IRC logs for #openttd on OFTC at 2018-12-27
⏴ go to previous day
00:00:26 *** Thedarkb-T60 has joined #openttd
00:05:11 <Samu> something that Albert teached me
00:06:06 <glx> you could just do local myCID = AICompany.ResolveCompanyID(AICompany.COMPANY_SELF);
00:06:35 <glx> and use myCID when needed
00:15:28 <Samu> I wonder how other AIs handle an AI that founds towns
00:16:05 <Samu> do they notice the new towns?
00:30:23 <Samu> if (perform_action && TestPerformTownAction.TryPerform(town, action)) {
00:31:04 <Samu> if (perform_action && TestPerformTownAction().TryPerform(town, action)) {
00:31:48 <Samu> the amount of redundancy in the names
00:42:09 <Samu> i start right away to decide not building HQ
00:42:43 <Samu> NoCarGoal delay is much longer with 14 AIs
00:43:07 <Samu> CompanyValueGS seems to be fast enough, about 3 days
00:47:22 *** Progman has joined #openttd
00:49:01 <Zuu> NoCarGoal has to register cargo delivery counters for all sources or destinations. If the map is not too large, it will be able to do so within mapgen, but if your map is too huge, then the mapgen ticks are not eounght.
00:52:20 <Samu> let me test in a 4k map brb
00:54:51 <Zuu> But I don't see how it is a problem if you get to know that it is a NoCarGoal game after building your first connection or having built the HQ. If that is a problem, your AI is not robust enough.
00:56:52 <Samu> the monthly report signs seem to pile up
00:57:16 <Zuu> Then don't ask for them. :-)
00:57:49 <Zuu> Or turn them off if they are on by default, but you don't read them fast enough.
01:00:18 <Samu> my ai has a very very long loop
01:00:34 <Samu> especially with the increased number of routes, vehicles, etc
01:00:49 <Zuu> You can always toss in calls for scp.Check() a few places in your loop if it is very very long.
01:00:49 <Samu> i get some loops lasting 5 years :(
01:01:17 <Zuu> But if it lasts 5 years, then yes that is very very long.
01:01:49 <Samu> it's the road vehicle management, i can't make it faster :(
01:01:58 <Samu> the aircraft management is adequate
01:02:43 <Samu> until it starts searching for a spot for intercontinentals...
01:03:28 <Zuu> If it cannot be faster, can you run it less often on each route and only do a few routes each main loop cycle? Or monitor a few key stats for all and using that select which ones to target with the slower management?
01:05:05 <Samu> i tried something that let it perform road management up to 2220 ticks
01:05:11 <Zuu> And look for marker signs that are cheap to check that could indicate a broken route that requires runnig the path finder etc. So you don't run it blindly on all routes to check that they aren't borken.
01:05:56 <Zuu> Compare profit over time for example.
01:06:28 <Samu> oh, i dont run the pathfinder while inside the road route management
01:07:01 <Zuu> CluelessPlus sometimes does run it in repair mode.
01:08:20 <Zuu> If not else, run it when you receive vehicle crashed event
01:08:48 <Samu> hmm, let me get my main cycle out there
01:08:55 <Zuu> Chances are that someone ran a track through your road and started to crash your vehicles. :-)
01:08:56 *** Wormnest_ has joined #openttd
01:09:32 <Zuu> And then you better see if path finder can swap that out for a bridge.
01:09:35 <Samu> my loop starts with Utils.RepayLoan();
01:09:41 <Samu> should be almost instant
01:10:10 <Samu> then it starts the scp check loop, which is supposedly fast enough
01:10:20 <Zuu> I use timers in CluelessPlus source that will register time spent on different parts and log that to console.
01:10:58 <Samu> RemoveLeftovers() is next, which is pretty much just checking if there's something that was left behind that needs to be cleared, so it's also instant
01:11:17 <Samu> dead routes, stations pretty much
01:11:48 <Samu> updateVehicles(); this is the road management, the more routes, the slower it gets :(
01:12:49 <Samu> BuildRoadRoute(cityFrom, buildManager.hasUnfinishedRoute() ? true : false); comes next, which runs the pathfinder, split into several attempts, each attempt is about 500-700 ticks
01:13:06 <Samu> wrightAI.ManageAirRoutes();
01:13:37 <Samu> is selling, adding, removing airports, all at once, it's almost always fast enough, unless it's dealing with too many vehicles in depots
01:14:01 <Samu> PerformTownActions(); FoundTown(); BuildHQ(); should all be very fast
01:14:07 <Samu> and that's the end of the loop
01:14:13 *** rocky1138 has joined #openttd
01:14:23 <Samu> oh, wait, there's the build air route after the manage air route, my bad
01:14:38 <Samu> usually fast, unless dealing with intercontinentals
01:16:00 <Zuu> It is long, but the thing I wanted to point out is TimerStart("name"); .... TimerStop("name");
01:16:39 <Zuu> And every 5 year it prints out the timer counts at the end.
01:17:04 <Zuu> This way I get to know which parts of my main loop to focus on to speed up.
01:19:20 <Samu> this.CheckDepotsForStopedVehicles()
01:19:27 <Samu> i really should prioritize depots indeed
01:19:44 <Samu> having buses waiting 5 years in a depot to be renewed is bad
01:20:09 <Zuu> On line 724, it even runs that check in the middle of managing individual connections.
01:21:07 <Zuu> Oh, and a CluelessPlus special is that it stores vehicle and connection states in vehicle names. :-D
01:21:43 <Zuu> So I can click on vehicles in the depot and see if they are waiting to be sold or to be replaced. :-)
01:23:13 <Samu> my ai, up to about 1500-2000 road vehicles, it's fast enough
01:23:26 <Samu> after that, it becomes a crawl
01:23:50 <Samu> seems to become exponentially slower the more it gets
01:24:02 <Samu> could never reach 5k vehicles :(
01:24:42 <Samu> it tops at about 3500, and that's already 5 years per loop :(
01:25:41 <Samu> prioritizing depots is now on my to-do
01:27:45 <Samu> the dynamically adjustment of the number of vehicles, a feature i recently added, is a slow feature in the long run :|
01:28:33 <Samu> adds, then removes, then adds, then removes, depending on the vehicles moving at 0 km/h
01:28:52 <Samu> nice idea in theory, though
01:32:32 <Samu> timing my stuff is a good idea
01:32:58 <Samu> i should do something better about the way i'm timing stuff on the ai
01:33:24 <Samu> what I have was an afterthought, implemented in a rush
01:39:53 *** Thedarkb-T60 has joined #openttd
01:45:23 <Samu> gonna see how long this takes in my 3.5k vehicles game
01:47:14 <Zuu> My timing stuff was added quite late too. It doesn't excuse making it nice though. :-)
01:51:38 <Samu> oh gosh, it's so slow i'm already bored lol
01:52:09 <Zuu> But in overall large programs get out of shape.. CluelessPlus is large and have many quirks. :-)
01:53:04 <Zuu> Backwards compability is one thing that will make it hard to recover from mistakes or bad design in the past.
01:53:33 <Samu> managed 154 routes in 0 ticks?
01:55:28 <Samu> oh no, i wasn't saving the lists of stuff in depots
01:57:56 <Samu> saveload wasn't ready for this, omg i'm so stupid
01:58:56 <Samu> it works on the aircraft side apparently, it wasn't forgotten there
01:59:08 <Samu> i blame the original LuDi
02:00:15 <Samu> he's storing these vehicles in AILists, but saving lists is slow
02:00:27 <Samu> saving arrays should be faster, right?
02:01:46 <Zuu> I think you can only store squirrel data types so no AILists only arrays, dicts, and simple data types. But maybe OpenTTD have special code for AIList, you have to look up the docs on that.
02:02:02 <Zuu> CluelessPlus just return {} in Save()
02:02:34 <Zuu> And upon a loaded game in Start() it will read the connections from the map.
02:02:59 <Zuu> By parsing order lists, station names and vehicle names.
02:03:24 <Zuu> Using this it rebuilds the connection data to continue manage the connections.
02:05:54 <Samu> hmm, i don't think i can do that here
02:07:02 <Samu> well, it checks for vehicles that are old on the 2nd loop, not all is lost
02:07:44 <Samu> perhaps i dont need to save this data afterall
02:08:31 <Samu> gonna wait for the 2nd loop, zzz
02:09:49 <Zuu> Mind that you have limited time in Save() and Load(). If you spend too much time there, your AI will get killed.
02:10:44 <Zuu> So any data that is slow to determine for saving has to be pre-cached already before Save() is called. Load is easier, just save the data to this._save and handle it in Start() and you are good.
02:18:43 *** wodencafe has joined #openttd
02:21:19 *** wodencafe has joined #openttd
02:35:28 <Samu> managed the selling/renewing part of 90 routes out of 154 in 4440 ticks
02:36:40 <Samu> 103 days just for this :(
02:37:09 <Samu> then there's also the add vehicle to route part, oh crap, this is seriously too slow
03:13:13 <Samu> addremoveVehicleToRoute is slow
03:13:35 <Samu> these are the biggest offenders
03:14:31 <Samu> gonna wait for the selling/renewing again
03:33:05 <Samu> okay, i see where part of the slowdown is coming from
03:33:44 <Samu> creating a list of vehicles of type air or road so that I could count them
03:34:45 <Samu> apparently, i got about a total of 8000 vehicles total
03:35:14 <Samu> if i only want to count road, or air, it takes about 100 ticks
03:35:55 <Samu> but inside loops this adds up to become a slowball
08:25:38 *** andythenorth has joined #openttd
08:30:08 *** sla_ro|master has joined #openttd
08:38:31 <Eddi|zuHause> every time i think a bottleneck is solved, a new one pops up
08:38:36 <Eddi|zuHause> and ships are hopeless :/
08:44:34 <andythenorth> sounds like they managed to recreate TTD pretty well
08:44:40 <andythenorth> Eddi|zuHause: make a ship mod :P
08:49:48 * andythenorth wonders if TF vehicles make integer tile lengths :P
08:53:52 <Eddi|zuHause> uhm... no.. definitely not
08:54:42 <Eddi|zuHause> andythenorth: platforms have lengths of 80/160/240/...m, but trains have things like 232m
08:55:10 <Eddi|zuHause> wagons of the same generation don't have the same length, engine lengths are all different
08:55:56 <Eddi|zuHause> and trains of length 240m don't quite fit into 240m terminus platforms
08:57:09 <Eddi|zuHause> and i somehow lost my buffer stops...
09:07:55 *** tokai|noir has joined #openttd
09:07:55 *** ChanServ sets mode: +v tokai|noir
09:10:27 <andythenorth> I don't think I could play that game :P
09:10:38 <andythenorth> some part of my brain spent too much time in grids
09:10:59 <andythenorth> tile grids, lego grids, design grids
09:14:09 *** Wacko1976 has joined #openttd
09:18:52 <Eddi|zuHause> that might get tricky in a gridless game
09:20:05 <andythenorth> free-form curves bother me :P
09:26:16 <Eddi|zuHause> so i replaced half my trains with modded trains, and now i'm out of money :/
09:50:04 <Eddi|zuHause> <andythenorth> sounds like they managed to recreate TTD pretty well <-- it's actually very different from OpenTTD wrt ships, because in OpenTTD ships have endless capacity (you can just add more) whereas in TF they block each other at docks, and the entry/exit time is so large
09:52:28 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #6985: Fix (#6974): Add filter widget to api (and a double dot from somewhere) https://git.io/fhkIq
09:53:58 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh merged pull request #6985: Fix #6974: Add filter widget to api (and a double dot from somewhere) https://git.io/fpoNj
09:57:35 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #6987: Fix: [Win32] WIN32 may not be defined, always prefer the compiler pre… https://git.io/fhkIl
10:12:34 <andythenorth> blocking ships :P
10:12:38 <andythenorth> such terrible idea
11:04:06 *** Progman has joined #openttd
11:25:58 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #6917: Fix iconv and clang version detection on OSX https://git.io/fhkmU
11:33:21 <TrueBrain> as it appears nobody wants to close PRs ......
12:03:01 <TrueBrain> still a downside of our regression, if no gfx can be found, it 'hangs'
12:11:06 *** Thedarkb-T60 has joined #openttd
12:14:09 <TrueBrain> awh, regression fails on Windows .. no clue why .. total disaster :D
12:37:20 <Eddi|zuHause> sounds like you're having a great day.
12:51:08 <peter1138> create mode 100644 docs/Readme_Windows_MSVC.md
12:51:08 <peter1138> delete mode 100644 docs/Readme_Windows_MSVC.txt
12:51:15 <peter1138> What do I use to read a .md file?
13:08:14 <Eddi|zuHause> .md is effectively wiki-formatting?
13:10:58 *** HerzogDeXtEr has joined #openttd
13:22:57 <TrueBrain> so for some reason OpenTTD doesn't want to run the regression, but it is not telling me why it doesn't want to :D
13:27:55 *** Thedarkb-T60 has joined #openttd
13:27:59 *** sla_ro|master has joined #openttd
13:28:01 <LordAro> peter1138: Eddi|zuHause: markdown, specifically
13:31:07 *** wodencafe has joined #openttd
13:37:26 *** TrueBrain is now known as TrueBrain_ii
13:38:18 *** TrueBrain has joined #openttd
13:49:54 *** Wacko1976 has joined #openttd
14:01:47 *** wodencafe has joined #openttd
14:05:27 *** wodencafe has joined #openttd
14:21:39 <TrueBrain> meh; so running regression via MSVC doesn't work currently. Seems it used to work in 2011, but we used the mingw build to do that for the last 10 years. Guess .. we won't run regression on the Windows builds :P
14:30:39 <Eddi|zuHause> regression is overrated
14:36:28 <FLHerne> Eddi|zuHause: It's sort of halfway between formalised internet-speak formatting and wiki markup
14:51:27 *** andythenorth has joined #openttd
15:07:35 *** gelignite has joined #openttd
16:07:17 *** andythenorth has joined #openttd
16:09:39 *** Wacko1976 has joined #openttd
18:09:20 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 merged pull request #6987: Fix: [Win32] WIN32 may not be defined, always prefer the compiler pre… https://git.io/fp9vd
18:33:21 *** synchris has joined #openttd
18:46:26 *** gelignite has joined #openttd
18:53:03 *** Thedarkb-T60 has joined #openttd
18:57:03 *** Thedarkb-T60 has joined #openttd
20:04:39 <nielsm> TrueBrain, maybe add a cmdline flag to redirect/tee debug output to a file
20:04:55 <nielsm> or whatever output the regression tests make
20:08:04 <TrueBrain> that is also a possibility
20:08:16 <TrueBrain> but yeah, as it is now, it is not going to work :D
20:08:25 <TrueBrain> well, guess running the Windows build is better than not running it :)
20:16:32 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain opened pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhka6
20:17:34 <Eddi|zuHause> how is that part of the main repo?
20:18:53 <nielsm> the same way the makefiles and VS projects are
20:18:57 <nielsm> it's build instructions
20:19:55 <andythenorth> I need a train nerd
20:20:00 * andythenorth wonders who will step up
20:22:49 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhka6
20:25:21 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhka6
20:26:01 <nielsm> ooh, it's checking itself
20:26:27 <Eddi|zuHause> andythenorth: one who tells you how your sprites have the wrong number of rivets?
20:26:29 <TrueBrain> Eddi|zuHause: guess you haven;t looked in the main repo for a very long time :) Or any other "modern" project :D
20:26:48 <Eddi|zuHause> TrueBrain: not really
20:27:09 <TrueBrain> IaC - Infrastructure as Code
20:27:18 <TrueBrain> any decent project these days defines itself how the CI / CD works
20:27:35 <TrueBrain> (and ... tracked in a VCS!)
20:28:09 <Eddi|zuHause> well, my intuition would have been that CI config would be a separate repo
20:28:21 <Eddi|zuHause> like you have separate repos for all the bots
20:29:59 <andythenorth> Eddi|zuHause: one who tells me which european electric loco can be squashed for UK loading gauge :P
20:30:09 <andythenorth> I need a fake freight engine for 2020 or so
20:30:14 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhka6
20:31:17 <TrueBrain> Eddi|zuHause: you won't see many repositories doing that; and for good reason. How the CI runs, is part of a repository. If you put it somewhere else, locality increases, and that makes it harder to maintain
20:31:27 <TrueBrain> its pretty nifty, how "the world" solved this
20:31:28 <nielsm> andythenorth, with hitachi being in the european market (and england) now maybe a japanese one could work too? remember they largely run on narrow gauge
20:32:40 <TrueBrain> come to think of it .. 'tox' already did it too .. which is older than cloud
20:33:34 <andythenorth> nielsm: seems Hitachi have a UK factory :P
20:33:34 <Eddi|zuHause> andythenorth: doesn't siemens build UK versions?
20:33:59 <TrueBrain> Eddi|zuHause: btw, the bots are in a different repository, but the configuration is per repo too
20:34:43 <TrueBrain> okay, job seems to run now :D
20:39:15 <nielsm> it's certainly better than 20 minutes :)
20:39:26 *** Wormnest has joined #openttd
20:42:17 <TrueBrain> now the question .. shall we just switch or not?
20:43:03 <nielsm> it's probably no worse than the current jenkins
20:43:39 <TrueBrain> guess I wait for the review, and then switch Jenkins to Azure .. and we will see how much it burns :D
20:44:17 <andythenorth> it's obvs. nearly time to switch :D
20:44:23 <andythenorth> what could go wrong :P
20:47:15 <TrueBrain> not much honestly :)
20:48:22 <nielsm> i'd say it's pretty well tested already
20:49:12 <TrueBrain> and it ONLY runs on CI currently :)
20:56:50 <nielsm> but the PR is not being run on jenkins so it's blocking the regular merge
20:57:06 <TrueBrain> just not going to change that till it is approved :D
20:57:21 <TrueBrain> and lucky enough .. approval has nothing to do with what CIs think :)
21:07:20 *** Thedarkb-T60 has joined #openttd
21:09:08 <andythenorth> yeah I just found that
21:09:21 <andythenorth> life imitates art again :)
21:12:18 <andythenorth> any visuals yet? o_O
21:13:59 <Heiki> I haven’t found any, but probably quite similar to Class 88
21:17:28 <TrueBrain> nielsm: I didnt say it with that many words, but I was kinda hoping you would approve my PR :D
21:19:05 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh approved pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhkrz
21:21:24 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain commented on pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhkrr
21:21:31 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain merged pull request #6994: Change: use Azure Pipelines instead of self-hosted Jenkins for CI https://git.io/fhka6
21:24:44 <TrueBrain> lol; I now blocked all existing PRs :D
21:25:04 <TrueBrain> if any need merging without them getting an update, let me know
21:25:07 <TrueBrain> but I doubt any will
21:28:57 <glx> always fun to break everything ;)
21:29:25 <TrueBrain> I am so happy we now run native MacOS and Win32/Win64 is back in the CI again :D
21:29:49 <TrueBrain> hopefully this weekend I can get DigitalOcean setup decently enough to start with some IaC hosting :)
21:29:50 <glx> I remember the C to CPP switch breaking all patches
21:30:11 <Borg> does anyone build OpenTTD w/ Mingw here?
21:30:42 <glx> disable freetype and it should work
21:30:42 <nielsm> before any nightlies can be published the revision detection needs to be fixed I think
21:30:51 <TrueBrain> its broken? Bah .. and yes, indeed
21:31:14 <glx> was broken before your work TrueBrain ;)
21:31:14 <nielsm> due to the azure pipeline doing a detached HEAD checkout
21:31:53 <TrueBrain> nielsm: that is easily fixable :)
21:32:43 <glx> Borg: hmm mingw32 or mingw64 BTW ?
21:33:01 <Borg> im still on best windows ever made..
21:33:43 <glx> ah it may build then, for mingw64 it definitively fails with freetype (because harfbuzz)
21:36:36 <TrueBrain> nielsm: I guess the main issue with a detached head is that branch detection fails?
21:36:51 <TrueBrain> what we normally do for nightlies, is force the version anyway
21:37:06 <TrueBrain> as we .. had a lot of issues with detection over the years :)
21:38:37 <nielsm> nightlies are just master anyway yes
21:39:15 <TrueBrain> yeah; I think we just need a new way of doing this
21:39:33 <TrueBrain> openttd-openttd-pr6698
21:39:45 <TrueBrain> so you can also get: truebrain-openttd-master
21:39:57 <TrueBrain> but something for another day :)
21:40:01 <nielsm> rather than depending on whatever bad branch name the PR submitter chose
21:42:24 <glx> I think with github desktop I get pr/XXXX as branch name
21:43:04 * glx likes github desktop for things like that
21:43:23 <glx> not for the silly merges I need to fix with rebase -i ;)
21:47:55 <glx> note for later: don't switch branch in github desktop while compiling
22:15:49 <glx> ok mingw32 still builds and run
22:17:01 <andythenorth> all modern locomotives look like cheese :P
22:17:20 <andythenorth> harder to draw, and kind of boring
22:27:50 <Borg> haha.. I just upgraded so old Network safe.. w/ new GRF..
22:28:12 <Borg> now some power stations are overcharging.... 141% capacity output :D
22:36:43 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 opened pull request #6995: Fix 59e42ea17, 1f083c3ac: make clean was not fully cleaning https://git.io/fhk6u
22:38:32 <Eddi|zuHause> 141% is sqrt(2), so i would expect that to be a reasonable output of a power station :)
22:40:27 <Eddi|zuHause> (given that for AC, this is the factor between average and max)
22:41:01 <glx> TrueBrain: did the PR trigger the CI ?
22:41:14 <Borg> the true reason is. that I added support for power station layouts.. so smaller produce less power (need less coal too)
22:41:29 <TrueBrain> glx: good point, I forgot to switch something on
22:41:35 <Eddi|zuHause> so many revolutions today, what's next, actual nightly builds? :p
22:42:02 <LordAro> woah now, let's not get ahead of ourselves
22:42:28 <TrueBrain> glx: you need to allow forks to run .. and you cannot do that in the YAML .. I keep forgetting that :D
22:43:07 <TrueBrain> glx: can you push again? (rebase, commit-message change, anything?)
22:43:13 <Eddi|zuHause> so much for locality?
22:43:26 <TrueBrain> glx: suggestion: 'make clean' between quotes :P
22:43:34 <TrueBrain> Eddi|zuHause: glass half-empty kinda guy, I see
22:44:08 <Eddi|zuHause> hey, i was just praising all the progress
22:44:41 <DorpsGek_II> [OpenTTD/OpenTTD] glx22 updated pull request #6995: Fix 59e42ea17, 1f083c3ac: make clean was not fully cleaning https://git.io/fhk6u
22:45:06 <TrueBrain> glx: there we go :)
22:45:25 <Eddi|zuHause> (on an unrelated note, sudden switching between high and low could be a sign of depression)
22:46:01 <TrueBrain> luckily, that is very treatable these days :)
22:46:25 <TrueBrain> glx: hmm ... let me investigate what is going wrong here :)
22:47:10 <TrueBrain> "Latest build not found" .. wuth?
22:49:12 <glx> at least linux and macOS work
22:50:31 *** Wacko1976 has joined #openttd
22:50:34 <TrueBrain> how weird .. it cannot find the dependencies in the artifact ..
22:51:56 <glx> but I like how the checks are detailled in github
22:52:28 <peter1138> I set up my own Jenkins just so I could copy what we'd done here, and now I have no clue ;p
22:57:18 <TrueBrain> a bit of trial and error .. lets see if I can find the issue
22:57:35 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain opened pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
22:58:06 <TrueBrain> no clue if it uses my new yml now, or the one in master :D
22:59:41 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:01:12 <TrueBrain> not what I would like, but fair enough
23:04:55 <TrueBrain> okay, this worked a moment ago .. lolz .. :D
23:05:58 <TrueBrain> their editor also lost track of the artifact ... which really is there :D
23:06:37 <nielsm> anyone more awake than me, who can check this for me? https://0x0.st/snTm.txt -- the "do not write multiple blank lines in a row" part does not work, it does produce multiple blank lines around a skipped block
23:06:46 <nielsm> (the actual block detection and skipping works)
23:13:57 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:14:43 <TrueBrain> hmm .. if this fixes the issue, I have another problem, but at least I know where that problem is :D
23:15:42 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:15:52 <TrueBrain> ugh, sorry for the spam .. not really another way to test this :(
23:17:16 *** andythenorth has left #openttd
23:17:37 <TrueBrain> this worked moments ago ... lolz
23:18:29 <nielsm> ah figured out my problem
23:20:30 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:23:28 <TrueBrain> okay, it works if I run it manually
23:23:31 <TrueBrain> it fails when it comes from a PR
23:24:02 <glx> depends on the username ?
23:25:10 <nielsm> hm did we use "remove:" or "update:" for removing unused strings from language files?
23:25:29 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:25:37 <glx> we used to use "cleanup:"
23:25:53 <TrueBrain> nielsm: the fact you ask, means we made it too complicated :D
23:26:21 <TrueBrain> fix/change/update is already confusing for me :D
23:26:35 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh updated pull request #6990: Fix: Correct display of industry requires/produces in Build Industry window https://git.io/fhTef
23:29:24 <DorpsGek_II> [OpenTTD/OpenTTD] andythenorth commented on issue #4410: Trains: var4A in purchase list (current railtype) https://git.io/fhkPy
23:29:39 <TrueBrain> I am very tempted to store the artifact somewhere else, so I can just download it :P
23:29:58 <glx> yup looking at the logs I see many Cleanup for removing strings
23:31:09 <glx> TrueBrain: the other option being to rebuild the deps each time I guess
23:31:29 <TrueBrain> glx: yeah .. adds 10 more minutes to the CI :(
23:34:19 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
23:35:16 <TrueBrain> now I come to think of it, it might be good to publish it on github as 'release' :P
23:35:18 <TrueBrain> but that is for another day
23:35:36 <TrueBrain> (possibly others find it useful to have the dependencies pre-compiled too)
23:38:12 <TrueBrain> I think it has to do with authentication of some sorts
23:39:43 <glx> because when you start it with your truebrain account it works
23:41:54 <TrueBrain> of all the issues, I did not expect this to be one :)
23:48:39 <TrueBrain> yeah; the issue seems to be in permissions inside Azure Pipelines
23:48:53 <TrueBrain> a build triggered from a PR has less rights
23:48:57 <TrueBrain> which seems to give issues here
23:50:16 *** sla_ro|master2 has joined #openttd
23:52:02 <glx> oh nice I have access to the configuration ;)
23:52:11 <glx> (don't worry I won't touch)
23:52:48 <TrueBrain> how do you have access?
23:52:51 <TrueBrain> (that is interesting :D)
23:53:19 <glx> only the basic settings, like repos selection
23:53:42 <TrueBrain> you can save it too?
23:54:18 <TrueBrain> yeah, ofc, you are owner there
23:54:43 <TrueBrain> but yeah, the issue is not with github
23:54:47 <TrueBrain> it is in azure pipelines
23:55:03 <TrueBrain> (the artifact is in azure pipelines, and it tries to download it (and fails))
23:55:49 <glx> hmm I could force merge my PR
23:55:56 <glx> to trigger a master build
23:56:09 <TrueBrain> there is no master build
23:56:15 <TrueBrain> which is a good point, there is no master build :P
23:56:34 <glx> of course because there's no nightly yet
23:56:49 <TrueBrain> yeah, but we can also validate master after commit
23:56:54 <TrueBrain> but not important atm :)
23:57:44 <TrueBrain> when I manually query the API, I get results .. hmm
23:58:05 <TrueBrain> so .... I need more debugging info
23:59:15 <DorpsGek_II> [OpenTTD/OpenTTD] TrueBrain updated pull request #6996: Fix: [AzurePipelines] download the correct Windows Dependency artifact https://git.io/fhkis
continue to next day ⏵
