Times are UTC Toggle Colours
00:01:52 <xarick> omg, how many trains/wagons exist in iron horse 3.14.0 ? 00:04:01 <peter1138> Some would say ... enough. 00:04:11 <xarick> my poor AI 00:04:43 <xarick> is experimenting all train/wagon compatibility of all available engines 00:04:55 <xarick> attachment testing 00:05:01 <_glx_> oh there are many variants 00:06:07 <xarick> never imagined the cause of my AI apparent slowdown was due to the insane amount of engines being tested 00:06:26 <_glx_> maybe API needs to know about variants 00:06:50 <xarick> there really should be a way to check if a train can pull a wagon 00:07:03 <xarick> without buying to test it 00:07:42 <_glx_> it's impossible, newgrf needs a real vehicle for callbacks 00:09:10 <_glx_> and in iron horse many wagons are just duplicates (variants with same data but different colours) 00:10:57 <Eddi|zuHause> maybe it's time to revive the AI callback? 00:11:01 <xarick> testing attachments since 1935, it's 1943 now 00:11:45 <xarick> it's a one time thing, but gee... 00:11:46 <Eddi|zuHause> given an engine and a cargo, and a usage pattern, the AI callback will return a fitting wagon 00:12:51 <_glx_> but newgrf rarely implement it 00:13:09 <Eddi|zuHause> yes, because OpenTTD never used it 00:18:38 <peter1138> Chicken & egg. 00:19:18 <Eddi|zuHause> back when noai was introduced, i was told off for suggesting to implement it... 00:21:34 <_glx_> if you want some fun, try dbset where some engines require a specific amount of passenger cars 00:21:51 <_glx_> or old UKRS and the mandatory caboose 00:21:59 <peter1138> And that's from 2004 or something :p 00:23:49 <_glx_> it might be simpler to use <https://docs.openttd.org/ai-api/classAINewGRF> and have predetermined consists hardcoded in the AI source 00:24:42 <_glx_> and fallback to your expensive method for unknown newgrf 00:25:05 <_glx_> https://cdn.discordapp.com/attachments/1008473233844097104/1219803866275647558/image.png?ex=660ca1e0&is=65fa2ce0&hm=cb3638c6d1b2dd932af2eb3eac1b5a13ba56ff891af3573f7a5034d120217e2d& 00:25:19 <_glx_> that's just a lot of combinations 🙂 00:25:29 <peter1138> Implementing the callback to allow AIs to use it doesn't mean AIs need to use it. 00:25:46 <peter1138> But, uh, difficult to test without much source material 🙂 00:26:15 <_glx_> I think dbset had the AI callback 00:27:16 <peter1138> IIRC the callback is able to return special AI-only vehicles which allow the AI to cheat. So that'll have to be filtered out. 00:28:02 <Eddi|zuHause> well, the command to actually purchase the returned vehicle should take care of that 00:28:29 <_glx_> oh and nml doesn't implement CB18 🙂 00:28:49 <Eddi|zuHause> that's a different problem :p 00:29:31 <_glx_> many newgrf are written with nml, so without nml support it's not very useful 00:30:21 <_glx_> ah and the CB doesn't support railtypes or NRT 00:33:36 <peter1138> Or multiple NewGRFs. 00:34:08 <_glx_> "This makes it possible to leave certain decisions to the previous handler and allows some cooperation between several .grf files. " 00:34:20 <_glx_> it's supposed to support multiple grf 00:34:24 <peter1138> Oh so it does 🙂 00:35:06 <_glx_> but "For trains, the track types (0=regular, 1=electric, 2=monorail, 3=maglev) are automatically mapped as appropriate depending on the electrifiedrailways/unifiedmaglev switches." 00:35:14 <_glx_> that's quite limiting 00:36:46 <peter1138> Almost like it wasn't implemented for a reason 😄 00:38:14 <_glx_> I really think it's simpler to check for newgrf presence and provide working association directly in the AI code 00:40:29 <_glx_> or introduce variants support in the API so it's possible to filter them out 01:42:03 *** Wormnest has quit IRC 01:48:17 *** tokai|noir has joined #openttd 01:48:17 *** ChanServ sets mode: +v tokai|noir 02:02:20 *** wallabra has joined #openttd 02:02:51 *** herms has quit IRC 02:03:42 *** herms has joined #openttd 06:01:35 *** Ox7C5 has joined #openttd 22:55:13 <rau117> andythenorth: Hmm... on the other hand, being able to place a fence (or some other object) on top of another tile would be nice. 22:55:13 <rau117> This will cause problems, but it will create a lot of new decorating opportunities 22:58:34 <peter1138> Remember when tile-edge signals was the meme... 23:27:05 <DorpsGek> [OpenTTD/OpenTTD] MontyMontana opened issue #12344: [Bug]: Root directory has changed since previous versions and can not be changed. https://github.com/OpenTTD/OpenTTD/issues/12344