IRC logs for #openttd on OFTC at 2025-05-20
⏴ go to previous day
01:28:13 *** jinks has quit IRC (Remote host closed the connection)
01:59:59 *** Wormnest has quit IRC (Quit: Leaving)
02:27:19 *** WormnestAndroid has quit IRC (Remote host closed the connection)
02:29:00 *** WormnestAndroid has joined #openttd
02:55:57 *** gnu_jj_ has joined #openttd
02:59:35 *** gnu_jj has quit IRC (Ping timeout: 480 seconds)
03:37:44 *** debdog has quit IRC (Ping timeout: 480 seconds)
03:54:58 *** D-HUND is now known as debdog
04:46:20 <DorpsGek> - Update: Translations from eints (by translators)
06:29:30 *** Heiki has quit IRC (Read error: Connection reset by peer)
06:29:31 *** soylent_cow[m] has quit IRC (Remote host closed the connection)
07:07:49 <LordAro> coffee might help me locate my enter key a bit better
07:12:28 <peter1138[d]> I have found coffee.
07:14:54 <pickpacket> LordAro: coffee may do the opposite. It erodes society
07:14:59 <pickpacket> Tea is where it's at
07:15:17 <LordAro> I made more tea by accident
07:15:33 <LordAro> I don't even have any milk left
07:19:41 <peter1138[d]> Black tea isn't the worst, but it's a mood...
07:21:27 *** reldred has joined #openttd
07:21:27 <reldred> I spent most of yesterday drinking cheap nasty mochas
07:21:47 <reldred> instant coffee powder and cocoa powder
07:25:11 <peter1138[d]> Are we moving away from Github yet?
07:36:24 <pickpacket> peter1138[d]: to...?
07:36:37 <pickpacket> I'm all for it. Github is shit
07:37:59 <reldred> gitlab hosted on andy's macbook
07:38:21 <LordAro> gitlab's pushing AI pretty hard too
07:38:53 <pickpacket> GoG on andy's macbook, maybe?
07:39:11 <reldred> what's that one my friend in duh fence uses that she absolutely hates
07:39:52 <reldred> (it's some disgusting archaic shite, I just can't remember the name of it)
07:40:04 <peter1138[d]> Oh however it's spelt?
07:40:15 <pickpacket> kinnda ironic that the gogs source code is on github
07:40:23 <pickpacket> peter1138[d]: codeberg is a thing, yeah
07:41:14 <reldred> Clearcase, that was it.
07:41:28 <reldred> It's apparently absolutely horrific
07:41:33 <peter1138[d]> I didn't know gogs was still a thing. I use gitea that was borne out of it.
07:41:55 <peter1138[d]> And forgejo is a fork of gitea after there was a bit of grumbling.
07:43:25 <peter1138[d]> (Codeberge uses forgejo)
07:45:44 <pickpacket> reldred: I've worked a lot in clearcase! For the first two or so years of my career
07:46:09 *** greeter has quit IRC (Remote host closed the connection)
07:46:10 <pickpacket> and also *really* expensive, and you need a license for each dev
07:46:38 <reldred> Oh yeah I wasn't actually suggesting it
07:47:28 <pickpacket> I know :) Nobody would seriously suggest it
07:48:26 *** greeter has joined #openttd
07:50:40 <LordAro> websvn & a maze of php & python & make build scripts
07:51:00 <pickpacket> sounds like a dream
07:51:01 <LordAro> it "worked" for #work for quite some time
07:51:26 <pickpacket> I think the biggest issue would be finding an equivalent of github actions
07:54:43 <LordAro> there's no direct equivalent that i'm aware of, gitlab-ci etc are all just "run this script"
07:54:50 <LordAro> which is arguably better anyway, less vendor lockin
07:58:50 <peter1138[d]> Other systems are mostly compatible with github-style actions. The bigger issue is the availability of computing time, especially things like the MacOS environments.
07:59:59 <pickpacket> LordAro: we use gitlab ci at $work and it's quite horrible to deal with
08:07:14 <LordAro> it's a bit funky at times, but i don't have any particular complaints about it, what do you find horrible?
08:07:53 <LordAro> (admittedly, i've spent the last 4 years largely buried in it, so my opinion may not be representative)
08:11:45 <peter1138[d]> Back when I tried it, gitlab was a proper pain to work with on the hosting side.
08:13:01 <LordAro> basic hosting is trivial, just installing a deb package
08:13:13 <LordAro> does get a bit more complicated when you start doing all the HA stuff
08:20:48 *** Smedles has joined #openttd
08:45:53 *** jinks has quit IRC (Remote host closed the connection)
09:26:49 <yiffgirl> i've been using forgejo on a friend's instance, it's great. i assume since codeberg is a thing that it'd be suitable for openttd-scale projects, but obviously my work is not openttd scale, so ymmv
09:28:52 <yiffgirl> in any case i wholly support any attempt to migrate off of github
09:51:54 <peter1138[d]> I suppose the cargo dest progress is not really anything more than "wouldn't it be nice if".
10:04:21 <xarick> how to make it shorter but understandable
10:10:59 *** Flygon has quit IRC (Read error: Connection reset by peer)
10:11:44 <_zephyris> peter1138[d]: Tempted to try this, just to annoy peter1138[d] 😉
10:23:22 <peter1138[d]> Hmm, maybe we should introduce a COC etc... :p
10:25:05 <peter1138[d]> xarick: Honesly it's highly unlikely that anyone else is reading all these debug messages your AI produces.
10:25:53 <peter1138[d]> It's not actually necessary to output messages for every single thing your AI does.
10:26:17 <peter1138[d]> it's only useful for you as the developer of it to develop it.
10:39:22 <peter1138[d]> _zephyris: Rights can be revoked... 😄
10:41:54 <_zephyris> I'm genuinely intrigued by it's capabilities
10:42:24 <_zephyris> I suspect anything which hits eg. window layout would give ai slop, but you never know
10:43:07 <peter1138[d]> I'm genuinely disgusted by even entertaining the idea.
10:49:46 <tabytac> BuT iTs tHe fUtUrE oF aI cOdINg!
10:50:09 <_zephyris> So you won't even test it?
10:56:15 <peter1138[d]> So yes, this "AI" lark is very polarising, And yes. I am completely opposed, and do not want to try it. I am not alone.
11:01:53 <andythenorth> also tempted to try copilot
11:02:52 <andythenorth> it's probably moderately good at html work
11:08:37 <talltyler> I am also an AI hater. 🙂
11:17:14 <peter1138[d]> These guys are unironically using an engine dyno that uses an old terminal system.
11:24:33 *** WormnestAndroid has quit IRC (Remote host closed the connection)
11:24:37 *** WormnestAndroid has joined #openttd
11:24:42 *** foodliker has joined #openttd
11:25:50 <andythenorth> putting aside the tech bro related problems, AI coding assistants are just llike a battery screwdriver
11:26:03 <andythenorth> I remember how controversial battery screwdrivers were when they came out
11:26:23 <reldred> Fuck I hate battery screwdrivers, or worse, technicians using an 18v cordless.
11:26:46 <reldred> Theyre still controversial to me
11:26:57 <kuhnovic> Just learn to use the chuck clutch, it's not that difficult 😛
11:27:18 <reldred> Hey look don’t get me wrong some industries they’re the only way to fly
11:27:19 <peter1138[d]> andythenorth: Entirely not similar at all.
11:28:11 <kuhnovic> It's more of a right-tool-for-the-right-job kind of thing. And AI is often the wrong tool.
11:28:31 <andythenorth> you mean battery screwdrivers don't depend on aggregating unprecedented power to a small elite, using intellectual property theft?
11:28:46 <kuhnovic> So it an eletric screwdriver when trying to take a car tyre off. Or an impact driver when repairing a phone.
11:31:49 <LordAro> i'm pretty sure you're just saying words
11:35:23 <_zephyris> One for the toolbox, not a panacea
11:35:44 <_zephyris> But visceral refusal to touch it confuses me
11:51:26 <foodliker> andythenorth: i bought a makita impact driver and ive unleashed manmade horrors beyond my comprehension again ):
11:59:17 *** WormnestAndroid has quit IRC (Remote host closed the connection)
11:59:28 *** WormnestAndroid has joined #openttd
12:03:44 <peter1138[d]> Okay, turns out Doom is playable with trackball as long as the sensitivity is cranked up enough.
12:04:02 <peter1138[d]> DPI seems to be fixed on it.
12:05:08 <_glx_> Hmm logs don't really help, but it seems it crashed before even filling settings
12:10:02 <peter1138[d]> Main game settings, or script settings?
12:31:04 <_glx_> main settings, all values are 0
12:33:50 <_glx_> my guess would be it's trying to read corrupted openttd.cfg, but that should not cause crashes (anyway with stack trace it's just a guess)
12:36:48 <_glx_> weird naming of the crash dumps
12:45:03 <peter1138[d]> Corrupted tar file?
12:45:39 <LordAro> crashing in a std::string constructor is a bit odd
12:45:43 <LordAro> if i'm reading that correctly
12:45:53 <LordAro> oh wait, std::filesystem
12:45:59 <peter1138[d]> Probably just garbage in, garbage out.
12:46:11 <_glx_> it's crashing on std::string filename_base = std::filesystem::path(filename).filename().string();
12:46:15 <peter1138[d]> Might be worth trying to get a copy of whatever broken file has been downloaded.
12:47:07 <peter1138[d]> Hmm. oh, if it's there then it's not even heading it yet.
12:48:14 <peter1138[d]> .native() returns a reference.
12:48:32 <peter1138[d]> But it's from a temporary, so the object lifetime may be wrong.
12:49:01 <LordAro> chained calls always make me a bit nervous
12:49:02 <peter1138[d]> Hmm, that change is 3 weeks ago, is that before or after...
12:51:50 <_glx_> might be already fixed actually
12:54:59 <_glx_> same trace, probably duplicate
12:55:13 <peter1138[d]> The line numbers don't seem to match beta2.
12:56:01 <peter1138[d]> Version of OpenTTD
12:56:01 <peter1138[d]> > 15.0-beta 2 Windows 11
12:56:47 <LordAro> given the sheer number of crash logs, there could be a separate issue in 15.0-beta2 :p
12:56:49 <_glx_> "revision": "14.1" in crash log
12:57:37 <peter1138[d]> The json logs are all dated recently though.
12:57:48 <peter1138[d]> So yea, the version number in the issue is wrong.
13:08:13 <xarick> as a league table expert *cough*...
13:09:45 <_glx_> ha and VS used 14.1 from symbol server when opening the dmp so definitely no 15.0-beta2
13:10:51 <xarick> transfers are fake profits
13:11:26 <_glx_> transfer money exist to fix this exact issue 🙂
13:13:16 <xarick> the official league table isn't all that great with newgrfs
13:22:21 <talltyler> More of a problem is routes that take longer than the measurement period (a quarter? A year?) to complete a full round trip. I guess we could scale the measurement period by map size but that feels messy.
13:23:39 <peter1138[d]> Logically, if scaling by map size, the income requirements should be higher... 😉
13:38:09 <frosch123> It's not the job of the game to allow every player to reach maximum score. It's the job of the players to adjust/exploit the requirements of the score, if they care about scores.
13:46:04 <_zephyris> Profit calculation shared among shared/identical orders?
13:46:59 <talltyler> That’s a good idea 🙂
13:48:05 <talltyler> Like group profits, although I don’t think those are calculated automatically for shared order “groups” since they’re not really groups
14:04:40 <peter1138[d]> My OrderLists-own-Orders branch is part of my first step to making Order Lists be less vehicle centric.
14:05:27 <peter1138[d]> In theory OrderList could have a profit history as well, just like groups.
14:12:47 <peter1138[d]> (It also simplifies storage and iteration of orders, mind you.)
14:13:04 <peter1138[d]> ((I say "my" but it's based on similar changes made in JGRPP))
14:42:03 *** tateisukannanirase has quit IRC (Quit: User went offline on Discord a while ago)
15:12:44 <peter1138[d]> Well, not anything relevant to the league table.
15:25:55 <yiffgirl> peter1138[d]: oooh, this sounds interesting
15:29:24 <peter1138[d]> Hmm, comparing some thing withs how JGRPP does it and...
15:31:07 <_jgr_> Apologies if my messy code offends at times 😛
15:41:09 *** gelignite has joined #openttd
15:52:51 *** Wormnest has joined #openttd
15:56:58 <xarick> group vehicle lists when?
15:58:36 <xarick> just kidding, I don't know what's eating performance yet
15:59:28 <xarick> the other day I suspected something Autoreplace related
15:59:57 <peter1138[d]> Sheer quantity, normally.
16:05:53 <xarick> I try to avoid an individual AI to reach the red
16:06:12 <xarick> but my recent additions are reaching that point 😦
16:07:27 <xarick> i made tons of changes recently indeed
16:10:11 <_jgr_> peter1138[d]: Which part of the code are you looking at, incidentally?
16:12:24 <peter1138[d]> SpriteCache zoom level changes.
16:13:10 <_glx_> xarick: easy, do less things per tick
16:13:23 <peter1138[d]> It's 1) quite different from the "simplify sprite cache" PR, 2) possibly similar to what I want for additional non-power-of-two sprite scaling.
16:14:28 <peter1138[d]> If your max opcodes is not 10,000 then any excess CPU usage is your own call.
16:14:56 <peter1138[d]> Throwing all the CPU time at it, and then wondering why it's using all the CPU time, is a bit weird.
16:15:29 <peter1138[d]> But I suppose optimising things if it's a sensible optimisation makes sense.
16:16:45 <peter1138[d]> (What I really want, I guess, is for zoomed in sprites to be scaled on the fly instead of upfront, but that's probably slow... which is why we scale upfront.)
16:28:12 <xarick> ScriptVehicleList_Station continues to be on top
16:35:19 <_glx_> how often are you recreating the lists ?
16:39:58 <xarick> everytime I am managing a route, i create a list for that route
16:40:04 <xarick> and there's tons of routes
16:40:38 <xarick> I was using manual handling of vehicles before
16:41:37 <xarick> better for the cpu I guess, but it was making the AI less competitive by doing manual care
16:42:12 <xarick> hmm yeah, it's autoreplace related 🙂
16:46:30 <xarick> scoping to just autoreplace
16:46:46 <xarick> GetGroupNumEngines is called very often
16:47:10 <xarick> let me check how many groups
16:49:41 <xarick> _enginerenew_pool.items 9439 unsigned __int64
16:49:54 <xarick> 9439 enginerenew entries
17:13:24 <xarick> Train::GetCurrentMaxSpeed also shows up often
17:14:16 <peter1138[d]> Compiling 901 files ;D
17:19:05 <xarick> constantly checking acceleration model for each train part seems redundant indeed
17:20:17 <xarick> it's mostly the iterator that's slowing down
17:20:33 <xarick> `for (const Train *u = this; u != nullptr; u = u->Next()) {`
17:28:32 <xarick> Vehicle::ShowVisualEffect but for trains is the one mass calling Train::GetCurrentMaxSpeed
17:31:26 <xarick> "one of" the slow parts
17:43:38 <peter1138[d]> xarick: One for you to test./
17:44:13 <xarick> nice, but dinner is coming
17:45:13 <peter1138[d]> Not if you drop it on the floor.
17:58:47 <peter1138[d]> I don't have any group-heavy games.
18:00:02 <yiffgirl> i have one, for a certain value of group-heavy
18:00:57 <peter1138[d]> AAAHogEx likes to use groups, but doesn't nest them.
18:01:58 <peter1138[d]> Although even without children, the game has to iterate the pool just to know that there are no children.
18:31:43 *** akimoto has joined #openttd
18:34:23 *** nielsm has quit IRC (Remote host closed the connection)
18:40:57 <xarick> hmm where would I place the TicTocs
18:49:43 <xarick> wow GetGroupNumEngines is no longer costing cpu
18:51:31 <xarick> gonna get a longer sample
18:53:12 <xarick> Master branch vehicle tile hash is only 7 bits, I was running my tests on 11 bits. It is generating different cpu usage statistics
19:04:51 <xarick> vehicle tile hash seems bad
19:05:08 <xarick> the SkipFalseMatches one
19:06:54 <_glx_> Switch back to pool iteration to reduce SkipFalseMatches calls 🙂
19:11:20 <xarick> completely different statistics
19:11:42 <xarick> GetGroupNumVehicles is at least no longer around
19:12:53 <xarick> yeah, it's way further down the list
19:13:44 <xarick> how do I switch to pool iteration?
19:31:43 <xarick> it's funny that ships are now the least cpu intensive
19:32:16 <xarick> oh, the AIs didn't make many ships actually
21:24:15 <peter1138[d]> I mean, you do realise that **something** has to be using CPU...
21:25:04 <peter1138[d]> If something that is costly is improved, then everything else will scale up to take its place. It's not possible for nothing to be happening...
21:28:45 *** keikoz has quit IRC (Ping timeout: 480 seconds)
21:37:38 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
21:50:54 *** akimoto has quit IRC (Remote host closed the connection)
22:03:33 <DorpsGek> [OpenTTD/OpenTTD] PeterN closed issue #14271: [Crash]: Assertion failure when displaying a set order speed limit with an unset travel time in the timetable window due to an incorrect string parameter offset in STR_TIMETABLE_TRAVEL_NOT_TIMETABLED_SPEED https://github.com/OpenTTD/OpenTTD/issues/14271
23:47:28 <peter1138[d]> xarick: Fancy sending me that savegame?
continue to next day ⏵
