IRC logs for #openttd on OFTC at 2024-03-24
โด go to previous day
00:49:29 <talltyler> A nice Steam comment ๐
00:50:33 *** notluke2578 has joined #openttd
01:33:22 <johnfranklin> For a whole day, I tried installing gnome desktop on WSL2, but always failed.
01:43:11 <wensimehrp> ^ not worthwhile at all
01:43:18 <_glx_> wsl2 is not meant to run a full desktop
01:43:45 <_glx_> but it can run GUI apps through windows
01:48:27 *** ChanServ sets mode: +v tokai
01:55:17 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
02:00:13 *** Wormnest_ has quit IRC (Quit: Leaving)
03:26:16 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
04:00:25 *** Leopold_ has joined #openttd
04:02:20 *** Leopold has quit IRC (Ping timeout: 480 seconds)
04:41:14 <DorpsGek> - Update: Translations from eints (by translators)
04:49:38 <johnfranklin> Finally solved, it is the problem of Windows scaling
04:56:22 *** Leopold___ has joined #openttd
05:02:50 *** Leopold_ has quit IRC (Ping timeout: 480 seconds)
05:03:38 *** Leopold___ has quit IRC (Remote host closed the connection)
05:08:44 *** Leopold has joined #openttd
06:53:56 *** ufo-piloot_ has joined #openttd
06:54:04 *** ufo-piloot has quit IRC (Read error: Connection reset by peer)
07:31:45 <Rubidium> it's Sunday, so that means a nice roast of some sort, right?
07:41:54 <merni> michiccviaGitHub: Hm, is it sending missed messages from yesterday?
09:01:26 <truebrain> merni: That is what Discord tends to do .. their webhooks are queued, and then something breaks, but the queue persists .. so when they fix it, it starts to dequeue slowly.. it is still not caught up ๐ฆ
09:02:15 <truebrain> Well, maybe it is, hmm
09:05:10 <truebrain> Seems also some IRC messages weren't bridged. But they too are now caught up ๐ just in a weird order ๐
09:37:27 <rau117> rubidium42viaGitHub: By the way, may I know how much more difficult his system will be to use?
09:37:27 <rau117> The only advantage of the current password system is that they are stupidly simple. You can easilly use something as simple asโฆ `ะ` and have a certain degree of protection
09:37:27 <rau117> But will it be that easy to add a player to the allowlist? Do I need to be in the company itself to add a player to the allowlist?
09:37:27 <rau117> For example, this situation: a friend went to bed and left the password for the company in Discord, so I can connect to him just like ctrl+v, without any additional mental operations. But what about allowlist, do we need to meet in-game?
09:37:27 <rau117> Or is it just being added later, and I'm going too far?..
09:49:36 <truebrain> there is a lot more code involved in passwords than I assumed, lol ๐
09:49:55 <Rubidium> rau117: those are all things that could be added in some form or another at a later stage
09:50:53 <Rubidium> rau117: but with the current PR you do need to be both logged in at the same time
09:59:58 <peter1138> * Type of the pool for link graph components. Each station can be in at up to
09:59:58 <peter1138> * 32 link graphs. So we allow for plenty of them to be created.
10:00:10 <peter1138> Hmm, 32 is probably the old cargo count, which is 64 now...
10:01:01 <peter1138> I wonder if the LinkGraph limit is ever reached...
10:01:38 <andythenorth> hmm....station slope calculation is different to objects? Is it rotated?
10:04:57 <peter1138> It's swapped depending on the axis of the station.
10:05:31 <michi_cc> peter1138: Very unlikely, especially stations ususally aren't the exclusive owner of each link graph.
10:06:06 <andythenorth> this is not right
10:21:30 <andythenorth> ok, so I won't be using one template for roadstop / object / rail station foundations ๐
10:21:40 <andythenorth> stations, don't ๐
10:29:06 <truebrain> sorry Rubidium if you didn't know GitHub just yells to everyone what your key is; it is useful and a bit annoying at the same time
10:29:28 <truebrain> a lot of cloudinit works with it, where if you spin up a VM you can point to your GitHub account, which automatically adds your key to the authorized list
10:35:15 <andythenorth> ok so stations are like all the ctrl-click livery stuff in Horse? ๐
10:35:15 <andythenorth> I implement something complicated and BAD, then the grf spec gets updated?
10:45:23 <xarick> so much dirt in the sky coming from north africa
10:46:27 <xarick> sky looks pale yellow, like a fog, but of dirt
10:47:19 *** nielsm has quit IRC (Ping timeout: 480 seconds)
10:47:54 <xarick> bad for my system's fans ๐ฆ
10:58:16 <michi_cc> I guess the client auth key should then simply not be called public key, but something like client key or so, to make the intentions clear.
11:02:14 <truebrain> technically it still is ofc a "public key", just how these keys are called ๐
11:02:49 <truebrain> but the problem at hand is the "do not track"; so keeping the public keys private (ghehe) for now, seems perfectly reasonable. This is already a lot of work, so small steps ๐
11:03:48 <michi_cc> Two buttons, "Join", "Join anonymously"?
11:04:38 <truebrain> like I wrote in my last comment, you also often see a switch between "Privacy" and "Public", where the first just generates another key per server
11:04:43 <truebrain> but your idea also works
11:04:47 <truebrain> can even ben an extension ๐
11:04:52 <michi_cc> Probably OTTD would then generate a new key and store it somewhere for the specific server key.
11:05:01 <michi_cc> Yes, what you wrote ๐
11:05:35 <truebrain> but yeah, I am already extremely happy if we can remove company passwords
11:05:54 <truebrain> so, baby steps ๐
11:06:07 <truebrain> (well, there is nothing "baby" about this step, but you get what I mean)
11:08:20 <xarick> wondering if I can abuse Try/Catch with Valuate
11:11:04 <xarick> this AI "crashes" but not really
11:22:21 <Rubidium> truebrain: those bastards! :(
11:22:31 <truebrain> in reply to what part? ๐
11:22:45 <Rubidium> github sharing public keys
11:23:11 <truebrain> you need to be able to verify if a commit signed by such key, is actually the one of the username you thought it was ๐
11:23:38 <truebrain> well, that would explain the GPG key, not sure why the SSH keys are there ๐
11:25:06 <truebrain> these conversations always really reminds me of IPv6 ... "assign a permanent unique IPv6 to every machine" - "shit, now ad-networks can track you in very high detail" - "quick, rotate the IPv6 every N hours" (which Windows does)
11:25:54 <truebrain> you have a fixed IPv6, and a temporary; the latter is used for outgoing connections, so once every N hours your IP changes for a new connection
11:47:22 <frosch123> i guess i'll use "codefix" for minor "docfix"es
11:55:58 <truebrain> `You should always develop a game from your own perspective, and if you dont play it anymore you should not be working on it. `
11:55:58 <truebrain> Steam has spoken, we had a good run, but I guess this is this.
11:56:42 <andythenorth> play again y | n?
11:59:01 <frosch123> hmm, i did not know the script api can return ``std::optional<std::string>``
11:59:44 <talltyler> We get the loudest complaints anytime we develop from our own perspective ๐
11:59:57 <talltyler> โI donโt use this feature, so I deleted itโ
12:00:19 <andythenorth> if we're triggering a small number of complaints from each end of the bell curve, we're doing it right
12:00:52 <andythenorth> ach there must be a var for station orientation?
12:00:53 <talltyler> Asking โWhat kind of data are you interested in?โ shows that they didnโt probably read the post
12:00:59 <andythenorth> they didn't read the post
12:01:29 <talltyler> Time to stop thinking about it
12:02:19 <andythenorth> we have the bingo card complete
12:02:24 <andythenorth> - spyware: check
12:02:37 <peter1138> > Many thanks but RC2 and RC3 are not working on macOS - any one else have this issue? (Already raised on bugtracker)
12:02:42 <andythenorth> - good design doesn't need evidence: check
12:02:57 <andythenorth> oh wait, nobody said "it should be opt-out"
12:03:01 <andythenorth> we didn't get bingo yet
12:04:22 <andythenorth> starts fine for me
12:05:33 <frosch123> another bingo item: can you trigger someone with this:
12:05:58 <truebrain> I identify as an invalid gender ๐
12:09:11 <rau117> In 14 rc3, fruits are labeled FRU**I** instead of the usual FRU**T**
12:09:11 <rau117> Is this change intended? This will break all old/unsupported newgrfs.
12:09:11 <rau117> *(that is, Iโm too lazy to go to GitHub, if necessary, Iโll make a proper bug report)*
12:17:46 <andythenorth> frosch123: yes, can definitely trigger someone
12:18:05 <andythenorth> we should just add 'random'
12:18:27 <andythenorth> the faces thing is such a minefield of identity issues anyway ๐
12:19:56 <andythenorth> station foundations are weird
12:20:15 <andythenorth> rail tiles are wrong for one orientation; non-rail tiles don't work at all
12:20:28 <frosch123> does someone know the difference between ``ScriptErrorType`` and ``ScriptError::ErrorMessages`` ?
12:20:42 <andythenorth> I refuse to learn the actual station spec, I'm just going to keep acting surprised that stations don't act as they should
12:22:39 <andythenorth> frosch123: is one mapping OpenTTD errors, and the other arbitrary?
12:27:28 <_glx_> Seems they are the same
12:29:48 <rau117> michi_cc: hehehe, up to date (=
12:29:52 <truebrain> frosch123: the first is an internal number assigned to an error, the latter is the error array of ScriptError, if I am not mistaken. ScriptTile for example also has errors
12:31:20 <frosch123> i was wondering why there were so few error messages
12:31:34 <frosch123> but there are tons of ErrorMessages enums in other classes
12:31:47 <xarick> Abusing valuate, is this acceptable?
12:32:00 <truebrain> and during generation, funky things are done to them ๐
12:32:13 <frosch123> i'll expand the documentation ๐
12:32:27 <frosch123> only a small futher step in this rabbit hole
12:32:49 <truebrain> the internal number is btw set by the first entry in the enum
12:32:51 <_glx_> xarick: What are you doing ?
12:32:53 <truebrain> `ERR_TILE_BASE = ScriptError::ERR_CAT_TILE << ScriptError::ERR_CAT_BIT_SIZE` and friends
12:35:41 <andythenorth> ach how do I get the station orientation?
12:35:57 <truebrain> `0 org.openttd.openttd 0x00000001023bcafb __cxx_global_var_init + 11`
12:35:57 <truebrain> `Exception Codes: KERN_INVALID_ADDRESS at 0x00000000023b4000`
12:36:06 <truebrain> that MacOS crash is weird ... it seems it can't even load OpenTTD
12:36:21 <truebrain> can I just blame 10.15.7?
12:37:54 <andythenorth> ha I'm just going to not do stations
12:38:03 <andythenorth> so many other newgrf projects, stations are stupid
12:38:06 <truebrain> seems it lost a `1` in the address reference .. very odd
12:38:52 <andythenorth> I'm opening Steam ๐
12:39:04 <andythenorth> obviously it has about 5GB of updates to itself to do first
12:40:18 <_glx_> andythenorth: You don't
12:42:51 <truebrain> that isn't good ...
12:43:00 <truebrain> also when you manually navigate to the steam folder and open it?
12:43:43 <andythenorth> if I click past the 'is damaged' warning, it boots
12:44:16 <truebrain> also when you go to the steam folder yourself, and look at the app?
12:44:31 <andythenorth> yes, that's where I can run it from
12:44:41 <andythenorth> I'm looking in package contents for anything obvious
12:44:43 <andythenorth> but not seeing much
12:45:03 <truebrain> the Info.plist should also be there, the Steam depot tells me
12:47:23 <truebrain> lol @ steamclient.dylib .. thought we checked that, but .. yeah
12:47:35 <truebrain> if you remove the social plugins, does that make any difference?
12:50:21 <andythenorth> yes, boots with no warning
12:50:30 <andythenorth> icon still missing but eh
12:50:58 <truebrain> well, that is a bit of a pita ... as we would need to ship a steamclient.dylib, signed by us, for that to work ๐
12:51:13 <truebrain> can't be arshed to look into this today; I will redraw the steam plugin for now, should fix the issue I guess
12:51:46 <andythenorth> andythenorth: thanks ๐ I'm going to skip stations for now I think
12:52:08 <andythenorth> _glx_: thanks ๐ I'm going to skip stations for now I think
12:52:58 <_glx_> Stations are not hard, they just don't work like objects/industry
12:55:29 <truebrain> andythenorth: I just pushed a change to Steam .. if you close and start Steam again, and let it update, you should no longer get social plugins
12:55:32 <truebrain> am I right in that?
12:56:57 <truebrain> I don't see how this works for any other application btw; any game integrating with Steam should run into this issue ๐
12:57:35 <peter1138> rau117: Not intended. Any other changed?
12:59:46 <andythenorth> _glx_: I can't even begin to understand how to add custom foundations for stations ๐
13:00:25 <peter1138> Think that's the only one.
13:00:43 <andythenorth> I know why it's "not just in the spritelayout"
13:00:55 <andythenorth> I'm trying to do it in the spritelayout
13:01:08 <andythenorth> because my templating is unified across objects, roadstops and rail stations
13:01:32 <_glx_> you can't, remember stations spritelayout are fully static
13:02:11 <andythenorth> yeah, this is the stuff that makes me conclude stations are a waste of time developing for ๐
13:02:24 <andythenorth> I'm going to stop CHIPS for now
13:02:58 <andythenorth> roadstops stuff was quite interesting though
13:03:25 <_glx_> roadstops are simple, they use the standad 123
13:03:49 <andythenorth> truebrain: RC3 now launches from Steam
13:04:12 <rau117> peter1138: Well, from vanilla cargoes I checked everything and everything except fruit works as it should... *at least in xussr.*
13:04:19 <andythenorth> the modified dates are weird in steam dir
13:05:15 <_glx_> xarick: ah yes it's not an abuse, scripts are free to handle errors
13:05:28 <truebrain> andythenorth: maybe date you first downloaded? I dunno
13:05:34 <truebrain> or when we first uploaded that folder, also possible
13:05:39 <truebrain> we merge every new release into the old version
13:06:44 <_glx_> xarick: but your test doesn't actually do anything at all, so it's just eating CPU for nothing
13:06:46 <andythenorth> oof I am bad at walking away from unfinished problem, now I'm trying to figure out how to resolve the station slopes
13:06:56 <andythenorth> it must be possible in varact 2
13:07:29 <xarick> it's pseudo doing GSBase.Rand() to tiles
13:07:44 <_glx_> yes and you remove them at the end
13:08:13 <xarick> whole_map isn't required
13:08:50 <_jgr_> Passing the entire list of map tiles to Valuate, or even storing them a list at all, is just a bad idea
13:09:42 <_jgr_> Script lists are inefficient, and should not be used for large numbers of items
13:09:47 <_glx_> well it's the usual worst case scenario test ๐
13:10:19 <andythenorth> maybe we need to teach xarick about pareto or 90th centile approaches
13:10:27 <_glx_> but indeed ScriptLists should only be used when communicating with API
13:10:30 <andythenorth> or the normal distribution and 2 standard deviations
13:11:48 <_glx_> allowing authors to build ScriptList objects was probably a bad idea ๐
13:12:16 <andythenorth> most of current GS is a bad idea ๐
13:14:56 <xarick> this might be the trick I need to speed up pathfinding, if I can repurpose Valuate to do the progress
13:16:29 <xarick> but there are some DoCommands in it, ๐ฆ
13:16:40 <xarick> maybe I can't do what I think
13:17:23 <xarick> also, the stalls ๐ฆ unbearable
13:17:29 <_glx_> your pathfinder uses ScriptList to store data ?
13:17:55 <xarick> i have number of iterations
13:19:22 <xarick> I was thinking using Valuate to run the progress of pathfinder, which runs for a certain number of iterations.
13:19:37 <_glx_> valuate is not made for that
13:20:12 <xarick> the list would contain the result maybe, 0 no path, 1 path found
13:20:28 <xarick> if I find an item with path found, they i know i have already stopped
13:21:26 <xarick> the real work would be done inside the valuator function
13:21:52 <_glx_> valuator function must be short
13:27:44 <_glx_> the intended usage for script lists is ask data from API (possibly with prefiltering for some of them since 14), apply valuator, keep/remove, decide what to do based on the list content
13:30:36 <_glx_> andythenorth: for me it is
13:40:19 <_glx_> andythenorth: looking at nml documentation I can see the fun with foundations, for industry tile it's a CB to draw or not default foundations, for object it's a flag, and for stations it's a CB to tell which spriteset to use to replace default
13:45:05 <peter1138> GRFv9 tile layouts for stations "when" ๐
13:45:25 <_glx_> and I don't see any mention of foundations in roadstops
13:49:16 <_glx_> most likely it's built over road, so it doesn't care about foundation
13:50:14 <andythenorth> peter1138: I subscribe to your newsletter
13:50:30 <andythenorth> grf v9 rail stations
13:50:56 <andythenorth> dare we also just have 'station decoration' tiles? split from rail stations?
13:52:23 <_glx_> you don't have to check landscape yourself at all
13:55:24 <andythenorth> I think I am just doing it the hard way
13:56:04 <andythenorth> I am trying to generate all the stations and objects out of a single definition
13:56:16 <andythenorth> and, as far as possible, unified nml templating
13:56:43 <andythenorth> the 'correct' way to do this is to manually provide the foundations within the spritelayout, like other methods
13:56:56 <_glx_> station and object spritelayouts are similar but different
13:57:14 <andythenorth> using the archaic 'OpenTTD magically does magic' route is....possible...but I'd have to completely fragment this part of the compile
13:58:00 <andythenorth> it's made more complex because the foundations are only used for coast tiles
13:58:16 <andythenorth> so I'd have to somehow return the base set foundations in the callback for non-coast tiles
13:59:27 <talltyler> I donโt know Python but can you make a base class that stations and objects inherit and then do different things with?
14:09:48 <andythenorth> ok let's fragment rail stations from other stations first
14:10:56 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
14:17:11 <_zephyris> andythenorth: Well, base set support for different sea level and non-sea level foundations would be nice...
14:20:09 <_zephyris> And, seconded, nml const sounds very useful!
14:22:58 <frosch123> ``base_sprite_foundations`` and ``base_sprite_shores`` do exist in NML
14:23:59 <andythenorth> ok so with stations, do I have to flip all slope checks depending on track orientation?
14:24:32 <andythenorth> e.g. I'm using nearby_tile_water_class
14:24:47 <andythenorth> so is X and Y transposed in that case?
14:26:01 <_glx_> nearby_tile use "The first parameter is the distance to move along the platform, the second parameter is the distance to move sideways (between platforms). Both parameters are signed numbers with range -8..7, negative numbers go northwards."
14:26:15 <_glx_> X/Y handling is automatic
14:28:23 <_glx_> so nearby_tile_something(0, 1) and nearby_tile_something(0, -1) checks tiles next to the platform, independant of orientation
14:29:35 <xarick> oh noes, it doesn't work
14:31:35 <andythenorth> so the first thing I need to do is find the slope, water can wait
14:32:15 <andythenorth> so first I need to find the station orientation
14:32:31 <andythenorth> is that via `rail_continuation`?
14:32:35 <andythenorth> what if there's no rail?
14:33:47 <_glx_> you don't need station orientation in most cases
14:33:47 <andythenorth> oh do I have to read `tile_type`?
14:34:03 <andythenorth> how do I invert the slope var results without station orientation?
14:34:20 <xarick> Where do I have these DoCommands?
14:34:34 <xarick> it could point me to one, but no, it's just a generic error
14:35:23 <_glx_> does your valuator tries to build stuff ?
14:35:53 <xarick> it tests if it can build bridges and rails
14:36:51 <_glx_> test mode is command without DC_EXEC, but it's still a command
14:37:23 <xarick> I didn't wanna modify the pf internally
14:38:09 <_glx_> using valuator for pf is definitely no right ๐
14:38:48 <_glx_> you are not using pf libraries ?
14:46:12 <andythenorth> can we have console logging for grf?
14:46:28 <andythenorth> I really want to know what values stations return for slope var
14:46:37 <andythenorth> I guess I could select debug sprites
14:46:52 <andythenorth> industries at least can write to storage for debugging
14:57:47 <andythenorth> ok well let's see if the var even works ๐
14:57:59 <andythenorth> I strongly suspect it doesn't
15:00:53 <andythenorth> slope NE is inverted
15:01:05 <andythenorth> the SE and SE faces are transposed
15:01:15 <andythenorth> so I do need to check station orientation
15:01:43 <andythenorth> oof newgrf vars that lie
15:02:24 <_glx_> the issue is you're trying to do things against the spec ๐
15:05:22 <_glx_> you're using nearby_tile_slope(1, 0) and nearby_tile_slope(-1, 0) here ?
15:05:45 *** Wormnest_ has joined #openttd
15:06:22 <_glx_> both should return 12 for the station in the center of the screenshot
15:07:02 <andythenorth> nearby_tile_slope(0,0)
15:07:12 <andythenorth> I want the slope of current tile
15:07:44 <andythenorth> I think I just need to detect orientation and reverse the provided spriteset
15:08:11 <_glx_> I think it would be easier to use foundation CB
15:08:13 <andythenorth> I can't see the var for orientation, but mod tile_type might do it?
15:08:36 <andythenorth> it's overall easier to do this the hard way
15:08:49 <andythenorth> I have one method for spritelayouts, with a horrible detail for rail stations
15:09:05 <andythenorth> or I have 2 entirely parallel approaches to tiles for objects + roadstops, then rail stations
15:09:54 <_glx_> trying to merge stations with everything else is asking for problems
15:10:42 <andythenorth> this is how progress happens ๐
15:12:19 <andythenorth> hmm, how do I set tile_type
15:13:49 <_glx_> only 4 values are allowed
15:14:15 <andythenorth> I suspect I haven't set it
15:14:37 <andythenorth> ok that's not set
15:14:54 <_glx_> you don't need too, it will use default from openttd
15:16:26 <_glx_> 0,1 is no building, 2,3 is building, 4,5 is half roof, 6,7 is the other half
15:20:26 <_glx_> hmm but the layout is selected based on tile_type by default IIRC
15:21:17 <andythenorth> ok so SLOPE_NE will be SLOPE_NW in the transposed context
15:25:32 <andythenorth> rail stations can't build on corner slopes, so this doesn't need many cases
15:27:08 <andythenorth> _glx_: thanks, think that works
15:30:59 *** Leopold__ has joined #openttd
15:31:57 <xarick> so... pathfinding won't work as easy as I though through Valuate abuse
15:32:12 <xarick> would need to rewrite large portions and I'm not up to it
15:33:05 *** Leopold has quit IRC (Ping timeout: 480 seconds)
15:46:48 <andythenorth> ok so what next in CHIPS ๐
15:46:53 <andythenorth> I did nothing with animation
15:57:15 <xarick> should my AI stop founding new towns?
15:57:39 <xarick> with infinite money it goes crazy
15:59:42 <peter1138> Founding towns is... a weird thing to do.
16:03:08 <xarick> I wonder if there are any AIs that build new industries
16:03:25 <xarick> I know there are GS'es, but i'm not interested in them
16:03:41 <xarick> wanted to see some AI do it for their advantage
16:07:44 <andythenorth> hmm Station Variants? ๐
16:11:43 <andythenorth> can't remember why these don't use the town index to select colour like FIRS objects and industries do
16:14:17 <peter1138> Becuase stations are company-owned and are therefore company-colour.
16:15:47 <peter1138> Don't really see why you'd want to make all different colour versions of them.
16:20:01 <jfs> > I've noticed RC3 is much slower than version 13, pressing fast forward is almost the same as normal speed, because my computer isn't fast enough.
16:20:01 <jfs> are RC and Beta builds made with asserts enabled?
16:21:07 <andythenorth> peter1138: solely so you can do this
16:25:04 <_glx_> as we are looking for bug reports ๐
16:29:52 *** gelignite has joined #openttd
16:32:15 <michi_cc> Seems that NewGRFs using rail staion DB 149 really are rare...
16:33:54 <_glx_> ah you made a typo for the second PR reference ๐
16:34:31 <peter1138> jfs: Also 27ms vs 30ms per tick will affect it slightly, if it was near the limit already.
16:34:44 <peter1138> Not actually running slower, but says it is.
16:34:51 <peter1138> But mostly asserts.
16:34:55 <michi_cc> _glx_: I don't see nothing... ๐
16:38:44 <jfs> _glx_: thanks, I wrote a response about that
17:13:25 <andythenorth> ok so nfo CHIPS had some magic to randomly build tile decorations (offices, vehicles)
17:13:33 <andythenorth> nml CHIPS does not have this yet
17:13:37 <andythenorth> wonder how it worked ๐
17:33:04 <xarick> i have a feeling garbage collector is triggering all at once for every ai
17:33:34 <xarick> how often does it occur need to check
17:37:07 <xarick> (CompanyID)GB(AI::frame_counter, 8, 4) this looks wrong
17:42:53 <xarick> ah, it's fine, nevermind
17:44:38 <xarick> if it's not the garbage collector triggering the stalls, then it must be NoCAB code itself
17:59:32 <xarick> is compressed memory a possibility for openttd?
18:00:23 <xarick> I wanna run 15 NoCABs ๐
18:00:24 <_glx_> it would worse for CPU usage
18:01:32 <xarick> dump file of a running openttd with 15 nocabs, and the zip of it
18:02:04 <Rubidium> xarick: just buy a cheap non-SSD/NVME disk and use that completely for swap
18:02:32 <Rubidium> then you've got hundreds of gigabytes if not serveral terabytes of "memory" for your AIs
18:06:18 <xarick> I wonder what NoCAB is actually doing with that much data
18:08:11 <Rubidium> well, just use the garbage collector to create a graph of the data that (a single) NoCAB has and see what it's using the memory for
18:09:24 <peter1138> I guess the main issue is you always always using 4kx4k maps.
18:09:57 <andythenorth> so grf v9 for OpenTTD 15? ๐
18:10:03 <andythenorth> or is that too much like planning?
18:16:35 <andythenorth> hmm the JGRPP multi-cargo ships though ๐
18:18:15 * andythenorth want to do some mailboats
18:19:33 <alfagamma7> Thanks to JGR for Multi SHARK
18:20:34 <andythenorth> I did look if I could understand the patch
18:23:08 <talltyler> Now would be the time for planning for 15.0. Start a discussion maybe? ๐
18:24:42 <andythenorth> I'm looking for the nml patch, I tested it with Sam
18:29:02 <andythenorth> ok found some stuff and added it to the discussion
18:30:30 <andythenorth> have been avoiding adding pax + mail boats to Sam, I find it weird to have those cargos split over 2 types of vehicle
18:34:16 <talltyler> Multi-cargo ships make sense to me
18:36:01 <peter1138> Yes, so used to seeing "not an issue" issue...
18:36:24 <merni> maybe close such issues? :p
18:37:44 <andythenorth> talltyler: how much realism would I do? ๐
18:38:03 <andythenorth> bulk ships have one hold
18:38:16 <andythenorth> cargo liners have multiple
18:38:39 <peter1138> merni: As in, not reported on github, only by "everyone knows this is wrong"
18:38:56 <andythenorth> freighters are a bit of a wildcard, they have separate holds, and can have tween decks
18:38:59 *** Leopold has joined #openttd
18:39:15 <peter1138> But what about my articulated log raft?
18:39:23 <andythenorth> three kinds of tree
18:40:00 <andythenorth> if I live long enough, I always wanted to do a logging economy ๐
18:40:06 <andythenorth> so like this, but for trees
18:40:38 <peter1138> They could be called... subtypes.
18:40:41 <andythenorth> we could give them subtypes?
18:40:53 <andythenorth> the JGRPP ships aren't articulated
18:40:58 <andythenorth> they're just articulated
18:41:10 <andythenorth> so the log raft is all one
18:41:27 <peter1138> I know, that's why asking to just backport the multi-cargo feature is a ... meh.
18:41:44 <andythenorth> add it to the discussion ๐
18:41:59 <andythenorth> I'm a bit confused about how articulated ships would work
18:42:10 <andythenorth> they'd need very long vehicle length, and then they will visually break up in turns ๐
18:43:35 <peter1138> Articulated planes?
18:44:18 <andythenorth> articulated zellepins?
18:44:48 <andythenorth> well maybe I should draw a set of generic ferry things, that can do mail or pax
18:45:23 *** Leopold__ has quit IRC (Ping timeout: 480 seconds)
18:57:27 <lucafiorini> Code to json-ify an orderlist
18:57:27 <lucafiorini> std::string OrderList::ToJSONString()
18:57:27 <lucafiorini> nlohmann::json json;
18:57:29 <lucafiorini> auto& SD_data = this->GetScheduledDispatchScheduleSet();
18:57:29 <lucafiorini> auto& headJson = json["head"];
18:57:29 <Eddi|zuHause> andythenorth: you'd probably have to make a special movement that the articulated parts move in unison
18:57:31 <lucafiorini> for (unsigned int i = 0; auto &SD : SD_data) {
18:57:31 <lucafiorini> auto &SDJson = headJson["scheduled-dispatch"][i++];
18:57:33 <lucafiorini> for (unsigned int i = 0; auto &SD_slot : SD.GetScheduledDispatch()) {
18:57:33 <lucafiorini> auto& slotsJson = SDJson["slots"][i++];
18:57:35 <lucafiorini> slotsJson["offset"] = SD_slot.offset;
18:57:35 <lucafiorini> slotsJson["flags"] = SD_slot.flags;
18:57:37 <lucafiorini> SDJson["name"] = SD.ScheduleName();
18:57:39 <lucafiorini> SDJson["start-tick"] = SD.GetScheduledDispatchStartTick().value;
18:57:39 <lucafiorini> SDJson["duration"] = SD.GetScheduledDispatchDuration();
18:57:41 <lucafiorini> SDJson["max-delay"] = SD.GetScheduledDispatchDelay();
18:57:41 <lucafiorini> SDJson["flags"] = SD.GetScheduledDispatchFlags();
18:57:43 <lucafiorini> const Order* o = this->GetFirstOrder();
18:57:45 <lucafiorini> if (o != nullptr) {
18:57:47 <lucafiorini> json["orders"][i++] = nlohmann::json::parse(o->ToJSONString());
18:57:49 <lucafiorini> } while ((o = this->GetNext(o)) != this->GetFirstOrder());
18:57:51 <lucafiorini> std::cout << json.dump();
18:57:51 <lucafiorini> return json.dump();
19:21:47 <Rubidium> truebrain: the point about options 2 is to *not* use the pubkey because I want to prevent someone spamming the allow list full. Using the pubkey instead, brings you to option 1. And option 3 is just sending both, so the server can validate that it at least is a known pubkey of some client instead of just garbage
19:22:12 <truebrain> ah; sorry, it was really hard to read with all the "clients" ๐
19:22:22 <truebrain> but you could just validate if the pubkey exists server-side
19:22:26 <truebrain> still no need for client-id ๐
19:22:49 <truebrain> although spamming is always possible; even with any other command, ofc
19:22:51 <Rubidium> yeah, that's basically #3
19:23:26 <truebrain> if we are afraid of spamming, we could also just add a rate-limiter ๐
19:25:22 <truebrain> so yeah, #3 without the ClientID being sent over sounds fine to me; that does allow future work with friends-list etc (where the server can validate if the pubkey actually exists, or something like that)
19:26:40 <truebrain> if we restrict the allow-list to an arbitrary size (fixing #1), I am actually not that afraid of people constantly manipulating the list. You can do far worse things. Like sending a chat message every ms ๐
19:27:10 <truebrain> but I do agree that it needs some limit, to avoid savegames getting gigantic ๐
19:27:15 <Rubidium> yes, but a kick/ban solves that
19:28:06 <truebrain> fair; so take the example of changing my company colour every ms. Or president face.
19:28:10 <truebrain> so many spammy things you can do ๐
19:28:14 <Rubidium> spamming some hidden id is much harder to spot, and kick/ban doesn't solve the problem
19:28:44 <truebrain> how does a kick/ban not solve it?
19:28:54 <Rubidium> the only thing that could come close are signs, but you could manually remove them
19:29:26 <truebrain> if I change my president face constantly, nobody is actually going to notice, but I spam the serveer. That would be similar too, right?
19:29:41 <Rubidium> kick/ban does not remove all the keys that were added to the list
19:30:22 <truebrain> not really sure what problem you are identifying and trying to fix, I have to admit ๐
19:32:33 <Rubidium> truebrain: I'm generally fine if the neighbour toggles their light switch every second (changing face/colour/name), but I'm not fine with them dumping a bag of garbage into their yard every second (adding sign/key). If I were to call the police, they stop the light switch toggling immediately and maybe need to toggle it off once. With the garbage, the cleanup is going to be way more involved.
19:33:09 <truebrain> but if we limit the allow-list to, say, 50, that is all addressed, not?
19:33:33 <truebrain> so I am missing the point of it being garbage, more than a light switch?
19:36:43 <Rubidium> if you limit it, nobody else can be allowed to join the company anymore. Or we need to add a load of code to clean up messes that were created.
19:37:35 <truebrain> over time, we also want to add code where you can remove people from the allow-list, ofc
19:38:11 <truebrain> unrelated to this conversation btw; just something people will request ๐
19:38:31 <Rubidium> well... it will be needed to combat griefers
19:39:24 <truebrain> either way, for the first version, I would be perfectly fine if you can only add pubkeys the server sees
19:40:02 <truebrain> so the DoCommand sends over the pubkey, the server checks if any clien thas that pubkey, and processes it
19:40:25 <truebrain> later on, we might add some ACL or what-ever, and allow you to remove people from companies allow-lists, etc etc
19:41:22 <truebrain> that means you also don't need to restrict the allow-list for now, I guess, as the chance on abuse is relative low
19:42:15 <truebrain> mostly why I mentioned: if we only do pubkeys in the DoCommand, and not ClientIDs, we can extend easier with it in the future ๐ And figure out how to prevent the grieving you mention here ๐
19:46:23 <xarick> no disaster happened, but my SSD is burst peaking at 600 MB/s writes
19:50:11 <xarick> okay, I'm stopping, unsure I can trust this SSD
19:51:07 <xarick> last screenshot before I go
19:52:08 <xarick> 82,8 GB memory allocated
19:54:24 <xarick> lol it's taking an eternity to unload OpenTTD
19:58:39 <xarick> what is it doing exactly?
20:03:08 <xarick> it's slowly getting there
20:11:41 <xarick> took about 20 minutes exiting openttd
20:13:28 <andythenorth> going to rework FIRS ports
20:13:45 <andythenorth> should they stick out into the water, or just limited to the coast tiles?
20:14:04 <xarick> then disable serve_neutral_industries
20:14:19 <xarick> force everyone to use ships
20:22:44 *** brickblock19280 has joined #openttd
20:22:44 <brickblock19280> I like how they stick out like this but it can be quite hard to expand them in any meaningful way. Something which would likely be easier if they were more centred on the coast
20:24:31 <reldred> andythenorth: Sticky outy. Love me docks.
20:44:56 <reldred> xarick: Why change it?
20:45:31 <xarick> i think there's a maximum number of subsidies of 256
20:45:57 <xarick> waiting 5000 years for a subsidy to expire is... meh
20:46:00 <talltyler> Thatโs a nice maximum that fits within the variable that stores the months remaining. 0 is already used as the โsubsidies disabledโ special value.
20:46:10 <andythenorth> is there any way we can make a cleared slope tile adjacent to water into a coast tile for grf var?
20:46:12 <talltyler> But 5,000 is essentially forever
20:46:22 <lucafiorini> could someone a bit more knowledagble about the inner workings of the order system help me help with a question in Discord channel #jgr-patch-pack ?
20:46:39 <andythenorth> /me trying to make FIRS ports more buildable
20:46:45 <andythenorth> but a cleared tile isn't a coast tile
20:48:28 <talltyler> Why would it be cleared?
20:53:29 <talltyler> Are you working on a GS? Or can GRF do terraforming Iโm forgetting?
20:53:49 <andythenorth> if the player terraforms to make the correct coast shape
20:55:57 <talltyler> Coast tiles flood pretty quickly, no?
21:00:24 <andythenorth> yes, it's mostly a QoL of issue for players
21:00:34 <andythenorth> the error message is just 'cannot build here'
21:10:16 <talltyler> Maybe a custom error would solve the problem?
21:19:16 <frosch123> if you allow building on cleared land, you may cause other problems further down the road
21:19:25 <frosch123> the shore-ness is stored in the industry-tile
21:19:55 <frosch123> and affects drawing default ground, canal/river borders and stuff
21:27:58 <andythenorth> hmm I would be better off writing a port layout rotator
21:28:03 <andythenorth> instead of doing them manually
21:28:09 <andythenorth> it's just matrix transformations? ๐
21:30:25 <Eddi|zuHause> grid rotations are just coordinate shenanigans
21:35:23 <wensimehrp> Can we extend `random_bits_tile`? 4 bits is very limited
21:38:52 <truebrain> Rubidium: that also made the PR a lot smaller; did not expect that ๐ฎ
21:39:15 <truebrain> looks good; will do a proper review and testing later this week, if nobody has done it before ๐
21:41:54 <_glx_> wensimehrp: 16 different option per tile (plus the whole 65535 of the station itself) should be plenty
21:43:17 <andythenorth> pff rotating the layout will also require rotating some sprites
21:46:22 <wensimehrp> I am thinking of making a station tile that displays containers. Each container have 4 possible design and 3 different recolouring. Additionally, each station tile contains 16 or 32 containers depending on the cargo amount.
21:46:22 <wensimehrp> And that means more than 35 sprites per tile... which is pretty much
21:47:21 <wensimehrp> And I want the game to randomly choose the design and recolouring for each container, also if the container is going to be displayed or not
21:48:10 <_glx_> 4 design * 3 colours is only 12 ๐
21:48:22 <Eddi|zuHause> sounds like a case for subsprites, but i don't know if stations support those
21:50:14 <wensimehrp> _glx_: As I said, each tile contains up to 32 containers, and I want each container to have a different appearance...
21:50:33 <_glx_> and 16/32 container on a single tile while the train itself will have only 2 (maybe 4 at most) on the same tile
21:51:03 <_glx_> they will fill very small on the station compared to the train
21:51:09 <Eddi|zuHause> you can "fake" colour variations by making them highly correlated
21:51:32 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
21:52:26 <jfs> it would be cool if ottd had real containerized cargo where units of a "main cargo" are wrapped into a shipping container, and the shipping container is then what is transported, but that doesn't fit very well into the current data model at all
21:53:40 <_glx_> btw usually if you see waiting cargo it means your network doesn't work ๐
21:53:55 <jfs> in a containerized cargo model, the individual containers could then have a fixed design that remains throughout its entire lifetime, so you could see the same container being loaded between train, ship, truck, disappear into a plane, and then come back out again, but yeah, would need a big rework of a bunch of things
21:54:05 <_glx_> it's due to how station drawing works
21:54:51 <andythenorth> _glx_: this is still puzzling, I can't think of any other way to show cargo ๐
21:54:58 <andythenorth> but it's silly for eye candy
21:55:08 <jfs> wensimehrp: try putting a train with containers next to that, see if the scale matches
21:56:54 <_glx_> well station copied from the loading/loaded mechanism of vehicles
21:57:11 <_glx_> not a wise choice for me
21:57:28 <_glx_> but so is the whole station spec ๐
21:58:11 <_glx_> luckily they found a better way for industries
22:02:28 <andythenorth> junction for this would be a nightmare ๐
22:02:34 <andythenorth> (port with layout gaps)
22:11:00 <wensimehrp> wensimehrp: These sprites are not referring to standard sized containers, rather smaller containers widely used in Japan
22:16:21 <wensimehrp> `STORE_TEMP(cargo_amount_waiting(bitmask(CC_PASSENGERS)), 13), // people waiting`
22:16:21 <wensimehrp> I can actually do this??
22:16:51 <wensimehrp> something must've been wrong
22:21:02 *** nielsm has quit IRC (Ping timeout: 480 seconds)
22:32:16 <peter1138> I dunno what you're trying to do.
22:34:15 <peter1138> For stations, you never have to look at variables to do with the amount of waiting cargo, you just use the numlittlesets/numlotssets stuff.
22:37:24 *** gelignite has quit IRC (Quit: Stay safe!)
22:43:40 <peter1138> Such faff. Not much point changing npf right now
23:04:02 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
23:11:25 <xarick> i need to be less lazy
23:11:31 <xarick> gonna test river generation
23:15:15 <reldred> andythenorth: i wish I could make my docks like this, but magically flipping trains mentally pains me
23:27:08 <xarick> AutoRestoreBackUp is not needed for River generation, just tested ๐
23:28:15 <xarick> AyStar has a 10000 max search nodes when it inits by default
23:28:37 <xarick> NPF uses AyStar, so inits with 10k
23:28:47 <xarick> river uses AyStar, so inits with 10k
23:29:38 <xarick> each findpath call still sets the value of max search nodes to what the user has in the settings
23:30:38 <xarick> i guess that's safe enough for between game starts, load saves
23:30:56 <xarick> uses the saved setting
continue to next day โต
