IRC logs for #openttd on OFTC at 2025-12-24
⏴ go to previous day
01:01:47 *** ChanServ sets mode: +v tokai
01:08:44 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
01:24:18 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
01:24:25 *** WormnestAndroid has joined #openttd
02:44:44 *** Wormnest has joined #openttd
03:47:47 *** rads has quit IRC (Quit: Page closed)
04:44:32 <DorpsGek> - Update: Translations from eints (by translators)
05:14:00 *** WormnestAndroid has quit IRC (Remote host closed the connection)
05:14:02 *** WormnestAndroid has joined #openttd
08:42:58 <locosage> so mp is still half-dead huh :/
08:43:14 <locosage> kinda awkward that openttd can't even connect to direct ip without a command
08:43:20 <locosage> when it could a few versions ago
08:59:50 <peter1138> If by "command" you mean "Click on Add server and enter the IP in the pop-up window"
09:01:05 <locosage> either way it requires entering address manually
09:01:28 <locosage> when it can be obtained from coordinator
09:02:13 <peter1138> And if you could enter the IP, it still wouldn't connect.
09:03:16 <locosage> wdym? it does connect by direct ip
09:03:42 <tranri> im using rdaminvpn to join my friends server
09:11:00 <reldred> CGNAT is now starting to be the norm rather than the exception.
09:11:09 <reldred> Makes things hard for people
09:48:57 <jfkuayue> pickpacket: i am drinking tea in sweden
09:57:52 <peter1138> Well, it's working now.
09:57:54 <ahyangyi> reldred: I wonder when IPv6 will be the norm rather than the exception
10:12:32 <peter1138> The statement "openttd can't even connect to direct ip without a command" is, of course, completely untrue.
10:14:25 <peter1138> Ppenttd won't connect directly by IP to a server in the server list, and doesn't receive the IP to be able to do so.
10:14:59 <peter1138> That appears to be the problem, but not what was said.
10:18:13 <reldred> ahyangyi: IPv6 has little to do with it, carriers straight up don’t want you routing unsolicited inbound traffic down retail domestic internet services.
10:18:24 <LordAro> ah, the age old describing the wrong problem
10:18:44 <reldred> I keep a VPS running with wireguard to push inbound traffic to my own servers these days
10:18:44 <LordAro> some people should know better though ;)
10:19:08 <ahyangyi> reldred: That's true as well
10:19:31 <reldred> Gonna slowly scale down that VPS til it’s just an inbound proxy once I can get my hands on enough ram and SSD’s for my little cluster
10:19:48 <ahyangyi> Though I just feel it's part of ISPs' general tendency of charging more money and providing less service
10:19:50 <reldred> I love my little 1L Lenovo tiny PC cluster
10:20:05 <reldred> It’s turned out so nicely
10:24:25 <peter1138> I should've done that while they were cheap...
10:24:54 <reldred> Yeah… I want a trio of M90Q’s but they’re so bloody dear now
10:25:26 <reldred> I’ve got M920Q’s but they’re i5 8500t’s and only have a single nvme slot.
10:26:39 <reldred> But I also still have sata so Proxmox is booting off a shucked 2.5” sata ssd wrapped in kapton tape and nestled alongside the 2xSFP+ mellanox nics I crammed in there
10:27:19 <reldred> And Proxmox hyperconverged has been a nice system to work with
10:27:56 <LordAro> i'm looking at using incus HA stuff for next work infra refresh
10:28:29 <LordAro> did initially look at proxmox, but it's "too different" to our current somewhat ad-hoc lxd stuff
10:29:10 <reldred> Yeah that’s fair if you’re already coming from lxd then incus looks like it’s similar enough in its approach
10:29:38 <reldred> But for me I grew up on ESXi/VSphere based solutions so Proxmox is right up my alley.
10:47:07 <LordAro> mm, it was a container/vm vs docker support thing for us - most of the things we're using don't really exist as ephemeral images
10:58:29 <pickpacket> jfkuayue: me too! 😄 Where in Sweden are you, if I may ask?
10:59:25 <peter1138> I used to use libvirtd back in the day, it's very manual though.
11:27:24 <locosage> what's untrue? how does one connect to a server without entering the ip:port?
11:27:32 <locosage> I mean when stun is down ofc
11:27:46 <locosage> haven't checked but I bet 1.9 could still connect
11:36:20 <jfkuayue> pickpacket: stockholm, but i was on the X2 back then
11:39:41 <peter1138> Entering the ip:port into the "Add server" pop-up is very clearly "without a command"
11:40:09 <peter1138> andythenorth, lunch?
11:43:46 <andythenorth> peter1138: yup, time
11:43:47 *** gelignite has joined #openttd
11:48:59 <Rubidium> locosage: I've done a simple test... I added Debug statements to the Receive_GC_*_CONNECT functions in network_coordinator.cpp (using 20251223-master-m98b0197d53). The result: `[2025-12-24 12:46:34] dbg: [net:0] GC_DIRECT_CONNECT`
11:52:19 <locosage> I don't understand what you mean with that
11:52:35 <locosage> It does dirrect connection afaict but still needs stun server to get ip or smth
11:57:40 <Rubidium> for direct connection without entering an IP you do need a working game coordinator (that what used to be the master server), as that's what's giving you the server listing. If that is not working, then you do need to enter an IP, but if the master server wasn't working you also needed to enter an IP. That has nothing to do with STUN.
11:57:57 <peter1138> It is "direct" connect via invite code, rather than via ip:port. So it still has to go via the coordinator.
11:58:16 <peter1138> The coordinator was not responding properly. It was not just STUN.
11:58:55 <Rubidium> the only 'problem' STUN has introduced is that not properly configuring your firewalls/portforwards etc will not be really visible anymore, as usually STUN/TURN will happily connect clients to you
11:59:19 <locosage> afaict both are true, stun server couldn't connect to server directly but client could also not resolve invite code
12:00:35 *** toktik is now known as Guest34334
12:02:21 <locosage> *stun server->gc I guess
12:05:14 <locosage> btw, I've also seen that rarely openttd suggests to use turn for server that available directly
12:06:20 *** Guest34334 has quit IRC (Ping timeout: 480 seconds)
12:11:53 <Rubidium> locosage: that log is from after the potential direct connect attempt. It's just that seems that it's very bad at logging that it's actually attempting that and failing. So I guess it would be useful to add more logging, especially about the GC communication, as that might give more insight.
12:13:10 <Rubidium> and that turn is sporadically used... might be a short packet drop at an unwanted moment. Might also be a more serious issue, but I have no way of looking into that
12:26:17 <talltyler> I assume no 15.0 release today, when do we want to do RC4?
12:56:43 *** blinkyistaken has joined #openttd
12:56:43 <blinkyistaken> Hi. I'm looking through the localization files in my language (nb_NO) and I'm seeing some errors. What's the best way to flag these to get them corrected?
13:02:37 <LordAro> talltyler: rc4 today/tomorrow, release next week?
13:32:08 *** Flygon has quit IRC (Remote host closed the connection)
14:07:29 <locosage> Rubidium, that is weird because I don't see why it would fail direct conection attempt via server list only to succeed a few seconds later via command line
14:07:36 <locosage> and that was repeatable, not a fluke
14:08:21 <talltyler> I should update my cacti PR then
14:08:54 <locosage> yeah, and gc couldn't reach the server either, that I know from server log
14:10:06 <talltyler> I've been toying with desert/rainforest generation that looks less like bathtub rings, but my current attempt at evaporating water from water tiles and blowing it inland, with mountains blocking the moisture to cause rain shadows on the other end, is not quite there. 🙂
14:10:47 <locosage> basically, as I understand it the problem is that if there is a connectivity issue on GC side clients will never even attempt a direct connection
14:11:21 <talltyler> Once again I've fallen down the rabbithole of "I should play a subtropic map" -> "I should make a subtropic industry set" -> "I should improve desert/rainforest generation" 😛
14:14:57 <Borg> talltyler: hehe yeah.. :)
14:15:33 <Borg> talltyler: I improved Tropic generation a bit in my fork... basically I just took Arctic params.. and it looks pretty well
14:15:43 <Borg> low mountains generate nice tropic areas
14:15:53 <Borg> there are plenty of deserts too.. so aint that bad
14:16:10 <Borg> but.. recently.. I switched back to Arctic.. and its now my favorite climate
14:16:59 <locosage> Imo, for servers that are supposed to have direct ip it should be attempted first regardless of what gc thinks
14:17:15 <locosage> how to know what server is supposed to have is a bit tricky but I guess simplest way is to add a setting
14:17:45 *** firepup650 has joined #openttd
14:23:02 <talltyler> Borg: I've also toyed with reversing the elevation of desert/rainforest, and like that better, especially with the arctic land generator. High desert plateaus towering over low rainforests look so much better. 🙂
14:24:17 <talltyler> Just need to figure out how to get this nicely into vanilla OpenTTD, the world generation GUI is already pretty full 🙂
14:26:15 <peter1138> Make it random so that different areas of the map are current way and other areas are new way.
14:26:22 <peter1138> And then no extra option.
14:27:22 <Borg> talltyler: just add extra entry in Land Generator [;
14:27:47 <talltyler> Land Generator isn't in the GUI anymore
14:32:53 <talltyler> There was some discussion previously about setting the normal/tropic/arctic elevation adjustments manually via a setting. I'll probably do this, but as always, the hardest problem is naming things.
14:33:25 <talltyler> frosch suggested "gamma" as the equivalent in an image editor, but not as the player-facing name, which I agree with
14:34:19 <talltyler> Maybe "average height" and rename the existing flat/mountainous/alpinist/etc. dropdown to "max height"?
14:38:35 <peter1138> We should really use multiple noise maps.
14:39:00 <peter1138> Using only height is a recipe for bland.
14:42:27 <talltyler> No disagreement, but that's a much bigger project
14:54:29 <ahyangyi> And also changes how heightmaps work (for the better)
15:31:41 <_zephyris> talltyler: I'll tell you about my tectonic experiments at some point...
15:32:07 <peter1138> My scriptable stuff was amusing, but probably not viable.
15:33:10 <peter1138> Probably should be more of a pipeline process than just rigging various heightmaps together.
15:45:46 <ahyangyi> _zephyris: I remember you experimented with map gen a few months ago. Was it the tectonic experiments?
16:00:57 <peter1138> IIRC it was numpy stuff.
16:02:42 <andythenorth> What was the themed irregular-shapes tiling we looked at?
16:05:22 <talltyler> Deserts above tropics, with arctic plateaus
16:15:22 <Borg> If I ever come back to play on Tropic.. Ill try the inverse mode too
16:19:24 <_zephyris> andythenorth: Voronoi?
16:19:39 <_zephyris> I was doing numpy algorithm experiments
16:34:43 <peter1138> Using perlin noise to place desert.
16:35:05 <peter1138> Dunno if it's better, but it is different.
16:38:03 <Borg> I think it would look better if that 5 tiles away from water rule would be off
16:38:15 <Borg> at least.. might be worth trying..
16:58:14 <peter1138> Desert that ignores distance from water is kinda weird.
16:58:33 <peter1138> Also I separated rainforest, so it's not simply rainforest = !desert now.
17:06:17 *** Wormnest has joined #openttd
17:06:35 <talltyler> There are plenty of places where arid landscapes meet the sea, with little to no greenery on the coast. I think desert coast tiles would improve the look of this, over one tile of rainforest right on the coast.
17:07:26 <Borg> but even w/o desert coasts it looks ok :)
17:13:35 <peter1138> Although comments wouldn't hurt.
17:15:23 <peter1138> And it is a bit slower, as it's doing slow perlin for each elligible tile. But it's not storing the noisemap, so doesn't need MiBs of memory holding data for each tile.
17:15:32 <Borg> actually.. you dont even need desert coasts ;) make it beach [;
17:15:47 <Borg> but it will contrast too much w/ green I guess...
17:15:54 <peter1138> The game does not have beaches.
17:16:22 <Borg> sprite recolor to the rescue? ;)
17:16:31 <peter1138> No, it would need sprites to be drawn.
17:16:42 <peter1138> So basically, desert coasts.
17:16:51 <peter1138> aka beach. Same difference really.
17:17:14 <Borg> okey so tiles cannot be recolored like sprites?
17:17:32 <talltyler> I’d love to see rocky coasts someday
17:18:01 <peter1138> talltyler, anyway, that's kinda what I mean by using noise. Turns out it isn't really a big project :)
17:18:56 <talltyler> Yeah, and I imagine it scales to the player’s chosen desert percentage much easier than my approach
17:19:28 <talltyler> (That feature is broken in my current attempt, anyway)
17:19:35 <peter1138> Sort of. I ignore the amount of water at the moment.
17:20:02 <peter1138> The noise generates a value somehwere between -1.25 and 1.25 (not sure why, normally you aim for -1.0 to 1.0)
17:20:10 <Borg> can I download it as .diff or .patch file?
17:20:18 <Borg> somehow I dont see any button at github to do so?
17:21:36 <peter1138> I'll let you figure out how to use git :-)
17:22:25 <Borg> I can fetch it to my repo... will be easier
17:22:49 <peter1138> talltyler, so I scale the value up to -100 to 100, then compare that against the provided desert percentage, which is why that's `* 2 - 100`;
17:24:20 <peter1138> talltyler, the other magic numbers for the PerlinNoise2D function are scale, and a prime (which is effectively a seed, which has the generation seed added to it internally)
17:25:14 <peter1138> So 0.35 biases it to large deserts, and 0.5 for rainforests mean they have smaller clumps... but still pretty large.
17:25:52 <peter1138> And using a different prime/seed means they don't line up, so there are now areas that are neither desert nor rainforest.
17:32:45 <talltyler> How does it handle 0% desert maps? Are some areas non-rainforest?
17:35:44 <talltyler> I will work on changelog and website PRs for RC4 in a bit, if someone can do the backport PR 🙂
17:39:04 <Borg> I have different results doing git diff master PeterN/noise-desert vs PeterN/noise-desert^^ PeterN/noise-desert
17:39:19 <Borg> the second one is correct (same) w/ github link you provided
17:39:32 <Borg> but master PeterN/noise-desert contain extra changes
17:39:55 <Borg> src/network/network_coordinator.cpp | 16 ----------------
17:39:55 <Borg> src/network/network_stun.cpp | 1 -
17:39:55 <Borg> src/network/network_turn.cpp | 4 ++--
17:40:00 <Borg> I wonder where this one comes from..
17:40:51 <Borg> $ git show --stat 900192e2cd848259d90
17:40:52 <Borg> commit 900192e2cd848259d90c56bff2d0155547b60215 (HEAD -> master)
17:40:52 <Borg> Author: Rubidium <rubidium@openttd.org>
17:40:52 <Borg> Date: Wed, 24 Dec 2025 14:48:01 +0100
17:41:23 <Borg> Rubidium commit slipped in [;
17:42:37 <peter1138> "I'll let you figure out how to use git :-)"
17:44:53 <Borg> anyway! I mirrored your stuff ;) might try it one day ;D
17:45:03 <Borg> when I figure out first.. how to improve my BSPI for Tropic
17:47:05 <peter1138> talltyler, "sort of". as the amount of rainforest is still proportional to the about of desert, it is mostly rainforest.
17:47:28 <peter1138> That's simple to change, but not simple to know what to change it to.
17:48:02 <talltyler> Pick an arbitrary number territory?
17:51:50 <peter1138> Right, shall I do the backports?
17:51:58 <talltyler> I would appreciate that 🙂
17:54:40 <rito12_51026> Do we gonna have RC5?
18:00:21 <andythenorth> Is it Christmas Day release? 👀
18:00:33 <andythenorth> Horse 4 isn’t ready 😛
18:00:50 <talltyler> Release the day after RC4 seems a bit premature 😄
18:01:07 <talltyler> New Year's Eve would be nice though, keep it within 2025
18:04:41 <soylent_cow[m]> a couple of days ago this started: i can't connect to a bunch of public servers from my home. it sits there for a short while at the beginning of progress bar, then says server is not responding
18:05:18 <soylent_cow[m]> i certainly changed nothing about my home network setup
18:06:05 <peter1138> soylent_cow[m], that was fixed about 8 hours ago.
18:06:33 <soylent_cow[m]> ah thank you! great to hear
18:06:46 <soylent_cow[m]> i am in the mountains rn, which is how i know it works from elsewhere
18:07:07 <soylent_cow[m]> can't test it from home for the next few days 🤣
18:07:15 <soylent_cow[m]> on the acct of xmas
18:07:26 * soylent_cow[m] sends snowfall ❄
18:09:37 <peter1138> talltyler, we could risk it and make RC4 the release? :p
18:09:54 <peter1138> No harm in a 15.1 ;-)
18:22:50 <_glx_> at least with RCs we have more testers than beta or nightlies
18:24:01 <_glx_> but it's annoying to receive crash reports for stuff added months ago only right after the RC release
18:29:39 <peter1138> Who knows, it's not merged yet.
18:58:20 <locosage> it's a feature though
19:03:42 <peter1138> we van merge features into master.
19:04:02 <peter1138> bloody phone keyboard
19:04:18 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
19:14:15 <rito12_51026> How can I download RC2 and RC1?
19:15:18 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
19:15:20 *** WormnestAndroid has joined #openttd
19:23:57 <rito12_51026> Badge filters seem to not work in RC3 and RC4, they worked in RC1
19:30:23 <peter1138> Yup, due to the change to 32bit values. Sigh.
19:30:25 <rito12_51026> And they work in RC2, so RC3 broke sth
19:31:11 <rito12_51026> Should I create an issue?
19:31:53 <peter1138> talltyler, hold off on RC4 just yet :)
19:44:27 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
19:44:27 *** WormnestAndroid has joined #openttd
19:45:38 <peter1138> Okay, not a complete fix :(
19:46:06 <talltyler> We did it, we avoided RC5 😛
19:46:17 <talltyler> At least for today 😄
19:47:43 <peter1138> Well. I need to head out. BBL.
19:48:18 <talltyler> Have fun at the pub 😛
19:51:02 <jfkuayue> It got me curious how UK is off 25-26, while EU is off 24 night, but only "partly off" 25-26
19:51:11 <jfkuayue> Protestant vs Catholic?
19:53:57 *** WormnestAndroid has quit IRC (Read error: No route to host)
19:54:19 *** WormnestAndroid has joined #openttd
20:00:05 *** WormnestAndroid has quit IRC (Read error: No route to host)
20:09:51 <xarick> the allocator is making me restructure the entire BPlusTree :|
20:12:54 *** WormnestAndroid has joined #openttd
20:25:52 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
20:28:19 *** WormnestAndroid has joined #openttd
20:45:33 <talltyler> Hmm, why is this text the wrong blue? Affected by text colour push/pop, maybe?
20:46:27 <rito12_51026> I have a fix for a bug with aqueducts
20:48:07 <peter1138> Look, "I have a patch for that" is my line...
20:52:19 <talltyler> Yeah, but do you have a patch for aqueducts? 😛
21:00:27 *** Borg has quit IRC (Quit: leaving)
21:22:01 <andythenorth> if i nerf off unfinished sprites, I can ship Horse 4 soon 😛
21:32:04 <talltyler> I just started a new Horse 3 game 🥲
21:46:04 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
21:46:34 *** WormnestAndroid has joined #openttd
21:50:10 *** toktik has quit IRC (Remote host closed the connection)
21:57:26 *** toktik has quit IRC (Remote host closed the connection)
22:02:47 *** toktik has quit IRC (Remote host closed the connection)
22:05:32 *** Wormnest has joined #openttd
22:05:34 <peter1138> No but I have another pint.
22:06:05 <peter1138> "I have a pint for that"
22:07:29 *** toktik has quit IRC (Remote host closed the connection)
22:10:28 *** toktik has quit IRC (Remote host closed the connection)
22:14:36 *** herms2 has quit IRC (Remote host closed the connection)
22:14:57 *** toktik has quit IRC (Remote host closed the connection)
22:20:56 *** toktik has quit IRC (Remote host closed the connection)
22:40:51 <kaji_kaede> talltyler: It seems sensible for this to be coloured that way deliberately. Cost to clear has a different colour, and I'd say both values are the only relatively significant ones to gameplay in this window.
22:41:55 <kaji_kaede> Maybe a different shade of blue isn't the best colour, though.
22:49:20 <_zephyris> (Almost) Happy Christmas y'all!
22:53:50 <talltyler> Nah, it’s accidental
22:56:46 <peter1138> Should I: 1) look at #14972 2) go to bed.
23:16:03 <Rubidium> locosage: if the problem is that the GC thinks the server is unreachable via direct connect, then it won't offer direct connect to the clients trying to connect
23:29:28 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:33:27 <rito12_51026> peter1138: I wouldn't call it a patch, it can cause game to crash, it breaks more than it fixes
23:43:39 <talltyler> More blue text. Seems to only be broken for articulated road vehicles.
23:48:41 <talltyler> Yeah, seems to be when strings are composed of multiple strings, but the colour code is on the first string. It no longer applies to the string appended to the end. For example, this one is `STR_VEHICLE_DETAILS_TRAIN_ARTICULATED_RV_CAPACITY`.
23:49:38 <talltyler> Or `DrawString(r.left, r.right, y, capacity, TC_BLUE);` might be forcing it to go blue
continue to next day ⏵
