IRC logs for #openttd on OFTC at 2020-05-03
⏴ go to previous day
00:14:20 *** Wormnest has joined #openttd
04:34:30 <DorpsGek_III> [OpenTTD/nml] MinchinWeb commented on issue #44: Move nml docs to be in repo, build with jekyll/sphinx/readthedocs etc https://git.io/Je8vh
05:27:40 <DorpsGek_III> [OpenTTD/OpenTTD] MinchinWeb commented on pull request #8091: Add: [Script] Native priority queue; useful e.g. for pathfinders. https://git.io/JfsJ3
06:52:08 *** Gustavo6046 is now known as Guest23961
06:52:12 *** Gustavo6046 has joined #openttd
08:36:19 *** Progman has joined #openttd
08:52:29 *** WormnestAndroid has quit IRC
08:52:42 *** WormnestAndroid has joined #openttd
09:03:46 *** andythenorth has joined #openttd
09:21:01 <andythenorth> _dp_ 'random' strikes again
09:25:59 <_dp_> at least this random can be disabled
09:27:00 <andythenorth> I *swear* that sometimes World of Tanks Blitz gives me a bad RNG seed for days.
09:27:18 <andythenorth> but somehow I doubt it's true
09:27:54 <andythenorth> back when I made games for my job, I stopped using true random, and used various kinds of weighted randoms
09:28:19 <andythenorth> people expect random to cycle through values
09:31:37 <_dp_> but then you get ppl exploiting it
09:32:15 <andythenorth> "I only play on Wednesdays, due to the RNG seed cycle"
09:34:30 <_dp_> I've seen so much of this nonsense that my brain just filters it out as spam automatically xD
09:36:02 <_dp_> though funnily sometimes it works coz, well, pseudo randoms are pseudo :p
09:38:50 <_dp_> btw, anyone knows what rocket crashed due to bad random? heard the story somewhere but can't find it anymore
09:39:37 <_dp_> bad random generator I mean
09:45:32 *** supermop_Home_ has quit IRC
11:41:58 <TrueBrain> hmmm .. people are renaming their package to the same name .. there is nothing in place to prevent that .. but when downloading, it will overwrite these files, making it impossible to get in a good-state :D
11:42:06 <TrueBrain> guess we need to prefix the names with the package-unique-id or something
11:44:25 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
11:44:40 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain commented on issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
11:51:50 *** sla_ro|master has joined #openttd
13:13:37 *** gelignite has joined #openttd
13:28:34 *** andythenorth has joined #openttd
13:58:37 *** frosch123 has joined #openttd
14:02:04 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 commented on issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
14:28:33 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain commented on issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
14:35:58 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain opened pull request #19: Fix #56: make sure filenames send to the client are unique https://git.io/JfsO6
15:14:18 <frosch123> TrueBrain: i could approve it, but black is still unhappy :)
15:15:38 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain updated pull request #19: Fix #56: make sure filenames send to the client are unique https://git.io/JfsO6
15:16:10 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 approved pull request #19: Fix #56: make sure filenames send to the client are unique https://git.io/Jfs3M
15:17:18 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain merged pull request #19: Fix #56: make sure filenames send to the client are unique https://git.io/JfsO6
15:19:08 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 commented on issue #18: Package with "name" and "version" of non-ASCII characters will have a filename "-.tar.gz" https://git.io/JfTEL
15:24:01 <TrueBrain> I even tested it! :P
15:39:54 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 commented on issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
15:39:54 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 closed issue #56: Different packages with same name causes OpenTTD client to mess up https://git.io/Jfsth
15:40:58 <TrueBrain> owh, I opened the ticket in the wrong repo .. nice :D
15:44:00 <frosch123> that was one of the essential parts of the news post :p
15:44:24 <frosch123> you are lucky the issue was addressed, and not just closed
16:25:39 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick commented on pull request #8091: Add: [Script] Native priority queue; useful e.g. for pathfinders. https://git.io/JfsZT
16:28:57 <Samu> is yexo around, or zuu, I really wanted to talk about something
16:45:38 <Samu> ah, planetmaker, can you make priority queue available on github, I would like to create a PR on it
16:57:12 <DorpsGek_III> [OpenTTD/OpenTTD] TrueBrain commented on pull request #8106: Change norwegian_bokmal.txt plural fix https://git.io/JfsnZ
16:57:21 <TrueBrain> can't believe nobody else closed that PR yet
16:57:28 <TrueBrain> it does remind me, we need to finish eints -> GitHub :)
16:57:41 <TrueBrain> and possibly run it on AWS .. hmm .. what a world that would be
17:06:13 *** Wormnest has joined #openttd
17:36:34 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh opened pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfscP
17:52:39 <DorpsGek_III> [OpenTTD/nml] nielsmh commented on issue #44: Move nml docs to be in repo, build with jekyll/sphinx/readthedocs etc https://git.io/Je8vh
17:52:55 <nielsm> (I might just have signed up to do that work myself.)
17:59:35 *** ChanServ sets mode: +v tokai
18:05:48 <frosch123> i still consider it the wrong thing to do. plain html doesn't even have a search function
18:06:01 <frosch123> at least doxygen would generate some js search thingie
18:06:21 <frosch123> but then, i don't get what's wrong with the wiki
18:06:36 <frosch123> we do not even have to maintain it :p (*rudge does)
18:07:49 <nielsm> the wiki is not versioned along with NML-the-program itself
18:07:56 <andythenorth> I am +/-0 to the change
18:08:02 <andythenorth> but the PRs come in
18:08:08 <andythenorth> and then I and other people have to write the docs
18:08:20 <andythenorth> separately, working backwards
18:08:36 <andythenorth> but the wiki is quite convenient otherwise
18:08:57 <andythenorth> I prefer it when docs + code are in same PR but eh
18:09:41 <frosch123> not every coder is good a writing docs
18:09:54 <frosch123> you may detail the implementation into the website pr mess :)
18:10:29 <andythenorth> the reason I am +/-0 is that the nfo docs have to stay in the wiki
18:10:37 <andythenorth> so we'd end up split brain, which is weird
18:11:06 <nielsm> I'd argue the NFO docs should be put into the ottd repos so they can be versioned along with ottd too :)
18:11:23 <nielsm> I know there's ttdpatch, it's a static target and the docs for that can remain
18:11:29 <andythenorth> maybe it would be enough to have a convention for 'draft' in the wiki?
18:11:42 <andythenorth> and then hold PRs until there's a draft docs page?
18:14:03 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfsWk
18:17:33 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh updated pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfscP
18:19:28 <andythenorth> nielsm I didn't link to web translator much in docs, I was confused about whether new translators can be registered on it
18:19:48 <andythenorth> probably me being stupid, but I couldn't find an answer
18:20:49 <andythenorth> AIUI it's done by email, and nobody knows who's responsible for checking that, but I might have misunderstood
18:24:10 <frosch123> andythenorth: it's about to change to gh auth
18:24:35 <frosch123> just hold your horses, or make them iron :)
18:26:28 <andythenorth> that was the other reason I didn't change it
18:26:50 <Wolf01> Hmmm, I have too many lego, I can't build what I want
18:27:10 <andythenorth> I'd approve that PR but checks failed :|
18:27:51 <nielsm> oh the opengfx dep for test
18:30:34 <frosch123> or rather, they all succeeded, but gh got confused or something
18:31:42 <DorpsGek_III> [OpenTTD/nml] glx22 dismissed a review for pull request #115: Fix #112: prevent __version__.py regeneration for source package https://git.io/Jf30t
18:31:42 <DorpsGek_III> [OpenTTD/nml] glx22 updated pull request #115: Fix #112: prevent __version__.py regeneration for source package https://git.io/Jf3ZN
18:34:38 <_dp_> do SLE_* really include "to" version? would be much more logical if they didn't
18:34:53 <DorpsGek_III> [OpenTTD/nml] glx22 commented on pull request #115: Fix #112: prevent __version__.py regeneration for source package https://git.io/Jfslk
18:35:00 <_dp_> as how I'm supposed to know what it will be once it gets merged
18:36:07 <_dp_> well, what constant? current - 1?
18:37:46 <DorpsGek_III> [OpenTTD/OpenTTD] James103 commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/Jfsln
18:39:54 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/Jfsl8
18:41:46 <_dp_> ok, so just comments are all wrong
18:41:47 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #8105: Fix: OpenGFX download link did no longer work. https://git.io/Jfslz
18:41:51 <_dp_> "to Last savegame version that has the empty space."
18:43:10 <_dp_> though, actually, wouldn't that work even it was included?
18:44:03 <_dp_> if (_sl_version < sld->version_from || _sl_version >= sld->version_to) return false;
18:44:13 <glx> it was included, but now it's not
18:44:48 <glx> was changed for the SLV_ enum
18:45:05 <_dp_> yeah, that explains wrong comments
18:45:40 <glx> comments are always forgotten :)
18:46:01 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 merged pull request #8105: Fix: OpenGFX download link did no longer work. https://git.io/Jf3kP
18:48:19 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh updated pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfscP
18:56:58 <nielsm> did the CI get stuck? :(
18:58:46 <glx> probably communication issue between azure and github
18:59:09 <frosch123> it was the same with the other pr
18:59:16 <frosch123> after approval it said "all checks passed"
18:59:24 <LordAro> someone should hurry up and switch the CI over to Actions :)
18:59:34 <glx> oh it's because the checks are done in 2 passes
19:00:34 <glx> well they are split in "checks" section
19:01:26 <LordAro> well azure itself certainly seems to think everything's working fine
19:01:31 <LordAro> so must be communication issue
19:02:11 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/Jfs8m
19:04:53 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/Jfs8a
19:16:34 *** WormnestAndroid has quit IRC
19:16:51 *** WormnestAndroid has joined #openttd
19:26:30 *** tokai|noir has joined #openttd
19:26:30 *** ChanServ sets mode: +v tokai|noir
19:44:11 *** Extrems has joined #openttd
20:02:52 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfsB7
20:03:08 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth approved pull request #8107: Update: Clarify some contribution guidelines https://git.io/JfsBF
20:30:10 *** Wormnest has joined #openttd
21:09:24 <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on issue #8093: Regular server desynchronisations from clients https://git.io/JfkmV
21:15:45 <_dp_> another indication of some company id shenanigans happening
21:19:10 <Samu> are you running a gs in that game?
21:19:42 <glx> _dp_: but modified client and server so...
21:20:00 <_dp_> funnily enough of all the things goin on there GS is not involved
21:21:33 <_dp_> glx, yeah, but I've no idea how any of the modifications could cause that
21:21:44 <_dp_> also unmodified is known to get into that state as well
21:22:48 <glx> there's at least #8103, but maybe some other areas can cause desyncs
21:23:00 <glx> and desyncs are always hard to track
21:25:02 <_dp_> glx, but can #8103 cause spectator to desync?
21:26:17 <_dp_> I'll try to get command log if this keeps happening
21:26:33 <glx> I think it only affects de client causing the free wagon reordering
21:27:07 <glx> but if that happens on the server all clients are ckicked
21:27:23 <glx> (non dedicated server of course)
21:27:45 <_dp_> me too, so should be the other cause
21:28:47 <Samu> i remember there was a change in squirrel api related to retrieving a company id causing a crash
21:29:49 <Samu> but the fix made some changes in how it handles companies
21:30:02 <_dp_> glx, not every crash is easy :p
21:30:19 <glx> true, but way easier to trace
21:30:24 <FLHerne> andythenorth: I wrote an nml lexer in nodejs to see what performance would be like without Python overhead
21:30:24 <Samu> probably not related to this, but...
21:30:30 <FLHerne> It's about 30x faster
21:30:53 <_dp_> glx, not when it crashes in completely unrelated place due to some memory corruption
21:30:56 <FLHerne> Of course, rewriting all of nml in JS might take a little while
21:31:29 <LordAro> FLHerne: i can't imagine that's purely python overhead
21:31:43 <LordAro> there must be inefficiencies in nml's implementation that can be fixed
21:32:19 <FLHerne> LordAro: The lexer is pretty much just using PLY
21:32:43 <FLHerne> Doing the same thing repeatedly, in tight loops, on strings, is about CPython's worst possible case
21:32:56 <FLHerne> And one where JS's JIT *really* helps
21:33:08 <_dp_> js easily can 30x vs python
21:33:28 <FLHerne> PyPy is about 5x faster than CPython there
21:33:55 <FLHerne> So I guess JS being ~6x faster than PyPy is the relevant number for performance-obsessed people
21:34:26 <FLHerne> I did look at rewriting just the lexer using CFFI
21:35:10 <FLHerne> But calling back into Python to create a couple of objects for every token hurts a lot
21:35:22 <FLHerne> Well, more than a couple
21:38:51 <andythenorth> I never writing anything in it
21:38:56 <andythenorth> writing / write /s
21:39:07 <andythenorth> but I have it for web stack deps
21:39:26 <andythenorth> I swore I'd never write javascript again many times :(
21:41:59 <andythenorth> FLHerne subprocess to a node lexer?
21:42:21 <andythenorth> the problem is what tree would it produce?
21:42:36 <FLHerne> Lexer is just a stream, so not too bad
21:42:54 <andythenorth> the only thing I use node for is subprocess for css compiles :P
21:42:59 <FLHerne> Hm, that might be easier than the approach I was trying
21:43:34 <FLHerne> Should be in C rather than JS, so LordAro doesn't kick me
21:43:36 <LordAro> i would much rather time was spent optimising the existing python, rather than rewriting it in a brand new language
21:44:58 <LordAro> nml just isn't well tested enough to reliably rewrite it in another language
21:46:36 <andythenorth> it only has to produce the same results from the same input
21:46:48 <FLHerne> andythenorth: That's not really practical
21:47:26 <LordAro> testing that on a significant enough range of inputs is nearly impossible
21:47:39 <FLHerne> Well, testing *that* would be easy
21:47:59 <FLHerne> Grab all the nml source on the internets and see if you get the same result
21:48:49 <FLHerne> The problem is that you can get the same semantic result from different grf output, and are guaranteed to without spending ages painstakingly duplicating all of NML's design mistakes
21:48:50 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #8103: Fix #8093: Build+Refit changed game-state in command test run, and thus caused desyncs https://git.io/JfsuS
21:48:59 <FLHerne> Which would be stupid :P
21:50:36 <andythenorth> don't checksum the exact bytes
21:50:44 <andythenorth> decompile both grfs with grfcodec
21:50:59 * andythenorth can't remember if grfcodec always imposes same order regardless of source grf
21:51:20 <andythenorth> ultimately there's a set of branched actions, they're either the same, or they're not
21:51:43 <andythenorth> I am not 100% serious
21:52:49 * _dp_ already wants to rewrite citymania before even finishing the previous attempt
21:55:03 <milek7> LordAro: it criticises netscape
21:55:08 <milek7> but firefox was success, after all?
21:55:55 <LordAro> after about 15 years of total IE dominance
21:56:05 <LordAro> and now it's fading into obscurity again
21:56:35 <milek7> ~30% market share, obscurity?
21:57:07 <LordAro> oh, i was led to believe it was lower
21:57:28 <milek7> apparently I'm stuck in 2010
21:58:26 <milek7> huh, globally under 5%
21:58:45 <LordAro> that's more inline with what i thought
22:00:21 <andythenorth> FLHerne do we know what % of a compile is spent in the lexer?
22:01:04 <milek7> apparently globally it was highest at 30% in 2009
22:01:05 <DorpsGek_III> [OpenTTD/nml] matthijskooijman commented on pull request #115: Fix #112: prevent __version__.py regeneration for source package https://git.io/Jfszv
22:10:33 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 updated pull request #8103: Fix #8093: Build+Refit changed game-state in command test run, and thus caused desyncs https://git.io/JfOjg
22:15:26 <FLHerne> andythenorth: It's about 50%
22:17:18 <FLHerne> andythenorth: NML's NFO output is nothing like a direct mapping of the input
22:18:56 <andythenorth> my thinking was that ultimately, the order of bytes etc is irrelevant
22:19:02 <andythenorth> if the graph of actions comes out the same
22:20:03 <andythenorth> but I don't know if we have a tool that will graph the actions
22:22:11 *** Thedarkb-Desktop has joined #openttd
22:24:14 <andythenorth> FLHerne is this a serious investigation of node, or just amusement? :)
22:26:25 <frosch123> imagine firs would need python, chameleon, nml, node.js and grfcodec to build :)
22:26:38 <frosch123> can you add wasm somehow?
22:26:43 <Thedarkb-Desktop> Hi, I'm having an issue with my server.
22:27:15 <Thedarkb-Desktop> I've had luck compiling my own SDL stuff for WASM with Emscripten actually.
22:27:37 <Thedarkb-Desktop> I'm getting dbg: [net] [udp] sendto([master.openttd.org]:3978 (IPv6)) failed with: 101
22:27:46 <Thedarkb-Desktop> And I'm running 1.10.1
22:28:00 <Thedarkb-Desktop> Server's been up for years, I'm not sure why it's gone down.
22:28:08 <Thedarkb-Desktop> Well, why it's stopped pinging master.
22:28:57 <Thedarkb-Desktop> Nevermind, it fixed itself.
22:29:07 <Thedarkb-Desktop> I just restarted it twice...
22:36:49 <Thedarkb-Desktop> No, wait, still not showing up on the master list.
22:37:08 <Thedarkb-Desktop> Nevermind, it just came up there.
22:37:14 <Thedarkb-Desktop> I think my connection is unstable.
22:40:30 <DorpsGek_III> [OpenTTD/nml] glx22 commented on pull request #115: Fix #112: prevent __version__.py regeneration for source package https://git.io/Jfsgr
22:42:22 *** Wormnest has joined #openttd
22:46:29 <FLHerne> andythenorth: Not sure, tbh
22:46:55 <FLHerne> I do share LordAro's concerns about reinventing the wheel
22:50:20 *** supermop_Home_ has joined #openttd
22:52:18 <andythenorth> frosch123 it would all run in a virtual machine, in wasm, in your browser
22:55:54 <FLHerne> I think you could run nml in Python in wasm right now
22:56:00 <FLHerne> If you really wanted to
22:59:21 <Eddi|zuHause> i'm fairly convinced someone is already running python in wasm in some production system somewhere
23:05:26 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #8103: Fix #8093: Build+Refit changed game-state in command test run, and thus caused desyncs https://git.io/Jfsgh
23:16:47 <_dp_> can #7912 be merged already? I'm missing it so much...
23:18:33 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 merged pull request #8103: Fix #8093: Build+Refit changed game-state in command test run, and thus caused desyncs https://git.io/JfOjg
23:21:23 <frosch123> i think town names in nml are completely broken :)
continue to next day ⏵