:36 <andythenorth> (not doubting frosch BTW, just don't trust tech)
21:39:38 <planetmaker> how else would one get stuff to that machine?
21:40:05 <andythenorth> how does Jenkins do it?
21:40:26 <andythenorth> doesn't it write to disk on same machine?
21:40:30 * andythenorth may be 100% wrong
21:42:08 <planetmaker> it actually has the server nfs-mounted
21:42:55 <planetmaker> but that does not make really any difference what to do with an artefact after it is build
21:43:05 <planetmaker> whether you issue a cp or an scp on it
21:43:07 <andythenorth> who does the magic with the version numbers? bundles or jenkins?
21:43:26 <andythenorth> e.g. release/LATEST etc
21:43:32 <planetmaker> neither really. The build script executed by jenkins
21:43:43 <andythenorth> that's the magic that I didn't want to replicate for AWS / Azure
21:43:44 <planetmaker> or the post_build script.
21:44:13 <andythenorth> so if we dropped artefacts on a dir somewhere, a bundles script could sort it out?
21:44:16 <LordAro> nothing particularly about storing (suitably locked down) credentials on azure/aws/actions/whatever
21:44:39 <andythenorth> it doesn't worry me, I just couldn't find which button to click
21:45:03 <andythenorth> I was able to give Azure an authed github connection with a revokable key
21:45:16 <andythenorth> and some dedicated AWS creds for push to an s3 bucket
21:45:25 <andythenorth> but generic creds I didn't find :)
21:45:57 <planetmaker> I would require login via ssh key... but that's easy... sorting where to put stuff... well
21:47:02 <planetmaker> currently it is easy as the post build script knows the project name and the version. So it writes to the dir identical to the project name, creates a dir identical to the version. And updates the symlink of LATEST to whatever it just created (if it built the 'latest')
21:47:29 * andythenorth tries to find the Azure FIRS project
21:47:41 <planetmaker> so bundles trusts the build node to store stuff in the right place and to do all the work with naming it correctly, too
21:47:44 <andythenorth> azure is like entering neverland :P
21:47:59 <planetmaker> bundles itself is absolutely dumb. It does nothing other than serving files.
21:48:12 <planetmaker> And that as plain html without overhead
21:49:21 <andythenorth> ok so the script is the clever part
21:50:51 <planetmaker> but it's the same script for every newgrf
21:51:03 <planetmaker> or rather even: every project
21:52:31 <andythenorth> maybe that's portable to other places
21:52:40 <planetmaker> with small adaptions: yes
21:52:47 * andythenorth lost in Azure and AWS, the UI is...ouch
21:57:14 <andythenorth> in some ways it was magical and easy
21:57:23 <andythenorth> but I spent a very long time in bad outdated documentation
21:57:35 <andythenorth> trying to set the bucket name, access policy, CORS policy etc
22:00:49 <planetmaker> so... what's wrong with bundles? :D
22:01:34 <andythenorth> nothing, I just don't have scp / sftp / ssh
22:02:01 <andythenorth> and in May this year, there was nobody around to look at this, so I tried alternatives :)
22:05:19 <planetmaker> the user management of the bundles server seems to be prime... most files belonging to users the system doesn't know :P
22:07:36 <Samu> it' still yesterday's game
22:08:23 <Samu> WormAI is orange, RailwAI is green, NoNoCAB is red, LuDiAI AfterFix is blue
22:08:52 <Samu> fast forwards too slow over here
22:09:29 <Samu> the limits are 500 500 200 300
22:12:00 <Samu> buoys may trigger the num of station ids limit
22:14:25 <Samu> wish there was an easy way to see how many station ids are in use
22:26:46 *** blackli0nxx_ has joined #openttd
22:27:40 *** blackli0nxx__ has joined #openttd
22:37:35 <Samu> wow, ships can walk over UFOs
22:37:41 <Samu> and nothing happens to them
22:47:46 *** blackli0nxx has joined #openttd
22:55:12 *** blackli0nxx has joined #openttd
22:59:13 *** blackli0nxx_ has joined #openttd
23:04:34 *** WormnestAndroid has quit IRC
23:04:50 *** WormnestAndroid has joined #openttd
23:12:37 <luaduck> random feature wish for a dev to take up: how difficult would it be to be able to append a reason string to a CLIENT_KICK (so we can give clients clear and visible reasons as to why they were kicked)?
23:13:09 <luaduck> right now the only way to give a client context as to their kick reason is to whisper to them immediately before kicking them, and hope they check their console
23:13:24 <luaduck> if there's interest I'll raise a github issue for it
23:13:39 <Eddi|zuHause> my guess is anywhere between "here, it's a trivial one-liner" to "holy hell, this requires a complete rewrite"
23:14:43 <luaduck> (and yes, I am too lazy to dig the source)
23:15:14 *** andythenorth has joined #openttd
23:15:29 <luaduck> adding it in the client would probably be trivial, but network is more interesting, though other packets like player name changes allow strings as packet data
23:16:09 <luaduck> unfortunately this isn't really something I'd feel comfortable contributing because C
23:18:08 *** andythenorth has left #openttd
23:20:34 <LordAro> luaduck: none of us were born knowing how to use it :)
23:21:09 <luaduck> fair point, but I don't need _another_ language to be actively learning, golang is my bff
23:21:24 <Eddi|zuHause> what happened to lua?
23:24:05 <luaduck> lua was a terrible idea
23:24:09 <luaduck> friends don't let friends write lua
23:24:29 <luaduck> (my new nick is actually duck. but I retain this one on freenode for reasons)
23:24:57 <Eddi|zuHause> i modified a line of lua once, i found it an overall unpleasant experience
23:25:16 <Eddi|zuHause> everything about it felt somewhat off
23:33:03 <luaduck> I wrote lua once and now I'm a terrible person
23:42:12 <LordAro> so i looked at the network code a bit
23:42:27 <LordAro> it's... unclear how difficult adding a message would be
23:42:34 <DorpsGek_III> [OpenTTD/OpenTTD] luaduck opened issue #7756: Allow server to supply a reason to kicked / banned clients https://git.io/Jen01
23:42:42 <Eddi|zuHause> ... so that's basically what i said? :p
23:43:41 <LordAro> it's possible to send a message to clients, and it's possible to kick a client
23:43:47 <LordAro> i can't tell if you can feasible combine those
23:44:04 <luaduck> LordAro: could the packet not be expanded with extra data?
23:47:29 <LordAro> luaduck: that does seem feasible, yes
23:47:48 <LordAro> i think the trickiest bit would be actually displaying the message on the other end
23:48:12 *** Wormnest has joined #openttd
23:48:44 <_dp_> sending kick msg should be relatively easy
23:49:25 <_dp_> though it never bothered me enough to make a pr
23:49:47 <_dp_> wheneven I kick people they usually already know why xD
23:49:58 <LordAro> i dare say you're probably one of the most knowledgable people regarding that area of the code :p
23:50:03 <_dp_> and never cared if the banned ones do xD
23:50:31 <luaduck> reason I'm requesting it is that on our reddit servers, we have an automated IP check that checks if the client is behind a proxy
23:50:58 <_dp_> luaduck, yeah, i noticed ;)
23:51:04 <luaduck> which kicks pretty much ASAP
23:51:28 <luaduck> and also a thingy which kicks players which don't change their name from the default if they try and create a company 3 times in a row
23:54:03 *** supermop_pdx has joined #openttd
23:54:35 *** blackli0nxx has joined #openttd
23:55:48 *** blackli0nxx has joined #openttd
continue to next day ⏵