IRC logs for #openttd on OFTC at 2021-07-21
⏴ go to previous day
00:00:52 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
00:00:52 *** Gustavo6046_ is now known as Gustavo6046
00:25:49 *** Speeder has joined #openttd
00:58:56 *** Gustavo6046_ has joined #openttd
01:02:05 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
01:02:05 *** Gustavo6046_ is now known as Gustavo6046
01:14:21 *** Progman has quit IRC (Remote host closed the connection)
01:16:11 *** Tirili has quit IRC (Quit: Leaving)
01:23:55 <esselfe> I found a bug in the dedicated server: The money reported from 'companies' is about half of the amount reported in the client
01:56:18 *** Wormnest has quit IRC (Quit: Leaving)
02:32:46 *** debdog has quit IRC (Ping timeout: 480 seconds)
03:14:50 *** y2kboy23 has joined #openttd
03:22:57 *** y2kboy23 has joined #openttd
03:45:50 *** y2kboy23 has joined #openttd
03:51:30 *** HerzogDeXtEr has joined #openttd
04:24:59 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
05:32:11 *** andythenorth has joined #openttd
05:57:20 *** tokai|noir has joined #openttd
05:57:20 *** ChanServ sets mode: +v tokai|noir
06:04:10 *** tokai has quit IRC (Ping timeout: 480 seconds)
06:07:36 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
06:11:00 *** sla_ro|master has joined #openttd
06:31:41 *** jottyfan has joined #openttd
07:04:47 <LordAro> mm, dollars vs pounds, most likely
07:32:39 <peter1138> Hmm, Twitch stopped working for me last night, and still isn't working.
07:52:35 *** jottyfan has quit IRC (Quit: jottyfan)
07:58:21 *** WormnestAndroid has quit IRC (Remote host closed the connection)
07:58:34 *** WormnestAndroid has joined #openttd
08:23:08 *** jottyfan has joined #openttd
08:33:42 *** HerzogDeXtEr has joined #openttd
08:45:25 *** ChanServ sets mode: +v tokai
08:51:00 *** jottyfan has quit IRC (Quit: jottyfan)
09:13:18 *** Speedy` has joined #openttd
10:11:50 <LordAro> the irritating thing about OGL stuff is that OTTD no longer starts if i've updated my graphics driver, due to a terminate() call buried within SDL
10:12:33 <LordAro> luckily, -v sdl is always an option
10:13:12 *** andythenorth has quit IRC (Quit: andythenorth)
10:24:47 *** Progman has joined #openttd
10:54:38 <DorpsGek> [OpenTTD/actions] glx22 approved pull request #33: Change: use GitHub composite actions over NPM-based for deploy-aws action https://git.io/JlJYc
11:07:19 *** aperezdc__ has quit IRC (Quit: issued !quit command)
11:47:37 <DorpsGek> [OpenTTD/OpenTTD] LordAro commented on issue #9071: [NRT] Building a tramtype over a roadtype with the NO_HOUSES flag set causes the flag to be ignored https://git.io/JOrJh
12:26:40 <DorpsGek> [OpenTTD/OpenTTD] LordAro commented on issue #9030: Global story book opens on top of company story book https://git.io/JOqgs
12:55:58 *** Compu has quit IRC (Remote host closed the connection)
13:05:52 <_dp_> this whole global/company thing is pretty confusing in goals and in storybook it's even worse
13:07:06 <_dp_> story book button can probaly just always open company story book, it has global pages anyway
13:07:53 <_dp_> and make company league table scriptable instead of global goals
13:08:53 <Samu> hmm, reset all settings is misleading
13:09:15 <Samu> i thought it would only reset game settings
13:09:28 <Samu> it reseted everything, even gui
13:10:35 <Samu> all setting types is what it should say
13:46:13 <Samu> STR_CONFIG_SETTING_RESET_ALL_CONFIRMATION_DIALOG_TEXT :{WHITE}This action will reset all game settings to their default values.{}Are you sure you want to proceed?
13:46:30 <Samu> "all game settings" is misleading, in the context where this window is shown
13:46:40 <LordAro> i would go with "ALL settings"
13:46:47 <Samu> i would go all setting types
13:46:50 <LordAro> (to really emphasise the ALL)
13:46:57 <LordAro> "types" isn't helpful here
13:47:08 <Samu> there's a preset named "all settings types" and another saying "game settings"
13:47:42 <LordAro> i'd probably rename that to "All settings", tbh
13:48:07 <LordAro> that's explicitly differentiating between different types
13:48:15 <LordAro> so "all types" is fine there
13:48:27 <LordAro> the "reset all values" is not, so "types" is not necessary
13:49:36 *** roadt__ has joined #openttd
13:49:59 *** sla_ro|master has quit IRC ()
13:53:11 *** virtualrandomnumber has joined #openttd
13:53:29 *** virtualrandomnumber has quit IRC ()
13:53:49 *** roadt_ has quit IRC (Read error: Connection reset by peer)
13:56:00 <Samu> honestly I would actually like to have a 'reset all settings of chosen type'
13:57:17 <LordAro> though i think the usecase for that would be even lower than the current reset all
14:52:11 *** tokai|noir has joined #openttd
14:52:11 *** ChanServ sets mode: +v tokai|noir
14:52:50 <Samu> oh man, working on the readme and changelog for my AI kinda sucks
14:58:32 <Samu> most AIs work fine on 256x256 maps, i realised
14:59:07 <Samu> then there's my AI, doesn't exactly sucks on 256x256, but it underperforms in relation to others
14:59:10 *** tokai has quit IRC (Ping timeout: 480 seconds)
15:06:06 <Samu> gotta mess with SuperSimpleAI, it has interesting settings
15:06:52 *** Beer is now known as CVE-2021-33910
15:48:52 *** Wormnest has joined #openttd
16:10:58 *** CVE-2021-33910 has quit IRC (Quit: Leaving)
16:15:38 <Samu> i'm tired of seeing this 'pptile' error from roadrunner
16:26:04 <Samu> his code is terrible to look at
16:40:28 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain commented on issue #9030: Global story book opens on top of company story book https://git.io/JOqgs
16:41:11 <TrueBrain> LordAro: re 9033, change of tile .. now that would be plausible :D Nice investigation .. now for a fix ;) :D :D
17:42:46 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
17:45:23 <DorpsGek> [OpenTTD/actions] TrueBrain merged pull request #33: Change: use GitHub composite actions over NPM-based for deploy-aws action https://git.io/JWj5c
17:45:37 <TrueBrain> lets test this out :D
17:45:56 <glx> lets find out if it works or not
17:46:18 <DorpsGek> [OpenTTD/master-server] TrueBrain opened pull request #50: Fix: don't announce a direct-ip if we already know it https://git.io/JlUyD
17:46:37 <glx> ah hoping it will fix the leak
17:46:47 <TrueBrain> yeah .. I should also fix it on the other side I guess
17:46:54 <TrueBrain> but let's start with not wasting bandwidth :D
17:47:51 <DorpsGek> [OpenTTD/master-server] glx22 approved pull request #50: Fix: don't announce a direct-ip if we already know it https://git.io/JlUSe
17:47:54 <DorpsGek> [OpenTTD/master-server] TrueBrain merged pull request #50: Fix: don't announce a direct-ip if we already know it https://git.io/JlUyD
17:50:41 <TrueBrain> at least it worked on the actions repo itself
17:52:15 <TrueBrain> yippie, seems to work fine
18:05:14 *** aperezdc__ has joined #openttd
18:10:10 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain opened pull request #37: Fix: don't announce a direct-ip already known to the stream https://git.io/JlU76
18:10:22 *** andythenorth has joined #openttd
18:10:47 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain opened pull request #38: Fix: ensure direct-ips of servers are unique https://git.io/JlU79
18:11:19 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain merged pull request #37: Fix: don't announce a direct-ip already known to the stream https://git.io/JlU76
18:11:41 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain updated pull request #38: Fix: ensure direct-ips of servers are unique https://git.io/JlU79
18:12:33 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain merged pull request #38: Fix: ensure direct-ips of servers are unique https://git.io/JlU79
18:13:47 <TrueBrain> let's see if that helps for memory issues :D
18:13:57 <TrueBrain> I do hope I didn't destroy IPv6 support, but it should still work fine :D
18:14:08 <TrueBrain> I remembered to do an "rpartition" on ":"
18:25:04 <TrueBrain> btw, just for realistion .. the Game Coordinator is running in ~40MB of RAM :P
18:28:45 *** sla_ro|master has joined #openttd
18:33:15 *** ioangogo_ has quit IRC (Read error: Connection reset by peer)
18:33:39 *** ioangogo has joined #openttd
18:52:08 <DorpsGek> - Update: Translations from eints (by translators)
18:57:50 <TrueBrain> lol ... the "my_client" is not cleaned up if the connection is closed by anything outside the clients control
18:58:12 <TrueBrain> the code is .. not pretty .. difficult to follow
19:15:34 <glx> <TrueBrain> btw, just for realistion .. the Game Coordinator is running in ~40MB of RAM :P <-- so controlling memory usage is very important
19:15:59 <TrueBrain> not really .. I could also just assign it 512MB RAM
19:18:42 <Samu> my canalpathfinder still consumes abnormal amounts of memory :(
19:21:20 <Rubidium> it's funny that you're talking about assiging something 512 MB of RAM when the data only needs a few hundred kB at most. Nobody seems to bother about memory anymore; just add more memory and that solves all problems
19:22:57 <Rubidium> but alas, apparantly adding memory to 32 GB in a laptop solves the issue with Teams wasting heaps and heaps. Yeah right... as if 16 GB isn't enough already
19:22:58 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain opened pull request #9453: Fix: crash when joining a server again after a TCP disconnect https://git.io/JlTev
19:23:19 <TrueBrain> Rubidium: I think you missed the part it is running on 40MB RAM, and it has 64MB RAM assigned ;)
19:23:33 <TrueBrain> not sure why you put me in the group "nobody" there :)
19:24:40 <Rubidium> I didn't miss that; if you think about it, it should not need more than a MB of memory. But alas, overhead for Python and friends (not that C programs have much benefit due to huge libraries getting loaded)
19:25:00 <TrueBrain> it runs in 40MB of RAM!
19:25:04 <TrueBrain> you think that is a lot?
19:25:34 <TrueBrain> a full interpreter is running in there too :P
19:26:17 * Rubidium wonders how memory much (q)basic used/needed
19:26:47 <TrueBrain> the only reason I mention the 512MB, is to not have people confuse my need to use as little memory as possible, as a must for it to run
19:27:09 <TrueBrain> I just like to use very tiny EC2 instances to run our full infra
19:28:13 <TrueBrain> for example eints requires at least 386MB to run in production .. which really annoys me :P
19:29:27 <TrueBrain> even the wiki runs in only 128MB RAM .. and that caches a lot of pages in memory :D
19:30:10 <Rubidium> don't get me wrong, 40 MB is much better than 512 MB for the GC and I like that you try to reduce the amount of memory required
19:31:41 <Rubidium> it's just that in the current state of things the frameworks already need like 90% of the memory needed for the actual service
19:32:27 <TrueBrain> "back in my day" :P
19:32:57 <TrueBrain> it just amuses me .. we run 100+ requests/sec over our services .. serve thousands of players a day with binaries and multiplayer .. and all that runs on < 4GB of RAM :D
19:32:58 <Rubidium> yeah, how much did MSU need? ;)
19:33:19 <TrueBrain> the C++ variant .. even more, because of the MySQL shit
19:34:23 <Rubidium> yeah, as I said... huge libraries/frameworks
19:34:48 <TrueBrain> what I like about Python, it needs so little bytes per socket .. which is really nice
19:35:00 <TrueBrain> comes down to like 100 bytes per client connected, worst case :)
19:35:09 <TrueBrain> which scales surprisingly well with little memory :P
19:35:36 <TrueBrain> I also like the trick I do for bananas-server .. a 1GB file you download via BaNaNaS over the TCP protocol only uses ~64KB of RAM on the server
19:36:02 <TrueBrain> it starts a HTTP GET on the S3 bucket, and when it receives the data sends it to the client, with a max buffer of 64KB .. if that buffer is full, it stalls the HTTP connection
19:36:12 <TrueBrain> which is a really nice trick to reduce the memory usage there
19:37:11 <TrueBrain> in the past it fetched the whole file no matter what .. which caused server-side crashes from time to time .. people can have REALLY slow connections
19:37:16 <TrueBrain> (down to 1KB/s connections I have seen)
19:37:30 <SpComb> a t3.micro with 1GB can barely run `yum update`
19:37:51 <TrueBrain> 0.5GB RAM, and yeah, depending on the age of the AM2 image, that is an issue :)
19:37:59 <TrueBrain> there are days I cannot reinstall the load balancers
19:38:04 <TrueBrain> I have to wait for the AM2 image to update :P
19:38:07 <TrueBrain> which is hilarious :D
19:38:24 <TrueBrain> well, the one with 0.5GB openttd.org runs :P
19:39:11 <TrueBrain> but, mostly, it works ;)
19:39:25 <SpComb> just don't sneeze or it falls ove
19:39:50 <TrueBrain> if it auto-heals because a bit fell over, it can be a bit troublesome
19:40:22 <TrueBrain> but .. yum .. yeah ... yum has other design issues :P
19:42:12 <TrueBrain> "SocketClosed openttd_protocol.wire.tcp in send_packet"
19:42:21 <TrueBrain> tells me exactly where the software is making a boo-boo :D
19:42:42 <TrueBrain> Task exception was never retrieved; future: <Task finished name='Task-3660' coro=<TokenVerify._conclude_detection() done ...
19:44:52 <SpComb> in other news, I'm experimenting with EC2 instance connect, and I now have an Ansible connection plugin that generates a new ssh keypair and uploads it via the ec2-instance-connect API... for every single Ansible task. Need to figure out how to optimize that next, particularly since it's actually using ControlMaster=auto multiplexing with ControlPersist, so SSH is really only ever connecting once...
19:45:27 <TrueBrain> that sounds wasteful, yes :)
19:45:34 <TrueBrain> otherwise EC2 instance connect is pretty nice
19:45:58 <TrueBrain> although I stopped provisioning EC2 instances via Ansible like that .. I just put it in the user-data
19:46:15 <TrueBrain> not sure it is "better"
19:46:18 <TrueBrain> just scales easier :P
19:46:25 <SpComb> user-data and hacky cloud-init scripts?
19:46:40 <TrueBrain> not sure it is hacky, as such, but yeah
19:47:25 <SpComb> I prefer the maintainability of Ansible roles, but I think there should be some happy medium
19:48:21 <TrueBrain> I have to say, I don't use that many EC2 instances anymore .. ECS is running most workloads for me these days
19:48:36 <TrueBrain> (which runs on EC2 ofc, but the images are provided by AWS, so I don't even know what is on it :P )
19:49:01 <SpComb> I kinda want something where you can reprovision an instance with a new AMI, and keep an EBS volume attached that just gets mounted on /var/lib/docker, and then it just autostarts the same Docker containers and keeps going
19:49:36 <SpComb> but there's many more nasty details to it than that, and yeah, at some point it would just make more sense to use something like ECS
19:49:52 <Timberwolf> We still only have two EC2 instances at work.
19:49:58 <TrueBrain> it also forces you to think in non-persistent containers :)
19:49:58 <Timberwolf> Production bastion and staging bastion.
19:50:03 <Xaroth> SpComb: you can always just provision the EC2 instances with a set key, and have ansible use that?
19:50:13 <Timberwolf> Everything else is either managed, or via lambda.
19:50:43 <TrueBrain> I still have to look into how to make good use of Lambdas
19:50:47 <Timberwolf> I suspect we'll hit cold start problems as we build more APIs where the clients can't tolerate "oh it took 15s+ to respond, nvm"
19:51:13 <TrueBrain> I want to run the wiki on cloudflare workers
19:51:16 <Timberwolf> Actually, GoLang cold starts aren't too bad. In the early days the C# ones were disastrous, especially if you used one of the slower-booting DI frameworks.
19:51:17 <TrueBrain> I think that is possible ..
19:51:24 <DorpsGek> [OpenTTD/OpenTTD] rubidium42 opened pull request #9454: Cleanup: remove some references to the old master-server https://git.io/JlTJj
19:51:50 <Xaroth> I really wish CloudFlare would improve their workers docs, it's so.... meh... I mean, it works, and when it works it works very well, but holy crap it's annoying at times.
19:52:15 <Xaroth> Also, the "dev server" for workers literally just uploads your code to a test worker and runs on that
19:52:19 <Xaroth> but it pipes from localhost
19:52:26 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #9454: Cleanup: remove some references to the old master-server https://git.io/JlTUq
19:53:11 <SpComb> lamba would probably be a poor choice for a fairly large spring backend
19:53:24 <TrueBrain> spin-up time: 3 minutes
19:53:29 <TrueBrain> sounds reasonable for any API? :P
19:54:06 <TrueBrain> I really have to invest some time moving parts of openttd.org to Cloudflare .. will save a lot of money :P
19:54:31 <TrueBrain> but their DNS thing makes it such an "everything or nothing" concept
19:54:36 <DorpsGek> [OpenTTD/OpenTTD] rubidium42 approved pull request #9453: Fix: crash when joining a server again after a TCP disconnect https://git.io/JlTUu
19:54:52 <TrueBrain> Rubidium: did you actually follow the issue? :P As my head still spins ...
19:55:27 <SpComb> Xaroth: you mean from cloud-init user-data? Yeah, but keys change over time, and that sucks for key management
19:55:34 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain merged pull request #9453: Fix: crash when joining a server again after a TCP disconnect https://git.io/JlTev
19:55:58 <Rubidium> well... mine's spinning a bit as well ;) It seems okay, and I looked for other places that might need some love in that respect, but could not find any
19:56:27 <TrueBrain> curious if there are now other cases that cause crashes .. but we will find that out sooner or later :P
19:56:34 <TrueBrain> I tested a lot of scenarios at least ..
19:56:42 <TrueBrain> I really hope this is the end of CloseConnection mess :D
19:57:36 <Xaroth> SpComb: you can always ask on #ansible-aws on Libera, pretty useful place most of the time, they've been very helpful last time I had to deal with AWS
19:59:09 <SpComb> yeah, I might need to start asking around, trying to understand the Ansible codebase is a pain... developing a custom connection plugin might be a bad idea anyways, from a future maintenance perspective
19:59:47 <Xaroth> They seem a bit lackluster in their way of tackling things
20:00:01 <Xaroth> before all plugins were separate, then they pulled it all into a monorepo, now it's all separated again
20:00:58 <Xaroth> I sent in one small PR for some docs stuff, and they still keep inviting me for their weekly documentation discussions
20:01:12 <SpComb> I think it's going to be problem with any massive Python codebase with a long history, the internal interfaces are never going to be very well documented
20:01:45 <Xaroth> I think it's more due to the scope of that they support
20:01:58 <Xaroth> it's an absolute truckload of plugins and stuff
20:05:27 <TrueBrain> glx: well, in good news, it indeed seems to have fixed the memory issue :D
20:05:43 <TrueBrain> it is one stable flat line now for longer than ever before :P
20:28:55 <DorpsGek> [OpenTTD/OpenTTD] rubidium42 merged pull request #9454: Cleanup: remove some references to the old master-server https://git.io/JlTJj
20:36:59 *** jottyfan has joined #openttd
20:41:00 *** ChanServ sets mode: +v tokai
20:47:55 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
20:57:41 *** Samu has quit IRC (Quit: Leaving)
21:06:56 *** sla_ro|master has quit IRC ()
21:09:43 *** tokai|noir has joined #openttd
21:09:43 *** ChanServ sets mode: +v tokai|noir
21:16:42 *** tokai has quit IRC (Ping timeout: 480 seconds)
21:25:00 *** jottyfan has quit IRC (Quit: jottyfan)
21:29:40 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
21:36:29 *** andythenorth has quit IRC (Quit: andythenorth)
21:46:32 *** nielsm has quit IRC (Ping timeout: 480 seconds)
22:57:53 *** Speeder_ has joined #openttd
23:05:27 *** Speeder has quit IRC (Ping timeout: 480 seconds)
23:07:28 *** azubieta60 has quit IRC (Quit: Ping timeout (120 seconds))
23:07:52 *** azubieta60 has joined #openttd
23:09:01 *** Webster has quit IRC (Remote host closed the connection)
23:09:54 *** Webster has joined #openttd
23:18:54 *** Wuzzy has quit IRC (Remote host closed the connection)
continue to next day ⏵
