IRC logs for #openttd on OFTC at 2024-02-22
โด go to previous day
00:01:14 <_jgr_> It's not a real command, it's just an alias
00:01:25 <_jgr_> You can use `list_aliases` to show those
00:05:59 <debdog> oh, I see. thanks _jgr_
00:06:50 <debdog> hmm, mayhap that should be mentioned in the output of the "help" command?
00:11:56 <debdog> btw. _jgr_, I have seen a (new?) branch diagonal_road_investigations in your repo. is this somewhat working, worth checking out atm?
00:13:18 <reldred> that's my grotesque mess @debdog
00:13:33 <reldred> currently broken, articulated vehicles break it.
00:13:50 <debdog> yah, I thought that might be the biggest issue
00:14:11 <debdog> ok, thanks for the info!
00:14:58 <reldred> I haven't looked at it in a while, but happy to collab. Disclaimer: I am not a C++ developer. I do not know C++. Any C++ that I write is purely by accident or inspired by the voices in my head.
00:15:32 <debdog> I've tried but gave up on C$anything years ago
00:15:58 <debdog> object oriented languges are just no my... what's the word...
00:18:32 <peter1138> reldred: Welcome to C++ developing ๐
00:20:02 *** johnfranklin has joined #openttd
00:20:02 <johnfranklin> I have read about two chapters of C++ tutorial and.. only programmed โhello worldโ. ๐
00:26:34 <_glx_> andythenorth: I think they can, but you need to do the magic yourself
00:40:07 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
00:46:33 *** Leopold_ has joined #openttd
01:06:32 <belajalilija> andythenorth: You mean so that hiding the last built one wouldnโt hide the whole list and youโd need to hide all the vehicles to hide the list?
01:12:03 *** Wormnest has joined #openttd
03:03:36 *** Wormnest has quit IRC (Quit: Leaving)
03:09:35 *** gnu_jj_ has quit IRC (Ping timeout: 480 seconds)
03:17:02 *** herms has quit IRC (Quit: Ping timeout (120 seconds))
04:01:05 *** debdog has quit IRC (Ping timeout: 480 seconds)
04:03:05 *** gnu_jj_ has joined #openttd
04:06:15 *** gnu_jj has quit IRC (Ping timeout: 480 seconds)
05:37:45 *** ajmiles has joined #openttd
05:37:45 <ajmiles> _glados_: I know it does look like much, but this is the GPU 'blitting' GfxFillRect instead of the CPU
05:39:38 <ajmiles> I'll never get back the hair I pulled out trying to get everything pixel perfect. Of course nothing is textured yet, which is going to be the next bit I imagine
06:02:27 *** _glados_ has joined #openttd
06:02:27 <_glados_> Definitely better that me i think I have perfectionist paralysis of overthinking and engineering the code instead of just writing it
06:03:37 <ajmiles> I think I've just noticed that Sprites in OpenTTD don't follow the usual ">> 1" size calculation as you go down the mips. I can see a sprite that has a width of 47 and then the next zoom out has a width of 24.
06:03:44 <_glados_> ajmiles: Whatโs your branch be cool to see how you have taken it compared to myself
06:04:26 <ajmiles> I've not pushed anything back up to a Github yet as I wasn't sure if I'd stick at it long enough to get something on screen, but a little more progress and I'll definitely do that
06:05:55 <_glados_> So how is the gpu rendering working with gl or is dx12 only?
06:06:15 <ajmiles> Just focusing on 12 for now. Though I've not done anything yet that wouldn't work with GL
06:06:38 <ajmiles> I don't know OpenGL, that would be the problem with me doing GL ๐
06:07:08 <ajmiles> I understood enough of the GL renderer to figure out how to mirror it in D3D12, but it's a learning experience
06:08:19 <_glados_> Still thinking dropping cpu rendering would be best all computers have has dx12 or vulkan & gl support for years
06:08:58 <ajmiles> I expect the developers need/want it to run on a potato. The GDI backend is still there for a reason I imagine
06:09:37 <ajmiles> And besides, if it can be done in such a way that the existing backends continue to work then there's no reason to take it out
06:10:05 <_glados_> It would run better on a potato because they were leverage, the GPU to do the graphics cycles on the CPU for the simulation
06:11:43 <_glados_> It also opens the door for bigger maps and more complex ai
06:12:21 <_glados_> And if it doesnโt run the game art style transparency
07:14:08 <andythenorth> _glx_: maybe I should ry again ๐
07:15:29 <andythenorth> belajalilija: just hide the parent? ๐
07:21:44 <belajalilija> andythenorth: That too
07:22:25 <belajalilija> What i was trying to say specifically is that hiding last built shouldnโt hide the list
07:26:01 <michi_cc[d]> _glados_: Obviously nobody has measured it, but the way the renders right now is basically lots of draw calls that do almost nothing. But to get anywhere close to maximum performance on a GPU, you need few draw calls that do lots.
07:26:11 *** ufo-piloot_ has joined #openttd
07:26:26 <michi_cc[d]> So basically youโd have to rewrite more or less all of the drawing code.
07:27:14 <michi_cc[d]> And cope with stuff like NewGRFs dynamically choosing what to draw each tick, and thatโs something that canโt practically go to the GPU.
07:28:54 <locosage> where do I check for updates for original graphics? ๐คญ
07:29:15 *** ufo-piloot has quit IRC (Ping timeout: 480 seconds)
07:30:35 <emperorjake> Shouldn't that message be disabled when loading TTD graphics?
07:48:47 *** Leopold_ has joined #openttd
08:06:52 <LordAro> that message shouldn't ever happen, as ottd_extra provides the... extra stuff
08:13:01 <andythenorth> Horse nml was 58 MB last night, with a unique varact2 chain handling "name" for every vehicle variant
08:13:22 <andythenorth> unifying that varact 2 chain to just the parent vehicle cuts the nml to 43 MB
08:13:45 <andythenorth> and knocks 8 seconds off the compile time (34s -> 26s)
08:14:07 <andythenorth> the only problem is....it doesn't work correctly ๐
08:16:24 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
08:18:25 <peter1138> But it's faster so clearly better.
08:29:25 *** Leopold__ has joined #openttd
08:36:34 *** Leopold_ has quit IRC (Ping timeout: 480 seconds)
08:44:17 <andythenorth> Seems 25% of the nml filesize was just for name string handling
08:44:20 <andythenorth> seems like overkill
08:44:34 <andythenorth> but 'working as intended' is also useful, sometimes
09:20:55 <peter1138> So paint tool instead of andy making thousands of different colour variants?
09:25:44 <peter1138> To do George's variant hiding stuff means I need to rewrite how engine selection works ๐ฆ
09:56:32 <andythenorth> peter1138: can it also paint 'random from list'? ๐
09:56:52 <andythenorth> maybe we need skins
09:59:12 <andythenorth> quite enjoy the many variants, reminds me of paint colour charts ๐
09:59:37 <andythenorth> but then again I collect "things" in multiple colours, some obsessive personality defect P
09:59:56 <andythenorth> trainers, trains, lego, pens, mugs
10:04:26 <emperorjake> How about real-livery-subvariants for engines and passenger stock too? ๐
10:05:29 <andythenorth> variant subvariants?
10:05:41 <peter1138> andythenorth: Is this good for gameplay or a bad feature?
10:05:58 <andythenorth> we're finding out empirically
10:06:05 <andythenorth> I enjoy creating it ๐
10:07:00 <peter1138> Ah, the meta gameplay.
10:07:33 <peter1138> Hmm, what's today's music.
10:08:57 <reldred> there. That's todays music.
10:21:21 <andythenorth> so repaint to arbitrary colours, or from a pre-defined list?
10:24:19 <andythenorth> the Horse implementation has two tiers:
10:24:19 <andythenorth> - different sprites
10:24:19 <andythenorth> - remapping of the first company colour to a pre-defined palette
10:24:37 <andythenorth> and there's the option to randomise both tiers ๐
10:34:06 <andythenorth> variants meets the 'different sprites' case really well
10:34:14 <andythenorth> then I've nested all the colours another layer down
10:48:00 <peter1138> If you had all the 'different sprites' cases handled by variants, and then the colour stuff was done by the game, as a separate 'use these colours', that ... could work,
10:50:22 <peter1138> Have we `master` -> `main` yet?
10:52:27 <peter1138> (Oh, and it'll be 2CC)
11:04:58 <SigHunter> I have a savegame that heavily relies on the new unbunching feature and I tried it for busses, mail, banks, aircraft and passenger ships (hovercraft). unbunching is somehow not working for the hovercraft and I can't figure out why. the unbunching order for shipyard also says "unloading" instead of unbunching. I have not found an existing bugreport. is unbunching not working for ships?
11:11:54 <peter1138> Ship unbunching works for me, but I have simple orders.
11:12:22 <peter1138> You've made a video but not it's fast-forward, so hard to see anything, and you haven't included the order list.
11:13:00 <emperorjake> I've also noticed the unbunching not working right with fast ships
11:13:46 <SigHunter> yeah order list would have been nice, I'll redo
11:14:04 <emperorjake> peter1138: Exciting stuff, will there be a PR soon?
11:15:21 <peter1138> Probably not, no new features for 14.
11:15:46 <peter1138> Well, there might be a PR but it can't be merged for a whil.
11:16:10 <andythenorth> peter1138: that's pretty much my hope
11:16:41 <andythenorth> ideally the grf could return a set of 'recommended' or 'pre-computed' palettes to a painter that also has a 'no I want my choice' option
11:17:18 <andythenorth> not sure how randomising would work but eh
11:17:36 <peter1138> Custom remaps are not really feasible.
11:17:46 <peter1138> That's getting into refit-territory.
11:18:19 <peter1138> (Evaluating an unseeable list of palettes)
11:18:55 <peter1138> But I guess the CC remaps are not very nice compared to your custom remaps. Hmm.
11:21:03 <peter1138> Are these all remapping only the standard company colours?
11:23:37 <peter1138> Hmm, do you have different versions of the sprites too... ๐ฎ
11:24:35 <peter1138> Interesting interaction...
11:27:09 <peter1138> So.... not sure this is going to give what you need ๐ฆ
11:33:19 <peter1138> Also custom palettes means autoreplace needs to know what to do with them, what engines can they be applied to, etc...
11:33:20 <_glx_> Is it one sprite per colour, or just a remap and the same sprite?
11:34:02 <peter1138> remap and same sprite, I think.
12:04:03 <SigHunter> peter1138: you said "You've made a video but not it's fast-forward, so hard to see anything" but the video was fast-forward. did you mean to not make it fast forward?
12:05:22 <SigHunter> just that I get it right the second time
12:05:34 <peter1138> I can't always type sentences correctly ๐
12:13:51 <merni> peter1138: Yeah... I looked at that too. Not simple.
12:16:25 <merni> the idea that the group *is* the last-selected is a bit painful for that issue
12:28:32 <peter1138> No, I just to revert it back to select-by-row instead of select-by-engine-id.
12:31:34 <peter1138> "revert" used loosely here.
12:32:11 <peter1138> Or select by two conditions instead of one *shrug*
12:32:50 <xarick> the crash report on my AI is because on one tick, the engine existed and was valid, and in the next tick, it doesn't exist anymore
12:32:55 <xarick> what are the odds of that?
12:33:52 <merni> well, engines get retired quite frequently
12:35:57 <peter1138> You could select english to make it easier for us ๐
12:37:12 <peter1138> Is that an AI or human player?
12:37:13 <SigHunter> true but I also wanted to show off the wrong translation ;-)
12:37:25 <SigHunter> singleplayer without AI
12:37:38 <merni> is "warte auf entladen" supposed to be unbunch?
12:37:41 <SigHunter> all ships have shared orders
12:37:54 <peter1138> I don't think the translation is wrong, I think you've managed to get the wrong order flag on it.
12:37:57 <SigHunter> yep, but it says "waiting to unload" (not even possible in dock)
12:38:08 <SigHunter> not dock, i mean depot for ships
12:38:54 <SigHunter> when I switch game to english, it says "wait to unbunch" but german translation is "wait to unload"
12:39:21 <merni> Hm, there is a button "entpulken", what happens if you click that
12:40:10 <SigHunter> I'll record to show you, one sec
12:41:20 <merni> if I select "Entpulken" the string shown is "warte auf enlanden" in RC1
12:41:33 <_glx_> Seems you go to the depot in middle of the trip, shouldn't be an issue but who knows
12:42:22 <peter1138> Okay, that's explains the translation... it is already fixed.
12:42:37 <talltyler> โWait to unloadโ is definitely a translation mistake
12:42:58 <SigHunter> nice, so this is just wrong translation, I thought maybe in the background it also does something wrong which could explain the unbunching not working
12:43:15 <merni> Yeah, translation sh ould be fixed in the next version, already must be fixed in nightly
12:43:23 <merni> But unbunching should still work...
12:44:37 <SigHunter> I'll relocate the shipyard, see if that changes anything
12:45:23 <_glx_> Can you test without the extra stop in the depot?
12:45:36 <peter1138> Yeah, I added the depot again and unbunching goes weird.
12:45:42 <merni> It looks to me as though they are unbunched when they leave the depot going northward, but get bunched up again when they leave the depot going southward... might be just my eyes
12:48:15 <talltyler> Oh, I think I see the problem
12:49:11 <SigHunter> works nicely with just 1 depot order but breaks with 2 orders
12:49:14 <talltyler> The second depot is somehow resetting their unbunching data, I think
12:57:02 <peter1138> static const IndustryGfx NUM_INDUSTRYTILES_PER_GRF = 255; ///< Maximum number of industry tiles per NewGRF; limited to 255 to allow extending Action3 with an extended byte later on.
12:57:06 <peter1138> Action 3 extension eh?
13:00:15 <SigHunter> I also wanted to praise ottd version 14, it is awesome, best version ever, thanks for ALL your effort after all those years
13:00:38 <andythenorth> peter1138: is that the randomised version?
13:01:31 <peter1138> Randomised plus some colour other the company colour.
13:04:19 <peter1138> Also with rgbcc that drop down list UI becomes awkward.
13:04:33 <peter1138> But I'm not sure on the future of that.
13:04:55 <peter1138> "Doesn't completely rewrite how recolour works"
13:09:55 <LordAro> peter1138: much lolz.
13:14:46 <peter1138> Also, nobody else wanted to do MTB last night, so I went to the pub instead.
13:23:29 <andythenorth> peter1138: wonder what we actually want
13:29:20 <ajmiles> michi_cc[d]: Right, I'm hoping that I can put together one draw call (or just a handful) to rule them all. One that has access to all Sprites and a buffer of blit requests that get translated into geometry on the GPU
13:32:20 <peter1138> [2024-02-22 13:32:11] dbg: [sprite:2] LoadNewGRF: Currently 128533 sprites are loaded
13:33:26 <peter1138> andythenorth: how does Iron Horse still have so many sprites? ๐
13:36:41 <peter1138> [2024-02-22 13:36:24] dbg: [sprite:2] LoadNewGRF: Currently 299094 sprites are loaded
13:36:41 <peter1138> [2024-02-22 13:36:26] dbg: [sprite:2] LoadNewGRF: Currently 415212 sprites are loaded
13:36:45 <peter1138> That's a big jump some how.
13:37:31 <peter1138> Also that lies, they are not all loaded, just referenced.
13:39:04 <andythenorth> guess I drew a lot of sprites ๐
13:39:30 <andythenorth> I should run the duplicate counter
13:40:18 <peter1138> iron-horse.grf has 422432 sprites, maxx 65, maxy 51, maxs 2866.
13:40:22 <peter1138> Where "sprites" means NFO lines.
13:40:34 <peter1138> (but does include action 1 sprites)
13:40:55 <andythenorth> onluy 1000 duplicates
13:41:25 <andythenorth> will that count varact 2 also?
13:41:51 <peter1138> Yes, everything that's an nfo line.
13:42:22 <andythenorth> I didn't compare the nfo, but 25% of the nml is just handling vehicle names
13:43:44 <andythenorth> 72 MB nfo file when every vehicle has a varact 2 chain for name callback
13:45:02 <andythenorth> 59 MB nfo file when all variants share the same varact 2 chain for name callback (but it doesn't work right)
13:45:17 <andythenorth> that's quite lol
13:45:59 <andythenorth> the name callback results are broadly the same as well, for each class of vehicle
13:46:11 <andythenorth> they only vary in a handful of the cases
13:47:32 <peter1138> > [2024-02-22 13:47:08] dbg: [sprite:2] LoadNewGRF: Currently 499712 sprites are loaded
13:47:42 <peter1138> Nice, half a million sprites available.
13:48:27 <andythenorth> all hand drawn ๐
13:48:28 <ajmiles> And by "loaded" here, we mean, actually in memory and could be drawn at a moment's notice?
13:48:54 <peter1138> No, just a reference to where the sprite lives on disk.
13:49:00 <peter1138> (File and position in the file)
13:49:14 <ajmiles> Ah, ok. Is there a metric for "truly loaded"?
13:49:17 <peter1138> So they could be loaded and drawn at a moment's notice.
13:50:38 <peter1138> There's variables holding that value but no debug level that outputs it.
13:51:54 <peter1138> Hmm, actually not sure there is, other than iterating all 499712 spritecache entries.
13:52:41 <ajmiles> How safe to assume is it that most of those sprites are <= 64x64 in size, even if there are some that are huge?
13:53:15 <ajmiles> D3D12 has the ability to allocate a 4KB-aligned, 4KB texture, which for 32bpp would be 64x64. And you could actually load 500K of those in 2GB
13:53:16 <peter1138> No idea. The size of the sprites isn't known until they're loaded.
13:53:55 <ajmiles> But it does sound like I'd need to handle dynamically loading and unloading of sprites
13:54:28 *** Leopold__ has quit IRC (Remote host closed the connection)
13:55:44 *** ahyangyi has joined #openttd
13:55:44 <ahyangyi> ๐ค I don't think most sprites in a 4x newgrfs would be <= 64x64
13:56:00 <peter1138> Vanilla uses more memory than "necessary" as it scales everything to 4x zoom internally (depending on the max zoom level setting), so a standard 64x32 ground tile is actually 256x128.
13:57:00 <talltyler> Wow, #12154 is really something. My algorithm is embarassingly bad ๐ฆ
13:57:03 <peter1138> But that's only to simplify the blitter, as you are making changes you could change that.
13:57:12 <ajmiles> That scaling sounds like something I'd be able to avoid in a GPU renderer
13:57:48 <peter1138> It goes a bit deeper though as the grf sprite loader does all the sprite scaling, and then passes it to the blitter to encode into whatever format the blitter likes.
13:57:58 <andythenorth> anyway, goes it too many sprites in Horse? ๐
13:58:17 <peter1138> It would only need information about whether to prescale or not, not that complex.
13:58:58 <peter1138> Timberwolf's Trains has around 68000 sprites, you have another 50000 on top of that :p
13:59:16 <peter1138> (Rough figures, not accurate, don't quote me)
13:59:44 *** Leopold has joined #openttd
13:59:58 <andythenorth> I automated more ๐
14:00:17 <ajmiles> *"peter1138 said exactly 118,000 sprites. No more, no less."*
14:01:52 <peter1138> So.... about 116000
14:02:15 <peter1138> They're all pretty small though.
14:10:42 *** Extrems has joined #openttd
14:11:36 <xarick> npf.cpp code comments aren't matching the code
14:14:03 <Timberwolf> The .grf file for Timberwolf's Trains (not including addon) is more than 10 times the size of the files on the original Transport Tycoon CD-ROM.
14:14:16 <xarick> > Ships don't really reach their stations, but the tile in front. So don't
14:14:16 <xarick> > * save the station id for ships.
14:14:16 <xarick> Proceeds to save the station id for ships
14:15:42 <Timberwolf> Although TTO is a contender for one of the most under-utilised CD-ROMs going, at ~2.7MB it's the smallest amount of files I have on one. Even the Gremlin Zool/Lotus/Nigel Mansell compilation is about 300KB more.
14:30:19 <_jgr_> talltyler: I've got an idea for what to do with it, I'll probably do some testing with it this evening
14:33:43 *** Leopold__ has joined #openttd
14:37:50 *** Extrems has joined #openttd
14:39:19 *** Leopold has quit IRC (Ping timeout: 480 seconds)
14:41:31 <talltyler> _jgr_: Thank you. Let me know if I can help. โค๏ธ
15:33:33 *** HerzogDeXtEr has joined #openttd
15:54:36 <locosage> feels like openttd should be affected too but avoids it because of caching
15:55:28 <LordAro> probably means we'll run into it shortly then
15:55:34 <LordAro> i'd report it as a bug
16:02:11 <locosage> problem is I don't really know what I'm doing here to properly report it ๐
16:10:27 <LordAro> "CI failure during XX: <failure string>" and a link to the failure ought to be enough
16:19:19 <ajmiles> SamuXarickviaGitHub: I like the idea of a ship being "momentously lost". Not just a little lost, but momentously!
16:32:41 <peter1138> I found my battery charger.
16:33:02 <peter1138> Been looking for about 3-4 months. It was, uh, on my desk all along.
16:33:33 *** D-HUND is now known as debdog
16:35:21 <LordAro> in need of a tidy, was it?
16:40:31 <Rubidium> or maybe just in need of a visit?
16:43:35 <peter1138> 410 SQL lines just because I want to change one field's length and add a few booleans ๐ฆ
16:44:07 <peter1138> Too busy making useless patches and riding bikes to do desk cleaning, I guess.
16:51:40 <LordAro> xarick: I think the word you're looking for is "momentarily"
16:51:48 <LordAro> temporarily works as well though
16:55:12 *** Wormnest has joined #openttd
16:59:27 <ajmiles> xarick: It wasn't meant to mock you, sorry if you felt I did. I figured it was probably "momentarily", it just gave me a funny mental image of the ship
17:11:14 <merni> rubidium42viaGitHub: Yeah...
17:12:34 <merni> Frankly CJK-language users are probably well used to digits being grouped with commas in groups of 3 on computers
17:12:58 <merni> At most perhaps we could have different comma grouping rules?
17:13:02 <LordAro> obvious question: is there a library that already does it for us?
17:19:16 <LordAro> correct thing to do, of course
17:19:45 <merni> Languages are too complicated
17:19:59 <Rubidium> merni: we *do* have different comma grouping rules, but apparantly they are not good enough as it should group every four digits, but apparantly you should not write 1X 0000Y 1000 for 1 0000 0000 but 1X 1Z
17:20:18 <merni> Well, that's if you use grouping words
17:20:41 <merni> All that can hopefully be avoided if you just use commas
17:21:44 <merni> I think the way to think about the 1X 0000Y 1000 -> 1X 1Z stuff is like "one million and one" vs "one million, no thousands and one"
17:22:10 <Rubidium> but then you get to abbreviating large currencies in graphs and you have 1 million, but... they don't have a word for 1 million they want to group in sections of 4 instead of 3 for there
17:22:12 <merni> or roman numerals, if there are no Xs just omit that but
17:22:32 <Rubidium> *that* was the whole reason to even start with this endeavour
17:23:06 <merni> just have a list of powers-of-ten that have names for each language, and use the closest smaller one?
17:24:24 <merni> so for english you would define 6=million, 9=billion; for european languages maybe 6=million, 9=milliard, 12=billion, for Indian languages 5=lakh, 7=crore, for Chinese 4=whatever, 8=whatever
17:24:26 <Rubidium> it needs the second closest smaller one, and then you come to the world of Hindi where the digit groupings are not consistent
17:24:49 <Rubidium> so would that be 1,000 crore?
17:25:17 <merni> 1,000 crore = 1000 00 00 000 = 10^10
17:25:40 <merni> The thing is, there is no proper standard in Hindi/other Indian languages for what happens beyond 10 crore
17:25:47 <LordAro> perhaps it could do with a similar system to the pluralisations - just have a hardcoded set of them
17:25:56 <LordAro> rather than try to make it configurable
17:25:58 <merni> You can do further groups of two, and there are outlandish and unknown names for each group
17:26:07 <merni> Or you can just express it as "x crore"
17:26:17 <merni> 1,000 crore, 10,000 crore, etc
17:27:29 <merni> Indian languages don't have nearly as much standard in numerical and scientific related usage as European ones do, since most of the advanced work happens in English anyway
17:28:13 <Rubidium> in any case, I'm not going to try to fix this. I had my go, and it failed
17:29:27 <merni> So now will the zeroes be retained and not abbreviated in graphs? Or will they use "million", "billion" etc and make translators find equivalents anyway?
17:29:57 <merni> (which could be just `,000,000` or the transliteration of the English if there really is no word :P )
17:30:58 <Rubidium> that's what it was, and what that PR reverts to. And million/billion is abbreviated
17:31:33 <merni> Honestly, the fact that openttd is translateable to so many languages already puts it above tons of games and (even open source) software :)
17:54:53 <andythenorth> so wagon colours then? ๐
17:56:13 <merni> variants as proper groups?
17:59:11 *** virtualrandomnumber has joined #openttd
17:59:25 *** virtualrandomnumber has quit IRC ()
18:06:29 <andythenorth> what are 'proper groups'? ๐
18:06:38 <andythenorth> we have variant groups already?
18:09:14 <xarick> I wouldn't give up just yet. At least you were receiving feedback
18:16:04 <locosage> andythenorth: where group is a separate entity and not just another engine
18:16:31 <locosage> so it can be nested freely etc
18:16:46 <locosage> and don't need those ridiculous callbacks :p
18:32:03 <andythenorth> or as I call them, 'helpful grf spec additions'
18:33:33 <frosch123> 'whack-a-mole' is the term of the day for 'helpful'
18:35:34 <frosch123> poor moles, they are cute, don't whack them
18:36:13 <DorpsGek> - Update: Translations from eints (by translators)
18:38:31 <andythenorth> might have to refactor how Horse is handling name callback though
18:39:13 <andythenorth> I think it's currently about 20 MB of nfo in a total filesize of ~60 MB
18:40:08 <andythenorth> the requirement to use the text stack does not aid it ๐
19:05:26 <andythenorth> hmm seems I could pass strings along to a procedure as params
19:05:43 <andythenorth> do we know if that causes nml to generate a lot of varact 2?
19:07:30 <andythenorth> and there are approximately 5000 of them including variants
19:07:56 <andythenorth> it's maybe as much as 1/3 of the nfo
19:08:13 <andythenorth> I am not sure of the best way to optimise it, except trying different things and measuring file size
19:09:48 <frosch123> maybe nml needs some debug output: print N most long living switch-names
19:22:30 <andythenorth> in my initial test, passing parameters (substrings) up a procedure chain makes a bigger nfo than just having a chain per vehicle
19:23:47 <andythenorth> compiles faster though ยฏ\_(ใ)_/ยฏ
19:52:38 <peter1138> What's that meant to mean?
19:59:35 <wensimehrp> It would be an amazing feature. We just needs someone who can figure this out.
20:02:31 <peter1138> Full requirements from the start would be useful.
20:02:32 *** _zephyris has joined #openttd
20:02:32 <_zephyris> I suspect with a thorough design document as a guide it'd be easier
20:04:20 <Rubidium> good luck finding the requirements analist coming up with such a document
20:05:06 <peter1138> Bah, I want to use zip but it's C++23 ๐ฎ
20:05:13 <peter1138> (And probably broken with clang)
20:07:19 <Rubidium> dropping clang would solve a lot of issues, right? ;)
20:24:51 <_zephyris> Is there a preferred way to do design docs?
20:33:01 <_zephyris> (is it preferred who design docs come from? ๐ )
20:35:19 <frosch123> in the past we put them on the wiki, now they are usually put on gist
20:36:57 <frosch123> (if that's what you asked for)
20:42:10 <peter1138> Reward for finding a secret: being smashed to smithereens :S
20:42:44 *** Leopold__ has quit IRC (Remote host closed the connection)
20:47:06 <_zephyris> I have thoughts on settings naming/organisation, I'll write them down at some point...
20:51:37 <_zephyris> Wow, yeah, that is a good format
20:54:01 *** Leopold has joined #openttd
21:01:21 *** Flygon has quit IRC (Read error: Connection reset by peer)
21:14:00 *** nanapipirara has joined #openttd
21:14:00 <nanapipirara> Is it an idea to add cargo icons to vehicle lists when buying vehicles? It would make it easier to find what you're looking for
21:14:45 <frosch123> how would that work for refittable vehicles?
21:15:07 <frosch123> either way, plenty of newgrf already add icons to the list, to group types
21:15:11 <nanapipirara> don't they have a default cargo usually?
21:15:54 <frosch123> btw. do you know about the cargo filter at the top?
21:16:06 <nanapipirara> frosch123: I do! ๐
21:16:18 <frosch123> it also refits the vehicle while buying
21:17:03 <_zephyris> You've already followed the good gameplay design rule - make the vehicles that carry the same cargo look the same.
21:17:55 <peter1138> More cargo icons eh?
21:18:41 <xarick> is v->running_ticks supposed to reset to 0 every Ticks::DAY_TICKS?
21:20:36 <_jgr_> It's used to calculate the running cost to charge each day
21:20:42 <peter1138> `v->tick_counter` doesn't though, but that wraps at 256.
21:21:30 *** berndj has quit IRC (Ping timeout: 480 seconds)
21:25:27 <wensimehrp> Possible but might bring a long and nasty list
21:25:27 <wensimehrp> Another way is to define cargo classes' icons
21:25:29 <talltyler> That seems like overkill. Canโt you just check the current speed, and if itโs 0 donโt block the ship?
21:26:18 <talltyler> You could remove the current check for VS_STOPPED since that would also be handled by checking speed.
21:26:44 <talltyler> (I havenโt looked at the code recently, but added the stopped flag check a while back)
21:27:30 <xarick> the ship could immediately enter that dock right in the tile it exits
21:34:31 <xarick> ok, seems fine after all, gonna do it
21:39:55 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:49:48 <peter1138> We have unbunching so no ๐
21:53:22 <andythenorth> flat unbunching?
21:57:56 <andythenorth> Flat Variants though
21:58:05 <andythenorth> like Variants, but without a group parent
22:03:53 <peter1138> That's easy, just remove the variant property.
22:07:20 <talltyler> We should unbunch variants and distribute them evenly through the purchase list
22:10:31 <peter1138> Is there a function to get the nth set bit?
22:11:02 <peter1138> We have FindFirstBit and FindLastBit...
22:32:09 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:45:48 <peter1138> Probably shouldn't be implemented, no.
22:51:04 <belajalilija> should certainly be optional
23:02:10 <talltyler> I love cargo icons, but the purchase list is too busy. The cargo dropdown is the correct way to find compatible vehicles, IMO, not looking at the list.
23:03:14 <talltyler> Sometimes I even wish we had a power type dropdown (diesel, electric, steam) to further filter the list, but railtypes might make a mockery of that ๐
23:03:36 <talltyler> Although we do have engine class, so maybe we could use thatโฆ
23:19:55 *** keikoz has quit IRC (Ping timeout: 480 seconds)
23:20:27 <reldred> I am already sick to death of grf authors loading their purchase list sprites with fifty million silly little icons. This would be a big 'no thankyou' from me.
23:20:52 <reldred> And not terribly useful given cargo refits
23:26:42 <locosage> It's pretty useful for vanilla vehicles though I'd probably put cargo icons on the right here.
23:30:10 <xarick> bed now, cyas goodnight
23:36:27 *** Tirili has quit IRC (Read error: Connection reset by peer)
23:44:24 *** tokai|noir has joined #openttd
23:44:25 *** ChanServ sets mode: +v tokai|noir
23:51:15 *** tokai has quit IRC (Ping timeout: 480 seconds)
23:57:37 <nanapipirara> peter1138: oops, that doesn' look great no.. But mostly because it's a mismatch of resolutions i think. oh well ๐
continue to next day โต
