IRC logs for #openttd on OFTC at 2024-04-19
⏴ go to previous day
00:02:19 <peter1138> Sorry if you spent hours looking at it 😒
00:38:10 <peter1138> Another old patch...
01:58:55 *** Wormnest has quit IRC (Quit: Leaving)
02:35:55 <pickpacket> There's some qol changes I'd like to make but I don't know where to start.
02:35:57 <pickpacket> 1) I'd like depots to list the number of vehicles in it like other vehicle lists do
02:36:19 <pickpacket> 2) I'd like the town listing to show the number of towns in the list
02:37:23 <pickpacket> 3) In the towns list I'd like to indicate which towns currently have a ratings modifier
02:37:57 *** gnu_jj_ has quit IRC (Ping timeout: 480 seconds)
02:39:29 <pickpacket> I also have an idea that I'm not sure about. I'm thinking that maybe a statue of company order should put a permanent positive ratings modifier on the town's opinion
02:41:26 <pickpacket> but anyway, the first three feel like no-brainers. But where do I start? I've looked around in different *_gui.cpp files but I haven't found anything that looks like what I'm trying to find
02:44:01 *** debdog has quit IRC (Ping timeout: 480 seconds)
03:20:32 <woobilicious> @pickpacket figuring out the GUI was a PITA when I wrote a simple patch to display (City) for cities:
03:45:21 <woobilicious> merni: IDK if you're a coder, it shouldn't be too hard to add trackpad for windows support see GET_X & GET_Y macros:
04:11:35 <locosage> pickpacket: it already shows a face for the rating
04:11:51 <locosage> pickpacket: done in citymania patchpack
04:20:16 <johnfranklin> Just knew Tyler has been a train dispatcher in real life
04:33:38 <pickpacket> locosage: well, that's in a patch pack :D I've never used those and would like it in the original game
04:35:00 <woobilicious> if it already exists it can be ported to vanilla.
04:40:41 <DorpsGek> - Update: Translations from eints (by translators)
05:20:22 <pickpacket> woobilicious: I'll try to have a look at that
05:20:38 <merni> woobilicious: I'll look at it
06:27:14 *** SigHunter has joined #openttd
08:19:22 *** Smedles has joined #openttd
08:42:21 <truebrain> 4% of BaNaNaS traffic was over TCP, in the last 24h .. that is a new low. So at least the HTTPS stuff had some effect 🙂
08:47:17 <truebrain> funny ... Azure: 100GB per month free, 0.08 dollar per GB. AWS: 100GB per month free, 0.08 dollar per GB. Oracle: 10000GB per month free, 0.008 dollar per GB. And no, those extra zeros are not a typo, it seems
08:53:24 <truebrain> but they sure are cheap
09:02:15 <truebrain> it might just be worth deploying those bandwidth hungry services like BaNaNaS TCP onto Oracle Cloud, tbh
09:02:28 <truebrain> the data is stored on Cloudflare anyway
09:07:13 <truebrain> I mean, AWS is overcharging on anything atm ...
09:07:37 <peter1138> I suppose there is, equally, "But, Amazon" and "But, Microsoft".
09:08:02 <truebrain> doing a rough estimation, our bill would go from 250 dollar -> 50 dollar per month
09:08:27 <peter1138> Okay, didn't know it was anywhere near that. Go for it...
09:08:48 <truebrain> AWS charges 0.09 dollar per GB
09:08:56 <truebrain> we transfer 1TB (mostly BaNaNaS traffic) out
09:09:04 <truebrain> owh, and that is before tax, ofc
09:10:41 <peter1138> Jesus wtaf Microsoft...
09:10:50 <truebrain> but, first eints to Cloudflare, as that should be like 50 dollar a month less too 🙂
09:10:59 <truebrain> peter1138: what did you do? 😛
09:13:35 <truebrain> it is funny, my migration to Nomad will help out if we actually want to migrate to Oracle .. as it is mostly just creating the infrastructure, and deploy Nomad on it. After that, it is "just working" 😛
09:13:48 <truebrain> it helps to not be stuck with a vendor-lockin 🙂
09:14:38 <peter1138> So my phone was ringing with a Teams call from my manager.
09:14:47 <peter1138> I pressed the button to answer it.
09:15:11 <peter1138> It then went to a "you need to sign in to microsoft" page, requiring a password to sign in...
09:15:50 <peter1138> truebrain: Surely we just deploy to andy's MBP.
09:16:18 <merni> I can't imagine the Microsoft higher-ups use teams or it'd have been made actually useable ages ago
09:16:31 <truebrain> peter1138: hahaha, that is priceless 😄
09:16:51 <truebrain> peter1138: ugh, that HN post ... those people really never hosted anything larger than a peanut in their life, I am afraid
09:17:14 <peter1138> Anyway, no Microsoft, I don't really have the time to start up my password manager, unlock that with biometrics, then get the password for Teams and paste it into your password box, then wait for it to sign in, probably requiring 2FA as well. WHILE A CALL IS RINGING...
09:23:12 <truebrain> `Oracle cloud hardware to reside in Azure datacenters – and Microsoft's good with that` fantastic headline, haha
09:39:26 <tateisukannanirase> Hi everyone, I've been working on the RV pathfinder lately and as part of my testing I was loading a save file, forwarding 10 years and logging the profit of each year. I'd been able to get identical figures between an older 13.4 release that I had downloaded and builds that I made myself. After 14.0 came out, I downloaded a version from openttd.org and ran my 'test' again and the profit figures
09:39:26 <tateisukannanirase> were higher (nice..).. but my local builds (even without my modifications) still have the same profit figures as my old 13.4 benchmark. Does anybody have any ideas why the version of 14.0 would give different results?
09:45:41 <locosage> tateisukannanirase: As you're basically getting different results on two builds of the same code it's either what you built isn't 14.0 or your testing methodology is bad.
09:47:09 <peter1138> Or it's being affected by the autoreplace order bug 🙂
09:47:48 *** Ox7C5 has quit IRC (Ping timeout: 480 seconds)
09:47:55 <locosage> I'll count that as testing methodology :p
09:48:02 <locosage> also, doesn't seem likely to me
09:49:40 <tateisukannanirase> locosage: I didn't build 14.0, I downloaded that from openttd.org. The local builds I have are all consistent (although admittedly I've only build against maybe one commit per week for the last month or two). Does the production/release build process usually differ significantly?
09:50:36 <tateisukannanirase> I guess there could be a bug that was introduced and fixed inbetween the last two times I built
09:50:54 <locosage> tateisukannanirase: then what are you referring to in "but my local builds (even without my modifications) still have the same profit figures as my old 13.4 benchmark"?
09:51:29 <locosage> if you didn't build 14.0 ofc 13.4 will work as 13.4 😵💫
09:54:52 <tateisukannanirase> OK, well I built with your commit 0d1fc47edb271f55dded207506961453c3ed0b5e today and it has the exact same profit figures on this save game as on 13.4 downloaded from openttd.org (in Nov2023 according to my file timestamps).. but this same saved game on 14.0 downloaded from openttd.org does not give the same figures. I hope that makes it a bit clearer!
09:56:22 <locosage> try building 14.0, comparing with master here adds a lot of variability here
10:00:50 <tateisukannanirase> OK I kicked off a build on release/14. It'll be a while on this old machine 🙂 One thing to add is that this map/saved game includes only RVs, which is a pretty stable part of the codebase from my observation.
10:02:16 <locosage> with all the C++fication going on nothing is really stable 😅
10:33:44 <merni> tateisukannanirase: could it also be related to either of these?
10:38:15 <peter1138> That could change it between 13.4 and 14.0, master should still be roughly the same as 14.0
10:40:44 <peter1138> (I'm assuming your "local builds" are based on master, and not something old...)
10:42:10 <tateisukannanirase> merni: locosage I just finished testing the local build of release/14 and it matches the downloaded version of v14. so I guess something was added, then reverted. But it doesn't matter too much, as long as I have a benchmark I can test my yapf changes against.
10:42:38 <tateisukannanirase> peter1138: yes I am building off of master including one build today!
10:42:57 <merni> what makes you say reverted?
10:44:07 <peter1138> Anyway, if you want to make sure your changes don't change your over-all profits, you should compare your version with the version you branched from.
10:44:07 <tateisukannanirase> because my 'profit benchmark' on a master build matches the results from builds before v14.
10:44:48 <peter1138> Maybe TB needs to include "profit" stats in his CPU and memory performance tests 😄
10:45:11 <tateisukannanirase> yeah true but I need to merge in the master to my branch sooner or later 😄
10:45:13 <merni> tateisukannanirase: could be something merged after 14 branched and not backported...?
10:46:44 <peter1138> Anyway, we don't really track changes to profit, there are sometimes decisions to deliberately change it, e.g. those PRs you highlighted.
10:46:53 <peter1138> Bugs can change it.
10:47:07 <peter1138> And of course random seeds can change it.
10:47:29 <tateisukannanirase> Yeah that's fine, I'm not suggesting the community should do it. I am just doing it myself.
10:48:06 <peter1138> Nobody is looking at PRs and wondering if a savegame from a year ago is still making the same profit. We might check if it still loads...
10:49:14 <tateisukannanirase> I presume that if I merged something that made the yapf worse and profit dropped (amongst other metrics) some players would be upset!
10:49:39 <peter1138> It's unlikely to affect profit though.
10:50:03 <peter1138> Players would mostly likely notice their vehicles taking different routes and complain about that.
10:51:48 <ahyangyi> Speaking of the max cargo age change, is the profit callback still limited to 256 days?
10:56:20 <peter1138> Yes, it's clamped to 255 if it's longer.
10:56:43 <peter1138> (although iirc it was never actually days)
10:57:04 <peter1138> The unit is something like 2.5 days? I can't remember off hand...
10:57:12 <ahyangyi> "some magical time units"
10:57:44 <ahyangyi> And, fractional time factor such as 1/16 is also impossible with the cb?
10:58:15 <ahyangyi> oh, the cb doesn't calculate the time factor, it calculates the profit...?
10:58:56 <ahyangyi> So I guess 255 periods thing is the only one to worry about
11:00:19 <peter1138> Well, you don't need to worry about it, it just behaves as it did before the change.
11:01:01 <ahyangyi> But that does make any claim that my grf "improves" the payoff curve more dubious 😛
11:01:05 <locosage> I can't think of any change that would cause profit difference between 14.0 and master
11:01:21 <locosage> tateisukannanirase: can you post your save and describe what are you measuring?
11:02:32 <tateisukannanirase> locosage: it's not really a big deal or anything that needs to be fixed. but if it's just your general curiosity, I can indeed share it.
11:09:00 <peter1138> ahyangyi: You can't improve TTD 😉
11:11:13 <peter1138> I don't see var 10 being used for anything in the spec for callback 39.
11:11:34 <peter1138> So we could put the 16 bit transit_periods into the first 16 bits of var10?
11:16:30 <peter1138> `telnet mapscii.me`
11:16:36 <peter1138> Then... use the mouse.
11:21:26 <peter1138> Hmm, I also wonder if savegame size might be another useful metric to watch over time.
11:22:18 <peter1138> (This is sort of a pivot from my usual "nobody cares about filesize" stance...)
11:28:07 <locosage> tateisukannanirase: Well, it may still be a sign of some big somewhere so won't hurt to check. So, yeah, post your save, I'll check when I have time
11:29:07 <locosage> Cb39 as a whole can use some rethinking as it's largely useless as it is
11:30:06 <tateisukannanirase> I DM'd it to you locosage , if that's what you meant
11:31:06 <locosage> peter1138: Well, you know my opinion on that 😅
11:33:45 <peter1138> Hmm, so we could just add a parameter that loads a savegame and then immediately saves it and quits.
11:35:18 <_glx_> Like `-vnull:ticks=1 -g ...`?
11:35:29 <locosage> Measuring save after running for a while can also be useful
11:36:07 <locosage> Some changes don't have immediate effect
11:36:19 <peter1138> _glx_: That doesn't save though, does it?
11:36:34 <locosage> Also uncompressed file size may be interesting
11:36:40 <_glx_> There's an autosave at exit setting
11:36:54 <peter1138> So we don't need to add anything \o/
11:37:13 <_glx_> But the setting removes the "are you sure"
11:38:07 <peter1138> Being able to set settings via the command line would be nice too.
11:40:20 <peter1138> ./openttd -vnull:ticks=1 -g ... -o autosave_on_exit=true
11:43:26 <peter1138> (Or rather `gui.autosave_on_exit`)
11:44:14 <ahyangyi> ./openttd -vnull:ticks=1 -g ... -o i_am_sure
11:46:31 <peter1138> Hmm, sometimes templates are quite annoying.
11:47:11 <peter1138> BaseSet is a template, so BaseSetTextfileWindow has to exist for all 3 BaseSet types...
11:48:36 <johnfranklin> efessel: We are thirsty for an open video conference software
12:04:53 <peter1138> > 0000000000897490 W BaseSetTextfileWindow::BaseSetTextfileWindow(TextfileType, std::basic_string_view<char, std::char_traits<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, unsigned int)
12:04:59 <peter1138> > 0000000000897520 W BaseSetTextfileWindow::~BaseSetTextfileWindow()
12:05:15 <peter1138> Interesting, I changed the type of one parameter, and the compile elided them all...
12:05:51 <peter1138> That constructor no longer exists, when I changed the string_view to a const std::string &.
12:06:01 <peter1138> (It's already a std::string, not breaking my new rule ;))
12:47:48 *** nielsm has quit IRC (Ping timeout: 480 seconds)
13:48:17 <xarick> CalcClosestStationTile
13:49:40 <xarick> there's a bit of YAPF trains that use it, but they don't use docks
13:52:17 <xarick> I was a bit lazy and also told not to focus much on NPF
13:52:35 <xarick> so I just made the changes to it very minimal
13:57:16 <xarick> NPFCalcStationOrTileHeuristic
13:57:35 <peter1138> If that function is only used for NPF, then don't add it 🙂
13:57:48 <xarick> yapf uses it, but only for trains
13:58:02 <peter1138> Or at least, keep NPF and YAPF changes to separate commits.
14:19:05 *** matusguy has joined #openttd
14:23:11 *** dinomartino has joined #openttd
14:23:13 *** dinomartino has quit IRC ()
14:58:56 *** Wormnest has joined #openttd
15:41:46 *** gelignite has joined #openttd
16:28:10 *** HerzogDeXtEr has joined #openttd
16:43:44 <peter1138> andythenorth: var61?
16:46:23 <peter1138> Iron Horse uses it a lot in the purchase list, where it does not exist.
16:48:35 <peter1138> Hmm, oh, maybe not in the purchase list. Hmm.
16:49:47 <peter1138> Ah it's used when it's not safe to. Fun.
17:00:43 <peter1138> vehicle ID 18000, nice.
17:01:18 <peter1138> Hmm, loads of them...
17:04:14 <peter1138> 6606 engines defined with Iron Horse. Definitely reasonable.
17:14:02 <andythenorth> did we do a wrong thing?
17:19:57 <peter1138> Well, using var 61 when it's not available might cause things?
17:20:26 <peter1138> The "avoid circular dependencies" limit.
17:21:45 <Eddi|zuHause> it might beat CETS
17:22:06 <Eddi|zuHause> i think that had just under 4000
17:23:01 <Eddi|zuHause> remember when i uncovered a bug in grfcodec when the sprite number rolled over 64k?
17:23:22 <Eddi|zuHause> or possibly it was nforenum
17:31:12 <andythenorth> peter1138: am looking...
17:31:35 <peter1138> It's all over the place, sorry.
17:32:34 <peter1138> Savegame bumps okay in master?
17:32:47 <andythenorth> peter1138: it will be some templating artefact
17:32:53 <peter1138> Or wait a bit for 14.x fall-out...
17:32:53 <andythenorth> just got to figure out which
17:33:13 <andythenorth> 9 templates with var 0x61
17:39:36 <andythenorth> peter1138: can you see any of the other var 61 params?
17:40:07 <andythenorth> 2068 uses in compiled Horse, figuring out which one is in the purchase context is ....pff
17:42:54 <peter1138> No, about that, I was mistaken, it wasn't in the purchase list.
17:43:20 <peter1138> I forgot I had 14 AIs set up, which were building vehicles.
17:47:30 <peter1138> Wonder if this used to be old-style PBS...
17:47:53 <peter1138> > Conversion from OTTD savegame without gamelog: version 16, 0
17:57:43 <_glx_> peter1138: entry everywhere? probably
18:00:12 <_glx_> but yeah looks like an old PBS conversion to previous signal in the enum
18:06:26 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
18:12:20 <peter1138> ^ If the chunks happened to be a different order somehow, that would've failed badly...
18:24:19 <Eddi|zuHause> did the old PBS get "downgraded" to combo signals, instead of plain signals?
18:24:55 <peter1138> Did old PBS allow mixing with pre-signals too? Can't remember...
18:26:08 <Eddi|zuHause> i don't know if that was the old PBS, or TTDPatch.
18:26:27 <Eddi|zuHause> but there's a reason why we have 2 unused signal types in the graphics
18:26:56 <Eddi|zuHause> because the 4 PBS types mirrored the 4 plain types
18:51:03 <peter1138> Rail type label conversion.
18:51:35 <peter1138> Should we 1) remove it, or 2) implement the missing functionality for road types?
18:52:45 <peter1138> Or 3) implement something more dynamic that would please the not-enough-roadtypes crowd...
19:05:56 <andythenorth> 2 roadtypes were enough 😛
19:09:57 *** felix has quit IRC (Ping timeout: 480 seconds)
19:21:43 <Eddi|zuHause> they were. but then they gave out a hand, and they wanted the whole arm.
19:21:57 <Eddi|zuHause> and then complained that the arm isn't enough
19:22:14 <Eddi|zuHause> (because it's never enough)
19:26:04 <andythenorth> ships in knots eh
19:41:01 <Eddi|zuHause> are there metric and imperial knots?
19:45:36 <peter1138> Just like metric and imperial metre.
19:51:35 <Eddi|zuHause> that would be "funny", an "imperial metre" that is close-but-not-quite the "metric metre", but convertible to other imperial units
19:52:03 <_glx_> "metric system is too hard"
19:59:55 <peter1138> Hmm, I don't understand opengl.cpp:1259
20:00:45 <peter1138> Specially the calculation at the end, which is then treated as a pointer.
20:01:33 <peter1138> Which is normally a pointer to image data.
20:07:32 *** Artea has quit IRC (Read error: Connection reset by peer)
20:07:42 *** Rubidium has quit IRC (Remote host closed the connection)
20:07:44 *** Rubidium has joined #openttd
20:07:44 *** ChanServ sets mode: +o Rubidium
20:17:01 *** frosch123 has joined #openttd
20:17:01 <frosch123> isn't "yard" the "imperial meter"?
20:20:47 <frosch123> peter1138: "If a non-zero named buffer object is bound to the GL_PIXEL_UNPACK_BUFFER target (see glBindBuffer) while a texture image is specified, pixels is treated as a byte offset into the buffer object's data store. "
20:20:54 <frosch123> blame C programmers
20:21:11 <frosch123> they don't know the difference between pointers and integers
20:39:17 <Eddi|zuHause> frosch123: not close enough
21:15:22 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
21:20:53 <peter1138> > ps auxH | grep chromium | wc -l
21:21:00 <peter1138> Guess how many tabs I have open in Chrome?
21:52:56 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:53:46 *** Wormnest has joined #openttd
21:57:39 <Eddi|zuHause> who would assume one tab equals one process?
21:58:09 <peter1138> Clearly we need more processes in OpenTTD.
21:58:56 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:07:16 <Eddi|zuHause> good idea, and each process handles its own pathfinder
22:07:33 <Eddi|zuHause> surely solves all the problems
22:39:21 *** matusguy has quit IRC (Remote host closed the connection)
22:43:42 *** felix has quit IRC (Ping timeout: 480 seconds)
22:45:57 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
23:09:30 *** Wormnest has joined #openttd
23:42:17 *** ChanServ sets mode: +v tokai
23:48:57 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
continue to next day ⏵
