IRC logs for #openttd on OFTC at 2025-08-26
โด go to previous day
00:06:46 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
01:52:53 <yiffgirl> allowing llmslop anywhere near the code starts a nasty precedent
01:55:17 *** tokai|noir has joined #openttd
01:55:17 *** ChanServ sets mode: +v tokai|noir
02:02:15 *** tokai has quit IRC (Ping timeout: 480 seconds)
02:29:50 *** Wormnest has quit IRC (Quit: Leaving)
03:16:24 *** Zathras has joined #openttd
03:19:41 *** Zathras_11 has quit IRC (Ping timeout: 480 seconds)
04:40:56 <DorpsGek> - Update: Translations from eints (by translators)
05:19:51 <andythenorth> FWIW, the first place I saw LLMs being used to write code was here
05:20:08 <andythenorth> I was pretty unconvinced prior to that
05:36:21 *** WormnestAndroid has quit IRC (Remote host closed the connection)
05:36:25 *** WormnestAndroid has joined #openttd
05:47:21 <rito12_51026> `Your connection is not private
05:47:21 <rito12_51026> Attackers might be trying to steal your information from www.tt-wiki.net (for example, passwords, messages, or credit cards).
05:47:21 <rito12_51026> net::ERR_CERT_DATE_INVALID`
05:59:08 <LordAro> andythenorth: but you were convinced after?
05:59:23 * LordAro ponders whether we need an "official policy"
05:59:56 <LordAro> (which i'm perfectly happy with it being "none at all")
06:01:13 <LordAro> rito12_51026: expired cert. the powers that be have been informed
06:26:48 <andythenorth> LordAro: yes, TB showed convincing results
06:27:01 <andythenorth> at least to the extent of LLMs being fancy autocomplete
06:34:15 <kuhnovic> An LLM is a tool, and tools can be used correctly and incorrectly. At work I use the integrated copilot in VS and it saves me quite a bit of time. It is mostly a fancy autocomplete although it can do much more if you want it to. Including letting it create all kinds of AI slop. The problem arises when people don't check the generated code, or worse, don't understand it. IMO that's not the fault of
06:34:15 <kuhnovic> the LLM, and banning LLM generated code altogether feels wrong to me.
06:38:37 <reldred> It's like a lot of tools in all trades. There are certain tools you don't leave the Apprentice alone with because they'll cut their hands off or otherwise make a mess of the job. Ethics aside, and I can definitely sympathize with the ethical concerns though I am much more a pragmatist, I think it'll be interesting to see what AI/LLM tools survive once the VC's get bored, everyone starts wanting an
06:38:37 <reldred> ROI, and the tools have to start costing what they actually cost to run.
06:39:53 <reldred> Maybe a "No AI/LLM from first time contributors" needs to be the rule, maybe it needs to be a straight up ban. People far smarter than me will have to wrestle with that one.
06:39:54 <andythenorth> well apparently there will be unprecedented economic growth from AI
06:40:02 <andythenorth> which drives the valuations
06:40:25 <andythenorth> this will be achieved by the massive reduction in headcount that most companies will be able to achieve via agentic AI
06:40:29 <reldred> yeah bunch of VC's are allegedly apparently wink wink nudge nudge say no more say no more getting a bit bearish on it.
06:40:34 <andythenorth> what in this picture doesn't work?
06:41:48 <reldred> all I know is that most AI diehards I know are also starting to border on total mental incompetence since they outsourced all their thinking, so it's put me right off it.
06:42:30 <reldred> And a manager once insulted me by asking me to feed all my work emails through chat GPT to 'make them read better' and I refused.
06:42:50 <reldred> not even on the disclosing customer data to a third party system point
06:43:04 <reldred> but just that I was offended he didn't like my writing.
06:44:31 <kuhnovic> Maybe he was hinting at your creative use of profanity. Which I find quite amusing.
06:45:25 <reldred> this email didn't even have any of that! No, the email he didn't like had a bit of a run-on sentence. He could have just said that and I would have been OK with it.
06:46:00 <reldred> But yeah that business was a complete joke
06:46:09 <LordAro> We've done an awful lot of messaging at work along the lines of 'don't put company data into any random LLM'
06:46:37 <reldred> I'm now on the other side of the country earning double what I used to deal with about a quarter of the BS.
06:46:45 <kuhnovic> It is scary that the knee-jerk reaction seem to be "run it through chatGPT" instead of "have another look, use your brain and change it"
06:47:03 <reldred> seems to be the missing element
06:47:51 <reldred> I do think that someone who is mentally disciplined could do great things with it, but it was like the early days of google, we had to have school subjects on using search engines, understanding and citing sources, copyright, etc. etc.
06:47:54 <kuhnovic> LordAro: Where I work they created their own LLM just to avoid leaking IP and customer data
06:48:25 <reldred> Yeah I've heard of some really great success stories of locally created LLM's trained on the internal knowledge base.
06:48:36 <reldred> Airgapped from the outside world
06:48:53 <LordAro> indeed, we've got a couple of those now too, though generally that's just paying money to services that promise not to use the data...
06:48:54 <reldred> And people could just go pepper it with all their dumb questions on things rather than harass the SME's directly.
06:49:50 <LordAro> one of the major things that have been done with it is to implement a wiki search (we use a wiki as a primary source of knowledge), which i still don't really understand the need for
06:50:37 <andythenorth> LLMs as fact searches are terrible
06:50:54 <andythenorth> but then so are wikis ๐
06:51:02 <reldred> Also generally yes ๐
06:51:31 <reldred> But Iโm one of those โdocumentation and procedures too often get used as an excuse for an engineer to not understand the work they are doingโ kinda people
06:52:05 <LordAro> for some reason my arguments that people should search for something that's actually on the page didn't wash
06:52:19 <LordAro> or better, change the page so it contains those search terms
06:52:37 <LordAro> i'm not sure why they're expecting google (of old) levels of search
06:53:30 <kuhnovic> Because people think that LLMs can think
06:53:31 <LordAro> reldred: oh i love those. our QA department is great at them
06:54:08 <LordAro> eventually i'll get an answer of it being needed by some external standards document, but it's not like anyone except the QA lead actually knew that
06:54:13 <kuhnovic> reldred: "Just put a note in the documentation so we point them to that"
06:54:29 <reldred> I mean not entirely, I think documentation and procedures are important, I just get annoyed at how they get used sometimes.
06:54:49 <reldred> or maybe I'm just annoyed at shitbirds who can't do their job.
06:57:12 <kuhnovic> Another good example of a tool being used and abused
07:01:23 <reldred> all I can think of now is 'oh shit its bobby WITH THA TOOL!'
07:01:44 <andythenorth> GPT is my friend
07:01:49 <andythenorth> it understands me and cares about me
07:02:04 <reldred> I just stick with having no friends
07:03:45 <reldred> I left them all behind halfway across the country to go chase money. Feels bad.
07:15:01 <_zephyris> Linear "thinking" and glorified autocomplete are good, connected thinking not.
07:34:03 <andythenorth> that is interesting ๐
07:45:04 *** zai2137 has joined #openttd
07:45:04 <zai2137> _glx_: I see you tried to fix Bee Reward in the past, have you been able to get it working with newer OTTD versions? the value of this.wanted_amount is good, but it seems GSText won't ever print the news message/goal entry properly here
07:45:49 <zai2137> the type of cargo, destination, and the reward are all good. It's only the amount that's missing, as if CARGO_LONG ignored the second param (int)
08:20:38 *** toktik has quit IRC (Remote host closed the connection)
08:29:00 *** Flygon has quit IRC (Remote host closed the connection)
08:38:00 <pickpacket> zai2137: What's Bee Reward?
08:46:49 <zai2137> It's a gamescript which used to work, but due to changes in OTTD (presumably) it stopped working properly
08:49:46 <_zephyris> Another day, another rebase
08:51:16 <_zephyris> Presumably would be best to remove TTFs from the repo, only build and upload as a release/to vendor for ottd. Then occasionally update the preview images.
08:54:13 <LordAro> How much effort is it to build the TTF via CI?
09:22:00 *** SigHunter has joined #openttd
09:30:16 <_zephyris> Shouldn't be hard, but I wouldn't know where to start ๐
09:32:16 <LordAro> in theory it's just calling build.sh / make from a github action. Of course, it's always a bit more involved than that (installing dependencies, running extra scripts etc)
09:33:22 <_zephyris> It should be a simpleish one (`make all`, requiring python, PIL/pillow, fontforge)
09:51:48 <LordAro> tbf on the preview images, it's often the only way i can tell what's changed
09:51:58 <LordAro> (with the swipe image diff view)
09:52:15 <_zephyris> Yeah, it is handy like that.
10:04:23 <_zephyris> Might be a good change, not sure TBH.
10:30:43 <_glx_> zai2137: Most likely worked by pure luck when GSText were not properly validated
10:42:30 <zai2137> _glx_: And have you been able to find any resources that'd help you troubleshoot it? I'm confused as to where the error occurs, whether it's due to the type of the variable being passed, the parameter's index or anything else. The value prints nicely, so does the type of the argument. Unfortunately, I haven't had that much experience with C and its derivatives in the past, so there might be a
10:42:30 <zai2137> language-specific troubleshooting step im not aware of.
10:50:50 <_glx_> zai2137: In theory there should be some info in script debug window
10:51:24 <_glx_> Validation is quite verbose
10:52:46 <zai2137> _glx_: Yeah that's the issue. There isn't. There were some validation errors in the repo I cloned, but after tweaking it it no longer throws anything at me. It's as if GSText defaults to "0", even if the parameter is correct
10:56:05 <zai2137> I'll take a look at OTTDs handling of string actions later to check if I see anything suspicious there
10:58:33 <_glx_> Using 14.1, 15-beta or nightly?
11:07:45 *** dh1 has quit IRC (Quit: My Mac has gone to sleep. ZZZzzzโฆ)
11:12:27 *** peter1138 has joined #openttd
11:12:27 *** ChanServ sets mode: +o peter1138
11:13:57 <peter1138> What's slopping today?
11:20:05 <peter1138> I haven't yet deleted my fork, so there's that.
11:24:53 <talltyler> Just a proposal, curious to see what others think (leave your comments on GitHub for posterity ๐ )
11:35:23 <andythenorth> looking what TB wrote with copilot
11:35:28 <andythenorth> might have mostly been TrueGRF
11:36:22 <andythenorth> oh possibly parts of Bananas
11:36:39 <andythenorth> Iron Horse is significantly LLM generated at this point
11:37:12 <peter1138> Until Iron Horse is upstreamed into vanilla that is your dubious choice.
11:38:24 <ahyangyi> Anyways, that PR makes me reconsider my statement "AI is good at writing tests"
11:38:36 <ahyangyi> AI is good at writing tests that I already know how to write
11:38:39 <andythenorth> AI isn't specifically good at anything
11:38:48 <ahyangyi> AI is specificially bad at something
11:38:52 <andythenorth> LLMs are specifically good at token stream prediction that pleases the end user
11:39:04 <ahyangyi> so might as well call the "not specifically bad" problems "good"
11:39:44 <andythenorth> LLMs don't lunch
11:40:47 <ahyangyi> Kind of weird to think that even the slowest AI is also *fasting*
11:48:52 <talltyler> LLMs are specifically good at allowing capitalists to continue their delusion that infinite growth is possible and desirable
11:52:45 <peter1138> > it doesn't know what words mean, just which words fit
11:54:50 <talltyler> Thatโs the best shortest description Iโve seen ๐
11:55:31 <talltyler> Of course some apologists will try to convince us that human brains work the same wayโฆ which is not true. ๐
12:01:55 <_zephyris> There is a fundamental attention span/range/distance limitation inherent to LLMs which (currently) is their limit for larger scale reasoning. The open question is how big does that span/range/distance need to be, and is the LLM architecture able to do that within compute availability.
12:02:21 <peter1138> talltyler, it will be true if they keep using it...
12:04:30 <talltyler> Yes, unfortunately. ๐
12:05:17 <_zephyris> Well, on the plus side, OpenTTD-TTF has no open bugs, and all without LLM help!
12:07:17 <_zephyris> How about if I write my own ML model, and use that?
12:07:57 <_zephyris> Sounds silly, but I was experimenting with making and training a pixel art upscale model.
12:09:42 <ahyangyi> Perhaps we can limit it to LLMs.
12:09:45 <zai2137> _glx_: jgrpp 0.65.4, so I'll test on vanilla to see if it's not JGRs fault
12:09:50 <Timberwolf> I tried to train a model once to convert the output of GoRender into properly Simon Foster style sprites (i.e. features perfectly aligned on pixels, that certain "crispness" they have) but never managed to get the reward function right so it'd get stuck on stupid stuff like "change the index of the transparent background by 1"
12:09:51 <ahyangyi> AI is a very vague term
12:10:02 <Timberwolf> Also lack of training data.
12:10:29 <ahyangyi> We certainly cannot ban pathfinders in OpenTTD, even if they were once part of "AI research"
12:11:47 <peter1138> zai2137, what are you testing?
12:13:02 <talltyler> That's a good policy
12:14:21 <zai2137> _glx_: it's always the simplest stuff - it works well on vanilla, thanks for reminding me about that ๐
12:15:24 <zai2137> peter1138: There was an issue with a string not being substituted properly by GSText, while it worked well on previous versions - I thought something might've changed in the game's source
12:15:46 <peter1138> Many things change in the game's source.
12:16:27 <peter1138> Especially with strings as I touched quite a bit of that.
12:16:35 <zai2137> That is exactly what led me to believe something broke in GSText itself
12:17:11 <peter1138> It's possible, though I did actually add a unit test for converting from old to new format for stored GSText strings.
12:17:51 <peter1138> Some GS are just broken though, like Bee Reward.
12:20:29 <zai2137> peter1138: oh? What do you mean?
12:20:35 <peter1138> (I think that may have been fixed, but then not released.)
12:21:33 <peter1138> For that? The version on bananas has some incorrect strings, but the version on the gitlab repository is corrected.
12:26:30 * LordAro pastes that into the relevant #work channel
12:29:49 <peter1138> I guess the author forgot to release? (Who would do such a thing...)
12:33:31 <zai2137> pretty sure they gave up when ottdcoop kicked the bucket
12:41:20 <zai2137> Anyway, after fixing the strings, the base functionality (generating goals and providing the rewards) seems to be working. I'll be publishing this corrected version on bananas soon, but first I'll see what in JGRPP might be causing these issues. Thanks all for thinking along!
12:54:02 <peter1138> Me: prepares a PR, also me: merges the PR, also me: oh crap, perhaps I should've included this as well.
13:00:58 <_glx_> indeed I failed, but we are lucky it seems to not cause any side effect
13:39:42 <peter1138> No idea how that didn't fail.
13:48:27 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
13:48:28 *** WormnestAndroid has joined #openttd
13:49:17 <peter1138> Should we continue to support this font configuration? :o
13:53:38 <peter1138> Side-by-side update takes account of the sprite font height regardless of the configured scalable font size.
13:54:05 <LordAro> i see no reason why not
13:54:13 <peter1138> (Even when the sprite font isn't visible in a string.)
13:54:34 <LordAro> ah, that might be worth considering
13:54:54 <peter1138> This is, of course, and extreme case that nobody normal would be using. But I was surprised by seeing "3" in the font size survey data....
13:56:22 <peter1138> Only occurs if you have smaller-than-default fonts. The more common case is default size or larger-than-default fonts.
14:04:22 <_glx_> it didn't fail for a simple reason, registering calls are direct calls and the script main loop is not yet loaded/running, so stack state is less vital
14:20:17 <peter1138> Hmm, PostgreSQL style users-are-just-roles-with-a-password, perhaps.
14:41:01 <rito12_51026> Why functions regarding to signals in `rail_map.h` are not documented with use of comments? Other functions in this file are
14:44:58 <talltyler> No reason besides nobody wrote them ๐
14:49:43 <rito12_51026> Should I then add comments in corresponding metro functions or leave uncommented to match rail ones?
14:50:18 <LordAro> given there's a near-zero chance this 'metro' stuff ever gets into master, you can do whatever you like
14:58:46 <rito12_51026> LordAro: Is there any chance of getting it as a newGRF only feature, same as trams?
15:00:27 <LordAro> I've no idea, all I know is that extending the map array in the way you're going is unlikely to be approved
15:00:41 <LordAro> Not that this should stop you! Some of the best features started out as massive hacks
15:02:53 <rito12_51026> I haven't extended map array, it's size remains as it was
15:03:21 <LordAro> then maybe my information is out of date
15:03:47 <LordAro> the general rule for new stuff is "do it properly" even if what's currently there doesn't
15:04:03 <rito12_51026> I've used all free bits that were there
15:04:54 <_glx_> what would be different between your "metro" and a railtype ?
15:05:35 <LordAro> _glx_: i believe the idea is to be able to build tracks under buildings and roads and things
15:05:40 <LordAro> and possibly stations too
15:05:40 <rito12_51026> LordAro: Ok, should I also add comments in rail_map.h or whould it be multiple things in one PR?
15:06:18 <LordAro> rito12_51026: as far as PRs are concerned, keep them focused, don't make unrelated changes that you don't have to
15:06:19 <_glx_> would be better to have that for any railtype
15:06:27 <rito12_51026> stations are a bit problematic though
15:06:35 <LordAro> if you want to add comments to functions that are already there, then by all means PR that separately
15:08:35 <rito12_51026> I have one more question, how to name static variables?
15:09:07 <rito12_51026> I couldn't find example in CODINGSTYLE.md
15:09:11 <LordAro> i don't think we have any particular rules about that
15:09:22 <LordAro> see what existing stuff does (and go with the majority)
15:11:06 *** Wormnest has joined #openttd
15:14:25 <_zephyris> Sorry talltyler, silly rebasing mistake.
15:31:37 <davidxn> Out of interest (as I mention it a lot in my videos), is the tendency towards bitwise operations in the OpenTTD code deliberate for efficiency, or a holdover from when the code was in assembly? (For example, calculating when a new quarter is starting by taking the number 1001001001b and checking if its [new month]th bit is 1)
15:36:48 <peter1138> For reference the value is actually "
15:37:05 <peter1138> For reference the value is actually `1 << 0 | 1 << 3 | 1 << 6 | 1 << 9` rather than 1001001001b.
15:38:17 <_zephyris> On the subject of LLM weirdness...
15:38:30 <peter1138> Anyway, it avoids dividing by 3, which is not really important.
15:41:30 <LordAro> _zephyris: i suppose *technically*
15:42:09 <_zephyris> I'm more worried about the memory not even being plugged in!
16:07:17 <_glx_> what I guess as the CPU behind it also look weird
16:17:19 <ahyangyi> Apparently GCC also optimizes "modulo 3" into a sequence of multiplication and various bitwise operations
16:36:08 *** Smedles has joined #openttd
16:43:19 <peter1138> _zephyris went to the effort of putting hyperlinks in OpenGFX2's readme file. But all add-on content is considered untrusted so none of the links are rendered or treated as such.
16:55:46 <_zephyris> Oh, that would explain it then... I had wondered what was going on, I assumed it was me.
16:56:13 *** gelignite has joined #openttd
17:18:05 <_glx_> ah indeed we only include english.lng for emscripten
17:21:15 <Rubidium> I already feared that when I wrote it, but alas... it first needed to build to be sure
17:21:58 <peter1138> Why is that? Lack of fonts?
17:24:30 *** Wolf01 is now known as Guest25061
17:25:19 <_glx_> and fontconfig, and freetype
17:25:36 *** Guest25061 has quit IRC (Ping timeout: 480 seconds)
17:25:39 <_glx_> so we decided it was better to have only english instead of possible ????
17:28:54 <michi_cc> truebrain: Have you seen the STUN issue? Might have solved itself by now of course.
17:28:58 <truebrain> Languages are big; websites should be small ๐
17:29:10 <truebrain> michi_cc: I did; I blame peter for saying it won't happen for another few years ๐
17:29:42 <truebrain> I restarted redis, solving the issue once more. But clearly it needs a deeper look ๐
17:40:16 *** gelignite has quit IRC (Remote host closed the connection)
17:43:17 *** gelignite has joined #openttd
17:55:59 <peter1138> So... if the requested width is narrower than the widest character in the string, the layouter will get stuck in an infinite loop. Probably then the loop should get if any progress is being made.
18:22:24 <truebrain> Did anyone mail orudge about the cert?
18:24:10 <truebrain> I will throw him an email ๐
18:32:30 *** SigHunter has joined #openttd
18:53:17 <truebrain> emailing orudge fixed the cert issue ๐
18:54:20 *** SigHunter has joined #openttd
19:12:28 <Rubidium> hmm... if we change strings in English, do those get to eints relatively soon or do we have to wait for the nightly eints-commit cycle?
19:12:50 <peter1138> Now need another excuse not to document station/roadstop flags.
19:13:52 <ahyangyi> "Someone already produced the documentation, but they used AI"
19:13:59 <_glx_> Rubidium: it's updated on next eints run
19:14:16 <_glx_> but a run can be triggered manually if required
19:14:39 <truebrain> Rubidium: eints only talks to OpenTTD repo on the push/pull happening every night. There is no other from/to (unless a human does it, ofc). We removed all other magic ๐
19:16:18 <Rubidium> so... I'll wait till tomorrow and not worry about the new language not showing up as translations just yet
19:17:09 <truebrain> Or trigger it manually; but yes, correct ๐
19:17:14 <LordAro> does the OTTD group need creating manually?
19:17:24 <truebrain> owh, there is a script for that I think ....
19:17:28 <LordAro> translator group in the OTTD org*
19:17:36 <dh1> does ottd store data about what languages are most used?
19:18:55 <truebrain> The `--generate-issue-templates`
19:19:00 <truebrain> but can only be done after eints ran
19:20:04 <LordAro> en_US, en & ru being the top 3 is surprising
19:27:33 <peter1138> Okay, something like that.
19:28:22 <peter1138> Haven't touched NML.
19:54:09 <peter1138> This is probably not exactly what these unicode sequences are meant to look like...
19:54:24 <peter1138> But it did automatically load *something*.
19:56:38 <_zephyris> Reminds me of when printers went haywire in the 90s
19:57:13 <peter1138> With Noto Emoji installed...
19:57:35 <LordAro> i guess the question is what it's supposed to be
19:59:22 <peter1138> Depends on the font, but generally a rainbow flag and a trans flag.
19:59:38 <_glx_> font detection using wrong encoding ?
19:59:49 <peter1138> No, it's working entirely correctly.
20:00:13 <peter1138> It's not a single unicode codepoint.
20:00:24 <_glx_> hmm blame symbol fonts then
20:00:53 <peter1138> The first was Symbola, which doesn't know to combine the sequence.
20:01:03 <peter1138> The second is Noto Emoji, which does.
20:01:16 <_glx_> I think windings can have the same issue
20:01:25 <peter1138> Noto Color Emoji does not work in OpenTTD as our font renderer was designed with the assumption that text is monochrome.
20:01:49 <peter1138> Wingdings is probably too ancient to have the glyphs anyway.
20:03:21 <_glx_> we skip SYMBOL_CHARSET fonts for win32 anyway
20:03:45 <peter1138> Dunno, I can only test the Freetype/FontConfig stuff.
20:08:00 <peter1138> Supporting 3 different native (well, FT/FC is not exactly native) font systems is a bit of a pain.
20:09:26 <peter1138> Hmm, well, with the icon fontcache system, I can draw colour sprites.
20:09:41 <peter1138> So potentially we could support colour fonts too. But.
20:11:41 <_glx_> hmm I can see `FC_SYMBOL` exists, so maybe that is true for emoji fonts
20:13:30 <peter1138> Anyway, it's not important.
20:14:46 <peter1138> Automatically loading CJK fonts if CJK characters are detected, without affecting fontsizes, is... quite nice.
20:15:12 <peter1138> (This is separate from language detection, which will affect fontsizes.)
20:15:54 <peter1138> (Also this bit isn't in that PR.)
20:16:37 <peter1138> truebrain, iirc the preview sites are never cleared up either?
20:16:53 <peter1138> (Thinking back to the size of them)
20:17:59 <truebrain> peter1138: Storage is cheap; but if you need 2 minutes to open the preview site, that is another story ๐
20:18:22 <truebrain> and in Emscripten world, you have the download all data
20:18:27 <truebrain> We can't load on-demand (yet?) ๐
20:29:36 *** bigyihsuan has joined #openttd
20:29:36 <bigyihsuan> peter1138: will that mean we won't get boxes/question marks/empty space when we use non-latin characters?
20:31:18 <bigyihsuan> if so,<a:yippee:1202488330688667729>
20:42:57 <peter1138> Hmm, there's a little bit of overlap of the non-latin glyphs.
21:03:55 *** raxitocl has joined #openttd
21:04:58 <raxitocl> hi, can someone help me? i'm creating a Music Set, but i don't know how to make them appear on the settings
21:11:39 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
21:13:29 <LordAro> raxitocl: how far have you got?
21:30:17 <raxitocl> i used openMSX as reference, and the .obm file are inside of the folder, but the game can't detect it
22:10:59 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:13:52 *** toktik has quit IRC (Quit: Ping timeout (120 seconds))
22:17:30 <_glx_> shortname in [metadata] section of the obm is what is used to detect all different obm
continue to next day โต
