IRC logs for #openttd on OFTC at 2025-01-10
⏴ go to previous day
02:05:06 *** gelignite is now known as Guest5462
02:05:09 *** gelignite has joined #openttd
02:12:26 *** Guest5462 has quit IRC (Ping timeout: 480 seconds)
02:41:11 *** gelignite has quit IRC (Quit: Stay safe!)
03:15:37 *** Wormnest has quit IRC (Quit: Leaving)
03:41:14 *** debdog has quit IRC (Ping timeout: 480 seconds)
04:05:24 *** gnu_jj_ has quit IRC (Ping timeout: 480 seconds)
04:06:03 *** D-HUND is now known as debdog
04:42:27 <DorpsGek> - Update: Translations from eints (by translators)
06:09:14 *** ufo-piloot has quit IRC (Ping timeout: 480 seconds)
06:09:54 *** keikoz has quit IRC (Ping timeout: 480 seconds)
06:40:25 *** ufo-piloot has joined #openttd
07:43:45 *** locosage has quit IRC (Quit: User went offline on Discord a while ago)
08:45:24 *** Flygon has quit IRC (Read error: Connection reset by peer)
10:47:21 <xarick> The following tests FAILED:
10:47:21 <xarick> 69 - regression_regression (Failed)
10:58:19 <xarick> the output differs from the expected result
10:58:30 <xarick> it caughts 2 CompanyRename events
11:04:22 <peter1138> talltyler, yeah, general rule is that translatable strings should be complete, not composed of parts.
11:06:22 <peter1138> talltyler, therefore probably the correct way is to expose a bit that the NewGRF can test for during callbacks.
11:07:09 <peter1138> I imagine they only need know calendar vs wallclock, nothing as detailed as minutes-per-year.
11:15:30 <andythenorth> production is in 'minutes', I'd expect grf-specific behaviour to mimic that
11:15:58 <andythenorth> I have NFI how FIRS should scale requirements to wallclock
11:16:04 <andythenorth> not sure I really care 😛
11:16:35 <peter1138> It shouldn't scale.
11:16:54 <peter1138> The only purpose is to make this text correct.
11:18:29 <andythenorth> FIRS behaviour runs on periodic processing callback, so that should be fine
11:18:58 <peter1138> I don't think there is any other way.,
11:19:18 <andythenorth> so it will just be n minutes
11:19:50 <andythenorth> if it's currently 3 months, it will be about 3 minutes
11:23:16 *** gelignite has joined #openttd
11:25:01 <peter1138> Okay, how do I fix Action 0 not allowing to change more than 255 items at once?
11:26:21 <peter1138> Bump the NewGRF version and make numinfo an extended byte?
11:28:53 <peter1138> Don't bump it and hope that nobody uses 255 items at once already? (My badges NewGRF does but that can be fixed)
11:29:47 <peter1138> Add an extension system that allows NewGRFs to enable changes as desired?
11:33:29 <andythenorth> bump the version?
11:34:36 <peter1138> If we bump the version, what are all the other things we wanted to change...
11:35:04 <peter1138> Number of act2 ids.
11:38:45 <peter1138> And do we do byte -> extendedbyte or just byte -> word?
11:40:47 <peter1138> Okay, 32768 act2 ids is enough?
11:50:42 <andythenorth> the compile times mitigate against using that many 😛
11:51:19 <andythenorth> I suspect nmlc will increase compile times at one of those O(log n) things I don't understand 😛
11:51:26 <andythenorth> for computing act2 IDs
11:59:35 <peter1138> Bump the size of switch ranges?
12:16:46 <xarick> yay, elon musk and poe 2 drama is hilarious
12:18:48 <pickpacket> is there a way to generate what the regression output file should look like?
12:19:14 <peter1138> Run the regression test. Copy your "failed" output over to the expected output. Your "failed" output is no longer failed.
12:29:51 <andythenorth> slow motion car crash
12:30:11 <andythenorth> under mild protest, we have some Wordpress, on WP Engine
12:30:17 <peter1138> 16-bit Goto labels?
12:30:41 <peter1138> More like 15 GOTO 40000
12:31:07 <andythenorth> what about, instead, we shard grfs 😛
12:31:26 <andythenorth> keep the current 'limits' and force a sub-grf format 😛
12:31:58 <andythenorth> varact 2 targets, but with a grfid 😛
12:32:07 <andythenorth> "what could go wrong"
12:35:46 <_jgr_> The ambiguity whether an action 7/9 value is an offset or a label ID is something that it would be a good idea to get rid of if the field was made wider
12:39:40 <peter1138> andythenorth, remember that NewGRF translation stuff... there is Action 13 -- translate GRF strings already.
12:40:32 <peter1138> numsprites can be a goto label...
12:45:14 <peter1138> Okay, So goto labels are 16 bits, not because that's all that useful but because it means that we can dedicate bit 15 of that numsprites to mean "this is a label".
12:46:01 <peter1138> (numsprites being 16 bits means you can also skip more than 255 sprites without using a goto label, although that is unwise)
12:46:11 <_glx_> Ah yes stupid if number happen to match a label you're screwed
12:54:26 <peter1138> So far I've got: make action 0 numinfo 16 bit. make action 2 id 16 bit. make varact 2 subroutine 16 bit to match. make varact 2 range count 16 bits. make randact 2 result count 16 bits (not really needed but consistency). make act 3 idcount 16 bit. make act 3 ctype 16 bit (allows translation tables with more than 255 entries). make act 4 num strings 16 bit. make act 4 base id always 16 bit (instead of
12:54:32 <peter1138> sometimes 8 bit, sometime extended byte, sometimes 16 bit). make act7/9 numsprites 16 bit, bit 15 must be set to use a label.
12:55:15 <peter1138> make act A num_sets and numsprites 16 bit (likely not needed but consistency)
12:56:17 <peter1138> make act 10 goto labels 16 bits (just consistency, mainly allows the lookup logic to be simplified)
12:57:15 <peter1138> make act 12 num defs 16 bit (consistency), num chars 16 bit (maybe useful) and base character 32 bit instead of 16 bit (so finally allows characters > 0xFFFF to be defined)
12:59:18 <peter1138> make act 13 num_strings 16 bit (also just consistency)
13:02:24 <peter1138> Would we want 1-byte/3-byte (readextendedbyte()) or 2-byte values (readword())
13:03:32 <peter1138> Hard to change existing NFO. NML would just need a rebuild I believe.
13:04:04 <peter1138> (And IIRC grf-py mostly always uses 3-bytes so that it doesn't have to deal with variable-length stuff)
13:05:50 <peter1138> andythenorth, lunch?
13:43:07 <peter1138> I'm sure there would be other things for a grf version bump, right?
13:50:25 <_glx_> hmm but act2 IDs will be 16 bit but limited to max 15bit
13:50:44 <_glx_> because bit 15 for callbacks results
13:51:06 <peter1138> That's why I said 32768 above.
13:51:18 <peter1138> True it's 15 bit, but it's stored as 16 bits.
13:52:02 <_glx_> 32K in a single scope is huge
13:56:44 <peter1138> So's 256, but still...
13:57:43 <_glx_> with procedures it's easy to consume the 256
14:06:09 <pickpacket> There should be a word for the sudden urge to play OpenTTD
14:12:37 <talltyler> What about the constant, neverending urge?
14:20:16 <talltyler> Will you believe that I played OpenTTD last night?
14:23:23 <pickpacket> talltyler: there should be a word for that, too
14:34:53 <andythenorth> peter1138: I should, I've overlooked it
14:36:13 *** kuka_lie has joined #openttd
14:44:45 <andythenorth> are crisps lunch?
15:57:20 <truebrain> talltyler: Not really, as I don't see any new bug reports
16:00:07 <pickpacket> running regression tests locally, is that just "make regression"?
16:05:50 <xarick> I wanna start work on MakeRiverWider
16:14:00 <LordAro> nothing stopping you from basing your work on your existing commits
16:23:42 <_glx_> pickpacket: `make test` IIRC, but you can also run `ctest` directly in build dir
16:26:26 <peter1138> Only if cmake is set to generate Makefiles.
16:40:25 *** Wormnest has joined #openttd
16:50:57 <_zephyris> So, if I try to do a python-only build process for OpenGFX2, what's the best way to do that? Just run a single script and it does everything?
16:52:40 <_zephyris> Hmm, I'm guessing that's pretty bad from a python dependency point of view
17:00:06 <andythenorth> all my grfs are Make calling python scripts
17:00:31 <andythenorth> I'd have an entry point script, and then break the rest out for different concerns
17:13:10 <LordAro> _zephyris: it'd certainly be nice to have incremental builds
17:14:55 <_zephyris> The python already checks if image sources have been updated, if that's what you mean
17:15:20 <_zephyris> And nml does caching too...
17:16:25 <_zephyris> Probably make triggering a single python script for building the graphics then
17:16:46 <_zephyris> andythenorth: I was just poking the iron horse for ideas 🙂
17:17:20 <andythenorth> I'm not sure how weird those are
18:06:36 <peter1138> Hmm, missed loads of bits.
18:39:41 <_tweez> pickpacket: withdrail.
18:42:07 <truebrain> Did anyone fix the nightly CI today btw?
18:42:14 <truebrain> Peter broke it yesterday 😛
18:47:34 *** tokai|noir has joined #openttd
18:47:34 *** ChanServ sets mode: +v tokai|noir
18:54:24 *** tokai has quit IRC (Ping timeout: 480 seconds)
21:01:58 <peter1138> Does/should NML have a way to select the output grf version?
21:03:10 <andythenorth> not mentioned in the wiki
21:03:58 <andythenorth> not mentioned in `--help`
21:05:01 <andythenorth> 'should it' is a good question
21:08:08 <peter1138> Or not, that's the version of the grf, not the grf verison ;D
21:10:49 <peter1138> Oh we should probably increase the house limit in NML from 255.
21:30:12 <peter1138> Hmm, okay, I have an idea for translation tables.
21:30:46 <peter1138> Either grfv9 it, or allow extending a table.
21:31:45 <peter1138> Instead of requiring the ID to start from 0, allow appending as long as the start point matches the current size.
21:41:55 *** gelignite has quit IRC (Quit: Stay safe!)
21:59:49 <peter1138> Hmm, I assume NML only ever skips with labels?
22:00:48 <peter1138> Obviously that's the only way that makes sense from .nml, but not sure if there are internal skips.
22:27:06 <peter1138> Hmm, I wonder why 16...
22:27:19 <peter1138> I guess it ensures there's space for labels.
22:29:54 <peter1138> I see, this Action7 code is a bit gnarly as it calls it "label" even when it's not.
22:30:12 <peter1138> Also, self.label = label is there twice :D
22:34:37 <peter1138> Hmm, how do I bulk-update all the regression tests...
22:40:59 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:50:56 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:06:30 <peter1138> Confused about action2var.py:95
23:11:43 <peter1138> Ah, because 'parameter' is used as both parameter and procedure refernce.
23:31:57 *** kuka_lie has quit IRC (Quit: leaving)
23:34:55 <_glx_> I use a powershell script to run and update all regression tests
23:36:30 <_jgr_> I usually just use `make -k` then copy all the files
23:46:05 <peter1138> make -k, that'll do, thanks :D
23:52:50 <peter1138> AssertionError: Expected 378 bytes to be written to sprite, got 393
continue to next day ⏵
