IRC logs for #openttd on OFTC at 2024-05-12
⏴ go to previous day
02:15:03 *** ChanServ sets mode: +v tokai
02:22:06 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
02:52:46 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
04:03:41 *** XYZ_ has quit IRC (Ping timeout: 480 seconds)
04:41:38 <DorpsGek> - Update: Translations from eints (by translators)
05:11:45 <wensimehrp> 🤔 I wonder why unifont won't work
05:12:08 <wensimehrp> And why the initial font configuration works
05:12:21 <merni> I think there is some log output that shows which particular character is missing but idk what it is
05:12:43 <wensimehrp> > I'm afraid that the game automatically determines that the requested font is unuseable when there is only one or two glyphs missing, while the rest are present. A research shows that ~3500 characters are enough to cover over 99% of common text. The required character count might be higher for Traditional Chinese, but will only be lower for Japanese and Korean.
05:13:04 <merni> > I'm afraid that the game automatically determines that the requested font is unuseable when there is only one or two glyphs missing,
05:13:18 <merni> So far that is intended behaviour since there is no way to use multiple fonts at a time
05:13:25 <merni> Which is what peter is working on now
05:13:57 <wensimehrp> merni: Maybe zephyris's script would work?
05:14:41 <merni> If you mean font, Zephyris's OpenTTD fonts only cover Latin, Cyrillic and maybe Greek and Hebrew
05:15:30 <merni> Yeah you could use that with some modifications to point to the right font
05:18:05 <wensimehrp> wensimehrp: Maybe it's the vehicle icons that are missing 😄
05:18:41 <merni> I don't think so, most fonts don't have those
05:19:01 <merni> By most I mean all except openttd sprite fonts and Zephyris's fonts :P
05:19:59 <wensimehrp> Maybe... The game don't find the vehicle icons, and determines the font is "unusable...?"
05:20:06 <wensimehrp> Well just guessing
05:20:10 <merni> Then every font would be unusable
06:11:39 *** HerzogDeXtEr has joined #openttd
07:24:11 *** gelignite has joined #openttd
08:02:23 <rau117> andythenorth: Yeah like 3 parameters with unknown value.
08:02:23 <rau117> What exactly does Low, medium and high mean?
08:02:23 <rau117> Ah no, it's just x1 x1.33 x1.5
08:06:14 <rau117> Ehhh… if it weren’t for that change “cities ignore the ban on road construction in scenario mode”...
08:06:59 <peter1138> So actually 0.75x, 1x and 1.125x
08:10:18 <rau117> PeterNviaGitHub: By the way, what about not_allowing the city to rebuild player-built buildings? Or make some kind of switch on allow / not_allow to rebuild this building
08:10:27 <andythenorth> Is that integer maths? 😛
08:14:09 <peter1138> rau117: Yes, but for now I will leave that to another change. This PR simply allows placing houses, whereas adding that also requires changing town growth behaviour and setting flags in the map array. Probably not a big job but enough for a separate PR.
08:14:48 <peter1138> Tesco pain-au-choclats are mostly disappointing...
08:15:37 <peter1138> The problem is I am not in France at a local patisserie.
08:21:08 <peter1138> This is the one I used to cycle to in the morning as a boy.... still around 30+ years later. Tractor's probably the same too 😄
08:21:18 <andythenorth> Maybe we should emigrate
08:27:41 <peter1138> Did anyone test the "copy existing vehicle" property yet?
08:29:34 <peter1138> Hmm, well, I guess it would break 14.x compatibility (or need to be backported)
08:29:56 <peter1138> Also I shouldn't've been lazy with the property number. Hmm.
08:36:52 <rau117> Hmm, nml… btw how about adding possibility to reject/alter random on dates? Now the randomness on the production start and end date is applied in any case, it seems like ± 512 days. Is it possible to add some parameter for NML so that it is possible to either control this value (512) or simply disable it?
08:36:52 <rau117> Like... the train model should appear strictly on 1950-01-01 and disappear strictly on 1955-06-06. At the moment, this is impossible to achieve such precision
08:36:52 <rau117> (whether this is necessary or not is another question)
08:56:50 <peter1138> If aystar is being kept it could do with a big RAII cleanup.
08:57:36 <peter1138> Like at least a destructor that calls Free() so it doesn't need to be done manually...
09:04:19 <Rubidium> I'm currently rabbit holing that area... doing river generation with yapf seems not to be a good fit, as that's heavily influenced by vehicle types and the likes. Then looking at what Free() does, and thinking the hash/binary heap should just be C++ objects in their own right which would remove the need for free. But that brings me to the binary heap... of which we got two... both referencing the same
09:04:25 <Rubidium> documentation and arguably both doing essentially the same thing I guess
09:05:06 <Rubidium> though there's also a std::priority_queue, so... how much effort is using that going to be?
09:05:07 <peter1138> Yes, well, rabbit hole is where I left it just now 🙂
09:14:23 <andythenorth> peter1138: I did not
09:15:48 <andythenorth> looks like the nml patch wouldn't be complicated?
09:16:46 <peter1138> Not sure, maybe NML has some required properties or something.
09:19:16 <andythenorth> ach nml has the magic ID assignment mistake
09:19:29 <andythenorth> so what value should prop 0x80 be if that's used by the author?
09:19:35 <andythenorth> there's probably some way to resolve that
09:25:18 <peter1138> Hmm, no, that shouldn't be necessary either. 😮
09:26:22 <peter1138> Value? It should be the engine ID you assigned. Same as used in the variant property.
09:56:57 <peter1138> Unlike my change which has an invalid free 😄
10:00:49 *** gelignite has quit IRC (Read error: Connection reset by peer)
10:13:29 *** gelignite has joined #openttd
10:28:09 <_glx_> andythenorth: Same as any other references to item IDs, nml knows how to translate to the real ID auto assigned or not
11:14:17 *** gelignite has quit IRC (Read error: Connection reset by peer)
11:27:16 <peter1138> Hmm, my sprite count guestimator says there are 2000 more sprites than needed (without NewGRFs)
12:04:07 <Beer> Debugging with facility net on level 9, going back to that "Client #{} (IP: {}) is dropped because it fails to send valid acks." error. I see Receive_CLIENT_ACK() coming from my client every 2 seconds
12:04:35 <Beer> Why would the token be invalid while clearly network communication is going though?
12:05:15 <Beer> Again, client actions are registered server side : creating a company and joining it is possible, and is still there if I reconnect after having been kicked out
12:10:16 <Rubidium> is either the server or client self-compiled?
12:15:43 <Rubidium> in any case, the server sends a challenge to the client every now and then, and if the client doesn't reply correctly to that you'll get such an error. Depending on the compilation flags there could be some debug fields in the network packets related to this, which makes server and client compiled with different settings potentially fail in such a manner
12:15:54 *** jenkings has quit IRC (Quit: User went offline on Discord a while ago)
12:17:10 <Beer> Yup it's compiled. But I have been doing this for a long time now, and I am not having this issue in 13.4
12:17:29 <Beer> Wondering what has changed in the netcode >= 14.0 triggering this
12:17:41 <_glx_> and you see "Sent ACK at" in client log ?
12:18:13 <Beer> I was asking if there was a client log yesterday, because I can't find it
12:18:28 <Beer> The client is merely isntalled from the official installer
12:18:46 <_glx_> same as server, needs high enough debug level
12:19:14 <Beer> Ok I'll start it from CLI
12:19:58 <Beer> 7 on the net facility I suppose
12:21:13 <_glx_> and there's `Debug(net, 9, "Client::SendAck()");` right when client send it so 9 might be better
12:21:59 <Beer> YEha, Receive_ACK is on net:9 too
12:22:16 <peter1138> Do you see Client::SendAck?
12:23:10 <Beer> Would be strange for the server to show Receive_CLIENT_ACK without the client ever sending it though, wouldn't it? :\
12:27:01 <peter1138> Sure but given it's not working...
12:28:07 <_glx_> "Sent ACK at" on client and "Receive_CLIENT_ACK" on server should show the same number
12:28:57 <Beer> Is there a possibility machines clock running on different timezones might have an impact?
12:30:44 <Beer> Client "Send ACK at ***" numbers match the server "Receive_CLIENT_ACK(): frame=***" messages
12:31:36 <_glx_> so one part of the data is correct 🙂
12:33:31 <peter1138> Have you posted the logs in your issue report?
12:36:47 <Beer> Haven't created an issue yet: nothing stands out and I'd like to wage an "It (doesn't) Work(s) For Me" war :oD
12:37:00 <Beer> I'm compiling Debug stances in to show values
12:37:44 <_glx_> hmm is server compiled with ENABLE_NETWORK_SYNC_EVERY_FRAME or NETWORK_SEND_DOUBLE_SEED ?
12:38:35 <Beer> cmake .. -DOPTION_DEDICATED=ON && make
12:39:14 <Beer> No idea on the default value: I trust the project's building pipeline
12:43:42 <_glx_> yeah add some debug output in `ServerNetworkGameSocketHandler::SendFrame` (to see what token server is sending), some in `ClientNetworkGameSocketHandler::Receive_SERVER_FRAME` to see what client receives and send back, and in `ServerNetworkGameSocketHandler::Receive_CLIENT_ACK` for the token received from client
12:46:01 <Beer> OK I'll do that next. Atm I'm digging the condition cs->last_frame_server - cs->last_token_frame >= _settings_client.network.max_lag_time
12:46:13 <Beer> Trying to see if I can output the diff'ed values
12:48:16 <_glx_> both values are set in `ServerNetworkGameSocketHandler::Receive_CLIENT_ACK`
12:55:59 <_glx_> anyway this mechanism has not been touch for 13 years 🙂
13:11:21 <Beer> Well I suddenly see SendSync server side and Reseive_SERVER_SYNC on client side
13:11:30 <Beer> That was definitely missing beforehand
13:11:37 <Beer> Still trying to understand why
14:25:12 <peter1138> Ah the well known f & g parameters.
15:18:44 <Beer> Could -DCMAKE_CXX_FLAGS_INIT="-DRANDOM_DEBUG" -DCMAKE_DISABLE_PRECOMPILE_HEADERS=ON options have an impact _glx_?
15:19:42 <Beer> Different ways of compiling OpenTTD create (or not) the problem
15:23:28 <_jgr_> Using RANDOM_DEBUG will make your build network-incompatible with normal builds
15:39:54 <Beer> I wonder why those options got burnt into the compiling process
15:40:09 <Beer> Were they needed sometime in the past?
15:41:13 <_glx_> RANDOM_DEBUG enables the 2 other defines I mentioned
15:41:39 <peter1138> Unless you were trying to debug a desync error....
15:41:59 <_glx_> and if client and server don't use the same then you'll have a mismatch when getting the token
15:44:36 <_glx_> Beer: are you copying flags from CI ?
15:44:40 <Beer> I have no idea where they came from
15:44:41 <peter1138> Ah ha, that's why FindMissingGlyphs does not work now 😄
15:44:57 <_glx_> because we use them there to check nothing is broken
15:45:20 <Beer> _glx_: Ah! That may explain their presence
15:45:22 <_glx_> CI dedicated does extra-cmake-parameters: -DOPTION_DEDICATED=ON -DCMAKE_CXX_FLAGS_INIT="-DRANDOM_DEBUG" -DCMAKE_DISABLE_PRECOMPILE_HEADERS=ON
15:45:38 <Beer> I seriously do not remember where the ycome from. I must have put them there ages ago
15:45:49 <Beer> _glx_: Yup that's a perfect match
15:45:54 <_glx_> but it's a test build, not meant to be used outside
15:46:21 <Beer> That must be it. Keeping OPTION_DEDICATED and getting rid of the rest
15:46:55 <_glx_> don't forget CMAKE_BUILD_TYPE=Release or RelWithDebInfo too
15:47:53 *** Wormnest has joined #openttd
15:48:03 <Beer> I'm not compiling a package: I'm using make install and copying the installation in a multi-stage container build
15:48:17 <_glx_> you still want a non debug build 🙂
15:48:18 <Beer> THe version is managed through the Git tag
15:48:55 <Beer> Where is RelWithDebInfo set? What's its use?
15:49:11 <Beer> Are those options documented in the compile docs? I don't remember them
15:49:16 <peter1138> It's set... with `CMAKE_BUILD_TYPE`.
15:49:41 <peter1138> It's documented in COMPILING.md.
15:50:02 <peter1138> (But indeed, I don't remember them either :D)
15:51:44 <andythenorth> cooperative multitasking?
15:52:18 <peter1138> Ahh, the old Ransom Note effect.
15:52:36 <Beer> -DCMAKE_BUILD_TYPE=RelWithDebInfo or -DCMAKE_BUILD_TYPE=Release then?
15:52:59 <Beer> I have no idea on their implications
15:53:04 <peter1138> The difference is basically unstripped vs stripped.
15:53:36 <peter1138> Unstripped means it still has debug symbols so it's more useful if there's a crash.
15:53:46 <_glx_> peter1138: yeah looks like letters from multiple newspapers 🙂
15:54:10 <Beer> SO it depends on the forsaw ability to run them through gdb, then
15:54:26 <peter1138> Mixing works good for the CJK stuff, not for Vietnamese...
15:54:56 <peter1138> Releaes is stripped, yes.
15:55:22 <peter1138> They look mostly like latin characters, maybe that's another task for _zephyris 😄
15:56:41 <_glx_> yeah they could go in OpenTTD fonts
15:57:24 <_glx_> some glyphs already are in it seems
15:57:40 <Beer> Might make sense that COMPILING.md mentions Release and not RelWIthDebInfo then, or at least explain the former option
16:52:32 <truebrain> _glx_: Seems you shouldn't give users options 😛
17:03:42 <_glx_> well it's just a "known" cmake thing
17:04:10 <peter1138> Oops, missing my fallback characters :S
17:04:31 <peter1138> Intentionally picking weird fonts that don't include everything...
17:07:02 *** SigHunter has joined #openttd
17:10:51 <truebrain> Lol .. it could be anything!
17:15:48 <peter1138> I think in this case, the font really does just have blank glyphs for cyrillic. That's anti-social!
17:16:41 *** _zephyris has joined #openttd
17:16:41 <_zephyris> peter1138: Oof, looks like a difficult problem. I'll have a ponder though
17:17:22 <_zephyris> Yeah, still lurking, just working on a house move!
17:17:32 <wensimehrp> They are Latin characters
17:18:23 <peter1138> Yeah, it does work.
17:18:34 <truebrain> today I fell in the rabbithole of font licenses ..... I so hate this world a tiny bit more now. So difficult to figure out how or what, and which font is published where, and what license, and blablabla .. ugh
17:18:36 <peter1138> But if the font says it includes glyphs but they're blank...
17:19:01 <wensimehrp> wensimehrp: > The term CJKV also includes Vietnamese, which was also historically written with Chinese characters.
17:19:12 <truebrain> if you use Adobe Creative Cloud, you can only use a font for as long as you have a subscription there .. and you can't host it yourself (for webpages) .. you have to use their CDN ... spy-much? 😛
17:22:32 <peter1138> Better or worse than Comic Sans MS?
17:24:01 <peter1138> Hmm, I guess I could look at that "bug" report from earliy.
17:24:22 <_glx_> peter1138: so fonts do like dummy 8bpp in newgrf
17:24:47 <peter1138> Those fonts aren't normally usable, because they miss other glyphs that OpenTTD requires.
17:25:08 <peter1138> But as that is no longer the case, they are loadable... and load junk :S
17:25:55 <peter1138> I could potentially also check whether each glyph has anything in it, but I imagine that's likely to be heavier than just checking if it exists.
18:11:52 <peter1138> 1) sprite is always loaded
18:11:52 <peter1138> 2) "DejaVu Sans" is a fallback font, loaded because it has the glyphs required for Arabic (the current language)
18:12:29 <peter1138> 3) "OpenTTD *" is always loaded after a fallback font.
18:12:29 <peter1138> 4) "Arial" is the manually configured font, so loaded last.
18:30:52 <peter1138> Hmm, so can we just include Unifont as a backup? :p
18:32:08 <peter1138> Ah it would override the sprite font currently. Hmm.
18:34:46 <peter1138> css is still unfinished.
18:35:26 <andythenorth> radius-top-left: 16px
18:47:07 <andythenorth> Maybe for OpenTTD 16
19:01:49 *** XYZ has quit IRC (Ping timeout: 480 seconds)
19:23:51 <_zephyris> Multiple combining diacritics is a pain... Really don't know how to fit that into the available space.
19:24:17 <_zephyris> That's why it ends up ransom note, depending on the renderer it shifts the composite glyphs to try to fit everything
19:24:45 <peter1138> No, it was a ransom note because I'm changing the font system to allow multiple fonts 🙂
19:31:09 *** nielsm has quit IRC (Ping timeout: 480 seconds)
19:36:33 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
19:39:53 *** Beer has quit IRC (Quit: Leaving)
19:41:24 <_zephyris> Ah, then my dev version has the same issue!
19:47:16 <peter1138> Urgh, owning vs non-owning pointers eh...
19:58:35 <peter1138> Hmm, I think os_handle isn't actually used on OSX.
19:59:22 <peter1138> `LoadCoreTextFont()` treats it as a prefetch font descriptor, but `SetFontNames()` (the only thing that can set it) always passes nullptr.
20:21:10 *** gelignite has joined #openttd
21:00:37 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:06:25 <peter1138> Ah ha, that was it 😄
21:20:31 *** keikoz has quit IRC (Ping timeout: 480 seconds)
21:41:34 *** gelignite has quit IRC (Quit: Stay safe!)
21:59:29 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:08:41 <klote> Hi i changed the minutes per year to 22 on my server
23:09:28 <klote> The maintenance for trains/ships/vehicle is now out of sync "i think"
23:09:49 <klote> what would i need to set these to... with a 22 minutes per year setting.
23:10:31 <klote> just double everything?
23:13:23 <klote> I dont really understand default valeu = 150 days...
23:13:40 <klote> but default shows up as 5 days...
23:15:12 <klote> and it wont let me set it higher then 30 days
23:17:04 <klote> if i enable percentage mode it also doesnt let me go higher then 90 days
23:25:10 <talltyler> klote: It’s not in days. You’re either choosing minutes or a percentage of the engine’s maximum reliability.
23:25:37 <talltyler> klote: So this is default of 5 minutes, maximum of 30 minutes.
23:25:58 <talltyler> The Settings menu does not show the right units but vehicles do
23:26:11 <peter1138> It's clearer in master.
23:26:28 <klote> So what would i have to set it to with 22 minutes per year?
23:26:47 <peter1138> How often do you want them to service?
23:27:00 <talltyler> Minutes per year has no effect on reliability decay, so you don’t need to consider scaling.
23:27:17 <klote> peter1138: the same scale as default when 12 minutes per year.
23:27:28 <talltyler> It would be approximately the same as it’s always been, considering 1 month is 1 minute 🙂
23:28:14 <klote> pHIllIPh(109.43.48.238) have reported an issue on OpenTTD-BuildEurope:
23:28:14 <klote> now time of service train is 150 minutes but one year is 400 seconds, time of going now is in time, before was in days in game
23:28:15 <talltyler> The default is 150 days, which is 5 minutes
23:28:34 <klote> so what does he mean by that then?
23:28:55 <peter1138> Pretty sure there is a settings bug somewhere, as I see 150 minutes by default.
23:30:23 <talltyler> peter1138: Oh yeah, I see that too 🤔
23:30:39 <peter1138> Not sure if it's left over old settings or what.
23:31:19 <talltyler> Yeah, I think it’s leftover somehow. I had a similar issue when we changed the units for linkgraph updates.
23:31:47 <talltyler> (Which of course I didn’t report 🙃)
23:31:49 <peter1138> Hmm, does a server's settings take priority over a player's settings?
23:32:34 <talltyler> Service intervals are a company setting, so who knows lol
23:33:00 <talltyler> Default service intervals are stored in saves
23:33:21 <talltyler> klote: Yes, click Default to reset it to whatever it’s supposed to be
23:33:30 <talltyler> Which would be 5 minutes
23:33:32 <peter1138> Okay, well multiplayer server + my friend's company who has never played the game before 14.0... 150 minutes.
23:34:10 <peter1138> Server is set to %, 50% for all. So it's not using the server's settings.
23:34:44 <peter1138> Hmm, but then I just joined and made one and it's 5 minutes
23:35:02 <peter1138> made one -> started a company and built a train.
23:36:24 <peter1138> Hmm, I need a tool to dump widget trees :S
23:37:31 <peter1138> Or I should just ban OpenGFX.
23:45:14 <wensimehrp> peter1138: which version? 😛
23:46:12 <talltyler> I only saw it when testing the house picker in the preview build 🙂
23:49:34 <peter1138> I checked and those didn't have SetFill before, so... weird.
23:53:12 <peter1138> A good test for these is Greek.
continue to next day ⏵
