IRC logs for #openttd on OFTC at 2024-05-13
⏴ go to previous day
01:28:14 <reldred> swap to the new teams they said, uses 40% less ram they said
01:28:43 <reldred> using 550MB of ram idling on a brand new user and a VERY quiet tenancy.
01:29:31 <reldred> computers are a mistake
01:45:03 <johnfranklin> The fact is that "Microsoft is bad, but you have to choose, because others are worse"
02:06:27 *** Wormnest has quit IRC (Quit: Leaving)
02:15:07 *** tokai|noir has joined #openttd
02:15:07 *** ChanServ sets mode: +v tokai|noir
02:22:19 *** tokai has quit IRC (Ping timeout: 480 seconds)
02:51:41 *** debdog has quit IRC (Ping timeout: 480 seconds)
03:30:21 <wensimehrp> reldred: I always have problems when logging in 😒
04:41:30 <DorpsGek> - Update: Translations from eints (by translators)
04:58:28 <merni> emperorjake: It was correct during the beta/RC phase :P
05:03:17 <emperorjake> speaking of social integration, it's now saying this
05:03:42 <emperorjake> and I'm pretty sure discord is running
05:11:30 <merni> Hm how do I create a new page on the wiki
05:16:54 <merni> Aha, if I navigate to a page that doesn't exist, the Create Page button shows up
05:24:26 *** keikoz has quit IRC (Ping timeout: 480 seconds)
06:12:36 *** SigHunter_ has joined #openttd
06:14:36 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
06:28:36 *** SigHunter_ has quit IRC (Ping timeout: 480 seconds)
07:13:12 <DorpsGek> - Add: summary for week 19 of 2024 (by OpenTTD Survey)
07:16:42 <kuhnovic> I guess Merni is happy that the PF usage didn't change all of a sudden 😉
07:18:33 <merni> I am actually looking at something completely different now, namely town-name generators 😆
07:19:38 <merni> Oh no 0.2% of people are using npf for ships in 14.1
07:19:42 <merni> What a tremendous increase
07:21:54 <merni> Ok, so why do towns only store the town name generator and the seed rather than the actual string name?
07:24:16 <merni> Is it purely "this is how it's been done so far" or is there some advantage to doing it that way?
07:27:17 *** SigHunter has joined #openttd
07:28:44 <kuhnovic> Technically speaking, saving two integers uses less memory compared to storing a string. But nowadays that's much less of a concern.
07:34:05 *** SigHunter_ has joined #openttd
07:34:55 <merni> If a town name generator can't find enough names, is it sensible to fallback to English (Original) in all cases? Or is it better to have some sort of customisable fallback mechanism, or append numbers to the names, or something else?
07:36:46 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
07:39:16 <LordAro> merni: yeah, it's a thing inherited from TTD iirc
07:39:46 <LordAro> and we've been saying "yeah this is a dumb way of doing it" for almost 20 years
07:47:31 <merni> I see, so that at least should be a somewhat less controversial change :P
07:48:17 <merni> Will open up the possibility to make changes to, or remove, the default generators without affecting compatibility
07:54:36 *** SigHunter_ has quit IRC (Ping timeout: 480 seconds)
07:57:31 *** SigHunter has joined #openttd
08:04:49 <peter1138> Well it will iif you load a game which doesn't have them saved yet.
08:05:58 <jfs> yeah I've considered before do make a patch for, all current built-in generators become "legacy generators", and all towns only store their textual name. on savegame load, all town names are converted to strings via the legacy generators. if someone wants to modify a town name generator, they can copy the legacy one and modify the copy.
08:07:16 <merni> That is basically what I'm thinking of
08:09:28 <peter1138> One thing I've seen people want to do is use multiple generators.
08:10:02 <merni> Also thought about that but one step at a time :p
08:10:24 <merni> It'd be nice to have an option to choose a generator when creating a town, and also to select multiple generators for a new map
08:11:05 <merni> Though presumably it'd choose at random between the generators, and people might say "I want all the English names on the west and the French names on the east" or something like that
08:14:20 <kuhnovic> The rabbit hole is getting deeper
08:16:28 <jfs> the annoying part is having to have a chunk of code in the game that just has to sit there forever, because of compatibility
08:16:44 <merni> saveload code is full of that kind of thing anyway
08:17:07 <merni> problems with guaranteeing compatibility for the last 29 years :P
08:17:26 <jfs> true, the afterload fixups is basically a big black box of "someone figured this works at one point, now never change it"
08:19:54 <andythenorth> how do town names work?
08:20:02 <andythenorth> there's some selectable generators, and then there's grfs?
08:21:17 <andythenorth> I have Irish Town Names grf in my game, but these seem like some of the original game names?
08:21:28 <andythenorth> I can't figure out how to enable the town grf
08:21:40 <peter1138> Add the town new GRF in your config.
08:21:56 <peter1138> Then when using New Game, pick it from the Town names list.
08:22:57 <peter1138> merni: if you add support for multiple generators then that dropdown can be disabled if newgrf town names are present, making it much easier for andy to use 😉
08:23:30 <peter1138> (Assumption is that if you add newgf name generators, you want to use them)
08:23:42 <peter1138> No doubt breaking some work flow 😄
08:25:57 <andythenorth> I did assume that adding a grf...makes it active in game
08:26:19 <andythenorth> why aren't town names GS?
08:26:39 <merni> peter1138: I think better would be to add all available newgrf town name sets to that dropdown and allow you to tick a mixture of default and grf generators to use...?
08:26:49 <andythenorth> I'm actually serious, GSTown.SetName() exists
08:26:57 <merni> Probably harder though code wise
08:27:06 <merni> andythenorth: Because they predate GS probably
08:27:17 <andythenorth> so that's the cause of how we got to here 😛
08:27:20 <andythenorth> but not where we could go
08:27:31 <peter1138> Then you have to scan ALL newgrfs, even unloaded ones, to find the names.
08:27:39 <jfs> how would multiple generators work? first pick a seed and then find a generator that makes a valid name for it? or first pick a generator and then try to find a useful seed?
08:28:05 <merni> jfs: I'm initially thinking the latter
08:28:05 <peter1138> And then you have to force-load the NewGRF with the one you select. And that NewGRF might have other content in it.
08:28:24 <merni> Do people bundle town names with other things in one grf???
08:28:42 <peter1138> Latter, I would think. Pick random active generator, and then use it with a seed.
08:29:32 <peter1138> I don't know but it's perfectly possible.
08:30:00 <merni> All right, even if you restrict to grfs that are selected for that game, I think it's still a good idea to allow mixing grf and original generators
08:30:27 <peter1138> Hmm, when I ask fontconfig to include family and style name information, the fallback font list comes up with a different 'best font' :S
08:30:31 <merni> Of course, that could be solved by releasing the original generators as grfs, since the code is GPL anyway...
08:30:47 <merni> And removing the built in generators except for compatibility
08:31:00 <peter1138> Then you need to maintain the code that generates thes the NewGRFs 😉
08:31:21 <merni> No, just convert them into grfs once and for all and remove them from the game code
08:31:33 <merni> Except for the legacy savegame-compatibility ones
08:31:37 <andythenorth> I was still serious about GS
08:31:41 <peter1138> Also there are some special cases that can't be down with NewGRF town names, I believe.
08:31:55 <andythenorth> GS has a lot more API capability, and there are a lot of free opcodes at mapgen
08:32:03 <peter1138> Town Names can be set by GS.
08:32:12 <merni> GS means only one at a time
08:32:12 <peter1138> But requiring a GS just to have Town Names is not reasonable.
08:32:15 <peter1138> What are you thinking/!
08:32:22 <andythenorth> merni: this is a myth 🙂
08:32:27 <andythenorth> GS libs don't have this problem
08:32:50 <merni> Also we have already a bunch of town name grfs
08:33:04 <merni> So maintaining 3 different town name mechanisms may be a bit too much
08:33:12 <andythenorth> merni: yes, this is hard to argue against 🙂
08:33:24 <peter1138> I want Outer Bongolia town names please, there's a NewGRF for it already. Nah, you can't use that, make your own Gamescript.
08:33:46 <andythenorth> automated grf -> GS converter
08:34:30 <jfs> how about then introducing the MS, Map Script, the same API as GS, but instead it gets to execute during mapgen, and you can add multiple
08:35:21 <jfs> a script could supply functions called to generate landscape, trees, towns, industries
08:35:40 <peter1138> I don't think NewGRF generators can do this.
08:35:59 <peter1138> I suspect most NewGRF 'generators' are just a list of names, though 🙂
08:36:51 <andythenorth> jfs: I suspect my GS suggestion is a trojan horse for whatever WASM world eventually arrives
08:38:09 <peter1138> It's annoying that the TGP noise generator is so fast.
08:38:33 <peter1138> A regular perlin generator has more flexibility in how it's used, but is also much slower.
08:38:47 *** asasnatirc has joined #openttd
08:39:03 <peter1138> I have a branch where I created some building-block functions for mapgen.
08:39:16 <peter1138> And I made a multi-threaded perlin generator
08:39:22 <peter1138> It was still slower than TGP.
08:39:34 <peter1138> Much more flexible though.
08:41:08 <andythenorth> "just limit map size"
08:41:19 <peter1138> Yes, 256x256 should be enough 😄
08:42:01 <peter1138> Hmm, any idea why the "missing glyphs for your language" error message is hardcoded instead of using the string system?
08:42:10 <jfs> (when did I last work on that, I wonder...)
08:42:20 <peter1138> My thought was it's to stop it being translated, and then the translation trying to use those missing glyphs...
08:42:49 <asasnatirc> I recently learned tha variety distribution actually relates to the terrain generator
08:43:02 <asasnatirc> I always thought it controlled the amount of decorative objects (lighthouses, transmissino towers)
08:43:37 <peter1138> Right there, in the middle of the terrain settings...
08:43:47 <andythenorth> isn't there only one useful value or something?
08:43:49 <peter1138> It is a piece of shit, mind you 🙂
08:44:02 <asasnatirc> nveer seen that tooltip
08:44:11 <asasnatirc> sorry my laptop keyboard sucks
08:44:49 <peter1138> Variety Distribution usually means "lots of flat"
08:45:02 <peter1138> Because it can only make mountainous areas flag, it can't make flat areas mountainous.
08:45:39 <peter1138> It's flawed, and mostly the reason why I was working on mapgen with building block pieces.
08:45:44 <asasnatirc> in a certain way I prefer the original generator anyway
08:45:55 <peter1138> Yeah but that got hidden
08:45:55 <asasnatirc> especially with how it works on arctic and tropical maps
08:46:17 <asasnatirc> didn't use it in a serious game yet though
08:46:34 <peter1138> Heh, that really only works well at normal map sizes.
08:47:40 <jfs> mnaybe the way to make it interesting on different sized maps would be to have various templates to apply to the overall map shape
08:49:14 <peter1138> "map shape" requires a different set of functions indeed.
08:49:43 <peter1138> variety distribution is basiclally elevation templates applied in chunks.
08:50:52 <peter1138> In procedural generation you get map shape by affecting elevation by position.
08:52:35 <peter1138> "amount of water" should, in the end, should equate to "large continents" or "small islands" etc.
08:52:50 <peter1138> We just flood the terrain 🙂
08:52:54 *** asasnati1c has joined #openttd
08:53:11 <peter1138> In theory that's is selected by "smoothness"
08:53:54 <peter1138> But that ends up with smooth round (squareish-round) shapes instead of useful features.
08:54:02 <jfs> I was thinking of having maybe general shapes like "Carribean", "Mediterranean", "Continent" (implied shaped by a large mountain range down through it)
08:54:19 *** asasnatirc has quit IRC (Ping timeout: 480 seconds)
08:54:45 <peter1138> So in my mapgen branch, the generator has the ability to make multiple perlin maps. None of them are drectly the landscape elevation, but they get combined in various ways.
08:55:44 <peter1138> God I am way too easily sidetracked.
08:56:03 <peter1138> I still have UI consistency changes to apply....
09:07:26 *** asasnati1c has quit IRC (Quit: leaving)
09:18:31 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
09:20:51 <locosage> fft noise is pretty fast and gives two maps at once
09:21:07 <locosage> though tiling it for chunks would probably be quite hard
09:27:32 <peter1138> `This method of producing noise is not the most efficient. In Fact I do not recommend using this in a production environment.`
09:28:25 <andythenorth> now reading (a) why Perlin noise is bad (b) why everyone has to use it anyway
09:28:33 <andythenorth> "because it's cheap"
09:29:10 <peter1138> Yeah, Voronoi is another one.
09:30:05 <peter1138> Years ago I was playing with a browser based one that would generate instantly...
09:30:05 <peter1138> In fact it was that one.
09:30:50 <andythenorth> it's not what we need, but eh
09:31:25 <locosage> peter1138: that's just some random article, I use it on citymania and it's sufficiently fast even with all the river stuff on top
09:31:50 <peter1138> It is just some random article, it's the one that crops up when I search for "fft noise"
09:37:21 <peter1138> Funnily enough a search for "citymania fft noise" does not yield any results.
09:44:16 <locosage> it doesn't even seem to have an established name afaict
09:44:26 <locosage> power law noise, 1/f noise, all the same thing
09:49:33 *** SigHunter has joined #openttd
10:08:16 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
10:10:48 *** SigHunter has joined #openttd
10:53:18 <peter1138> When you refactor something and then realise it's not used...
11:38:02 <johnfranklin> Today is the 15th Anniversary of Minecraft
11:43:20 *** SigHunter has quit IRC (Remote host closed the connection)
11:44:00 <dwfreed> damn, I remember playing that in college
11:46:15 *** SigHunter has joined #openttd
11:50:11 <johnfranklin> I am young, and it is my child memory
11:58:47 *** D-HUND is now known as debdog
13:57:56 *** gelignite has joined #openttd
14:11:04 <johnfranklin> Finally! I purchased Minecraft on its 15th anniversary, ending the pirate life for nearly 10 years.
14:33:15 <peter1138> I remember when I used to play Minecraft.
14:35:29 *** gelignite has quit IRC (Quit: Stay safe!)
15:09:10 <peter1138> Hmm, my Windows version of OpenTTD does not use OpenTTD Sans.
15:17:41 *** APTX has quit IRC (Quit: Farewell)
15:53:28 <peter1138> What mac hardware do I need to start testing mac OS builds? 😄
15:54:00 <peter1138> I assume anything for a reason price on CEX is going to be 1) stolen 2) too old for the required version of Mac OS.
15:55:31 <truebrain> M1 or higher I would think, but no actual clue 😛
15:56:09 <alfagamma7> Or does it not work on virtual machines?
16:02:39 <andythenorth> can't we rent a mac in the cloud? 😛
16:03:28 <andythenorth> peter1138: child #1 had the 1st gen M1 Macbook Air, which hands down beat my i9 for both OpenTTD performance and compile times
16:03:35 <andythenorth> it was £999 new, not sure how much they go used
16:03:58 <andythenorth> he had to swap it for a Macbook Pro with a fan though, because the Minecraft server ran the thermals too hot
16:05:03 <peter1138> Expensive version of a Raspberry Pi, innit...
16:05:13 <andythenorth> most of the RAM use on a mac is just Chrome
16:05:23 <LordAro> "macs use memory more efficiently than other processors"
16:05:26 <LordAro> or some other nonsense
16:05:37 <alfagamma7> Firefox eats a lot of RAM sometimes as well
16:05:43 *** urdh has quit IRC (Quit: Boom!)
16:10:12 <andythenorth> alfagamma7: nobody uses FF on a mac 😛
16:10:18 <andythenorth> they don't have enough RAM for that
16:21:14 <emperorjake> andythenorth: I do
16:21:29 <emperorjake> but I got the 16gb upgrade
16:23:06 <peter1138> Upgrade or factory option...
16:25:24 *** gelignite has joined #openttd
16:25:44 <peter1138> £200 for 8GiB extra... jaysus
16:26:09 <peter1138> Buying new is too much just for testing builds 🙂
16:36:58 *** Wormnest has joined #openttd
16:38:11 *** Smedles has joined #openttd
16:38:27 <andythenorth> emperorjake: same and same
16:47:02 *** HerzogDeXtEr has joined #openttd
16:49:30 <andythenorth> oof I have an i9 I should sell
16:49:59 <andythenorth> kept it as a spare because availability in the pandemic was patchy
16:52:23 <peter1138> "This works, but how?"
17:16:13 <peter1138> I assume OpenTTD Sans does work out of the box with Windows, and it's just my mingw/wine that's being a pain...?
17:18:29 <audigex> peter1138: I'd say 2020 Mac Mini is gonna be the cheapest option that will be compatible for a few more years, maybe the 2018 version
17:18:29 <audigex> 2014 already doesn't have the current version of OS X, though, and I'd say 2018 probably doesn't have long either being X86 rather than Apple Silicon
17:19:13 <audigex> Only problem with the Mac Mini is they're not very popular so there aren't enough used/refurbed ones around to keep the price down
17:37:41 <talltyler> peter1138: I use it on Windows 10
17:38:11 <peter1138> And it just works without having to configure it manually I hope
17:38:32 <peter1138> On Wine it won't find the font 😦
17:39:11 <_glx_> could be the way we register it
17:39:38 <peter1138> We don't register it, we try to load it directly.
17:39:48 <peter1138> But Wine just loads Arial instead.
17:40:59 <_glx_> we use AddFontResourceEx() to register it, as we do for any other font files
17:42:36 <_glx_> then we use an undocumented function
17:44:45 <_glx_> loaded dynamically so can be unavailable
17:47:30 <_glx_> GetFontResourceInfo (stub)
17:48:10 <_glx_> and GetFontResourceInfoW (not documented)
17:48:29 <_glx_> so it should be present I guess
17:49:54 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
17:51:44 <_glx_> does it fail to find small and large too ?
17:53:14 <peter1138> yeah, arial for all
17:54:08 <_glx_> you'll need to add debug message to trace it
17:57:40 *** SigHunter has joined #openttd
18:02:50 <peter1138> It takes 7 minutes to link :p
18:06:23 <peter1138> I'm not sure how to list unicode points above 0xffff :S
18:07:22 <peter1138> Using FreeType and FontConfig on Windows and mac OS would make things so much simpler :S
18:07:54 <peter1138> Such dependencies 🙂
18:37:42 *** gelignite has quit IRC (Quit: Stay safe!)
19:04:32 <peter1138> > This is similar to GDI's GetFontUnicodeRanges, except that it returns the full Unicode range, not just 16-bit UCS-2.
19:04:45 <peter1138> Well. So it's just not possible with GDI.
19:12:26 <peter1138> DirectWrite is the moderner way.
19:20:59 <_glx_> not like macos APIs being deprecated every major release
19:21:08 <peter1138> In this case I'm trying to improve how fallback fonts work.
19:21:31 *** gelignite has joined #openttd
19:21:35 <_glx_> there should be more modern stuff
19:21:57 <peter1138> But that's a whole separate thing.
19:26:13 <peter1138> Hmm, alternatively, I can read the font table directly...
19:26:23 <peter1138> (Kinda undoing #12677 :))
19:28:11 <peter1138> > Fonts that support Unicode supplementary-plane characters (U+10000 to U+10FFFF) on the Windows platform must use encoding 10 with a format 12 subtable.
19:30:44 <_glx_> ah it's possible to convert IDWriteFont to LOGFONT
19:32:58 <peter1138> I vote for dropping all non Linux support. Users can dual boot 🙂
19:33:43 <Rubidium> peter1138: seconded ;)
19:35:55 *** gelignite has quit IRC (Quit: Stay safe!)
19:39:51 <peter1138> I guess what I could do is "brute force" check
19:40:16 <peter1138> Instead of asking what glyphs are supported, ask if each specific required glyph is supported.
19:41:08 <peter1138> We already know which glyphs are required so that it's not too much drama.
19:42:25 <peter1138> Can perhaps use a combo, only testing surrogate pairs and using GetFontUnicodeRanges for the BMP.
20:17:35 <peter1138> Hmm, that might a rabbit hole...
20:26:16 <peter1138> Hmm, okay, which NewGRFs were using special sprites to do custom 'graphics' in text windows?
20:40:47 <talltyler> That’s the only know I can think of that replaces sprites. Others use the existing train and bus characters to signify certain things in roadtype names, and I recall someone using them in buy menu text too.
20:42:08 <peter1138> Someone had one that had "components" of engines that were put together to draw a consist.
20:43:09 <talltyler> Hungary Voxel World, perhaps?
20:44:38 <wensimehrp> Is it possible to assign more icons in the private zone and allow newgrf strings to access the them? E.g. cargo icons, road icons, the warning icon, etc. Not just the vehicle icons.
20:50:33 <truebrain> Rubidium: the whole searchpath system is hacky and fragile; no need to focus that on a single element of it 😉
20:50:45 <truebrain> and if you search for WORKING_DIR, you will see the exceptions .. they are weird, but nobody dares to touch them
20:58:00 <peter1138> Something went horribly wrong.
20:58:48 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
20:59:39 *** nielsm has quit IRC (Remote host closed the connection)
21:15:14 <peter1138> Ah, too much removed 🙂
21:20:25 <truebrain> Rubidum: I was talking about #11431. That change makes SP_WORKING_DIR always lose in case it is a duplication of any of the other SPs. It is not that SP_BINARY_DIR wins. I was just adding context for you to understand what is going on.
21:28:04 <truebrain> Rubidium: I once fiddled with the idea of making 2 searchpaths, one for reading, and one for saving
21:28:11 <truebrain> that would solve more of those weird corner cases we have
21:30:08 <Rubidium> I wonder whether just adding the ' && !_do_scan_working_directory' to the check in FillValidSearchPaths does the right thing. It's not obviously broken for me
21:33:41 <truebrain> it is so complicated to understand how the searchpath interacts with everything 😦
21:34:31 *** keikoz has quit IRC (Ping timeout: 480 seconds)
21:41:33 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:44:46 <talltyler> That’s a funny PR title 🙂
22:15:22 <peter1138> I mean, it seems logical.
22:19:40 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:25:16 <peter1138> Preview loads, is that good enough? 😄
22:26:33 <peter1138> Btw that "useless" line that glx commented on last time... is no longer useless.
22:28:07 <peter1138> Not sure what the symptoms before were though.
22:28:27 <peter1138> Did it preview fail to start or was it something more subtle?
22:30:34 <truebrain> It failed to load anything, if I remember correctly
23:05:14 <peter1138> So "hopefully fixing" means we don't know?
continue to next day ⏵
