IRC logs for #openttd on OFTC at 2021-03-17
⏴ go to previous day
00:11:04 <_dp_> oh, visual effects are part of the game state... that's "fun"
00:11:27 <_dp_> what happens if you try make a train out of smoke I wonder xD
01:02:26 *** WormnestAndroid has quit IRC (Remote host closed the connection)
01:04:40 *** WormnestAndroid has joined #openttd
01:18:59 *** gelignite has quit IRC (Quit: Stay safe!)
02:05:08 <Wuzzy> any MIDI experts in here? is there a software to automatically verify if a MIDI file is 'valid'?
02:56:19 *** jgx has quit IRC (Remote host closed the connection)
03:00:29 *** Wormnest has quit IRC (Quit: Leaving)
03:51:26 *** debdog has quit IRC (Ping timeout: 480 seconds)
04:24:54 *** didac has quit IRC (Remote host closed the connection)
04:36:22 *** snail_UES_ has quit IRC (Quit: snail_UES_)
05:31:59 *** Wuzzy has quit IRC (Remote host closed the connection)
05:51:49 *** didac has quit IRC (Remote host closed the connection)
07:22:46 *** sam__ has quit IRC (Remote host closed the connection)
07:23:35 *** andythenorth has joined #openttd
07:56:06 *** sla_ro|master has joined #openttd
08:09:21 *** jottyfan has joined #openttd
08:15:03 <DorpsGek> [OpenTTD/OpenMSX] nielsmh commented on pull request #15: Change: Stop calling OpenMSX a 'replacement' https://git.io/JmvXY
08:18:16 <DorpsGek> [OpenTTD/OpenSFX] nielsmh commented on pull request #36: Change: Stop calling OpenSFX a 'replacement' https://git.io/Jmv5Z
08:40:29 *** jottyfan has quit IRC (Quit: jottyfan)
08:41:14 *** HerzogDeXtEr has joined #openttd
09:15:26 *** jottyfan has joined #openttd
09:19:44 <DorpsGek> [OpenTTD/OpenMSX] orudge commented on pull request #15: Change: Stop calling OpenMSX a 'replacement' https://git.io/JmULP
09:20:31 <DorpsGek> [OpenTTD/OpenMSX] Eddi-z commented on pull request #15: Change: Stop calling OpenMSX a 'replacement' https://git.io/JmUqG
09:22:17 *** andythenorth has left #openttd
09:51:03 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
11:44:33 <Xaroth> that generally means that you have something misconfigured.
11:57:03 *** snail_UES_ has joined #openttd
12:30:54 <spnda> TrueBrain: Is the Bearer auth token from the bananas-frontend-cli the access token from the GitHub API, or what is it? (Specifically github.com/login/oauth/access_token)
12:31:01 <_dp_> interesting, it doesn't desync when command fails on the client
12:32:39 <spnda> TrueBrain: The bananas-frontend-cli also gets the access token from there, but mine is a lot longer and bananas-api does not like it.
12:33:17 <_dp_> though I guess it makes sense now that I think of it as by the time command is sent to clients it's not executed yet
12:40:27 <spnda> ah the bananas-api does if len(bearer_token) != 32:... Mine is 40 chars long
12:44:42 <spnda> oh so the bananas-api handles the github authentication.... but what if I don't want to do that?
13:02:22 <Xaroth> then you make your own bananas, with blackjack and apples.
13:04:41 <supermop_Home> apparently ohio is opening vaccines to everyone over 16 next week... whereas new York isn't even on over-60 yet
13:06:46 *** Kitrana1 has joined #openttd
13:07:25 <supermop_Home> power icons to left or right of purchase menu sprite?
13:11:46 *** Kitrana has quit IRC (Ping timeout: 480 seconds)
13:38:58 *** Pendoluso has joined #openttd
13:39:23 <Pendoluso> who can help me translate a word using gender ?, i'm having errors
13:40:08 <Pendoluso> {G 0 "outro" "outra" "outros" "outras" }
13:40:17 <Pendoluso> gives me: Error: Expected 5 string arguments for {G ..}, found 5 arguments
13:41:21 <LordAro> well that error message is unhelpful :D
13:41:59 <LordAro> looking at other languages, the '0' is the issue?
13:43:27 <LordAro> oh i see, the '0' is necessary
13:43:34 <LordAro> in which case the issue is that there are only 4 genders there
13:46:20 <Pendoluso> ok, i'll check other languages, to see how to fix, thanks
13:55:33 <spnda> wtf? Exception: redirect_uri is invalid for this client_id
13:55:43 <spnda> "if not redirect_uri.startswith(_clients[client_id]):"
13:55:47 <spnda> but why would that be??
13:56:43 <spnda> the redirect is localhost and so does bananas-frontend-cli do it aswell
13:57:07 *** D-HUND is now known as debdog
13:57:12 <spnda> len(cmd_args) = length of cmd_args string
13:57:28 <LordAro> Pendoluso: internal eints stuff you (as a translator) don't need to worry about
13:57:38 <LordAro> spnda: cmd_args is an array, but more or less the same thing
13:57:46 <spnda> ah, couldn't have known lol
13:58:20 <spnda> also why did anyone decide to use "ape" as a client id for bananas-frontend-cli
14:01:17 <spnda> TrueBrain: Could you explain why the API throws "redirect_uri is invalid for this client_id"? redirect_uri is "https://localhost:3977/". client_id is "ape"
14:01:56 *** Pendoluso has quit IRC (Quit: Page closed)
14:04:08 *** gelignite has joined #openttd
14:05:08 *** Gustavo6046 has joined #openttd
14:21:23 *** Gustavo6046_ has joined #openttd
14:23:49 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
14:23:49 *** Gustavo6046_ is now known as Gustavo6046
15:00:02 <DorpsGek> [OpenTTD/OpenMSX] Wuzzy2 commented on pull request #15: Change: Stop calling OpenMSX a 'replacement' https://git.io/JmOPQ
15:09:00 <Wuzzy> Does eints automatically update when strings in a baseset like OpenSFX have changed?
15:14:31 <DorpsGek> [OpenTTD/OpenMSX] orudge requested changes for pull request #15: Change: Stop calling OpenMSX a 'replacement' https://git.io/Jm3JD
15:34:55 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
15:44:07 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
16:24:25 *** Gustavo6046 has joined #openttd
16:37:18 *** Gustavo6046 has joined #openttd
16:40:38 *** fnutt has quit IRC (Quit: Ping timeout (120 seconds))
16:41:05 <DorpsGek> [OpenTTD/OpenSFX] LordAro commented on issue #37: Make new release when enough new translations arrive https://git.io/JmOpx
16:43:30 *** Gustavo6046 has joined #openttd
17:14:30 *** fnutt has quit IRC (Quit: Ping timeout (120 seconds))
17:14:39 <spnda> I'm still very much confused about it getting triggered
17:26:15 <milek7> seems like standard oauth2 error?
17:28:17 <spnda> I don't see why it does it though. Why is it checking if my redirect uri includes the client id. Those don't have to be related?
17:28:41 <spnda> Or rather the IP of the client
17:36:08 <milek7> it seems to just check whether given redirect_uri has valid prefix for this client_id
17:36:51 <spnda> that.... should be checked for in the api possibly?
17:45:35 *** frosch123 has joined #openttd
17:45:52 <milek7> maybe it doesn't like https
17:48:21 *** Wormnest has joined #openttd
18:20:51 *** gelignite has quit IRC (Quit: Stay safe!)
18:24:30 *** Progman has joined #openttd
18:39:47 *** andythenorth has joined #openttd
18:48:15 <_dp_> either gs or savegame magic
19:03:43 <DorpsGek> - Update: Translations from eints (by translators)
19:07:59 * andythenorth wonders how much of JGRPP is 'just GS'
19:30:52 <DorpsGek> [OpenTTD/OpenTTD] ldpl opened pull request #8872: Fix: Recompute road/railtype availability after disabling the engine https://git.io/JmcGg
19:37:47 <LordAro> Maximum map size is now 256M tiles, ranging from 16k x 16k to 256 x 1M.
19:37:48 <LordAro> (The NewGRF debug inspection window is disabled for all map coordinates longer than 27 bits).
19:47:19 *** gelignite has joined #openttd
19:54:50 *** jottyfan has joined #openttd
19:56:01 *** jottyfan has joined #openttd
20:22:44 <_dp_> I think I invented a new way to debug desyncs
20:23:00 <_dp_> take 2 saves from the server and advance the first one with command log from admin port
20:37:35 <LordAro> interesting - this (debug build, mingw) game appears to be "stuck" to ~4x on FFWD, despite the framerate window showing no significant time spent anywhere
20:37:46 <LordAro> 1ms in graphics rendering, some of the time, maybe
20:39:39 <LordAro> 1ms in graphics rendering, some of the time, maybe
20:40:49 <LordAro> seems to be the FFWD limiter
20:40:59 <LordAro> it's set to 2500%, but is only going 4x
21:55:55 <TrueBrain> LordAro: you can always disable it; but found out what was causing it?
21:57:28 *** Samu has quit IRC (Read error: Connection reset by peer)
22:14:39 *** andythenorth has quit IRC (Quit: andythenorth)
22:14:53 <LordAro> TrueBrain: not really, it seems like there's some sort of calculation error in the setting
22:15:04 <LordAro> i didn't investigate further
22:17:08 *** Progman has quit IRC (Remote host closed the connection)
22:18:49 *** sla_ro|master has quit IRC ()
22:21:32 <TrueBrain> quick to blame the setting :P No, it is interesting if disabling it does work
22:21:45 <TrueBrain> as we use steady_clock, and I wonder when we find the first that is a non-high-precision variant :P
22:22:33 <TrueBrain> as I kinda have been waiting for that to happen ....
22:23:34 <TrueBrain> so if you have some time, some investigation might be nice :D
22:23:44 <TrueBrain> if not .. shrug .. MSYS :P
22:24:50 <TrueBrain> (and just to rule it out, 2500% is 25x for me, so it does "work for me" :P)
22:25:12 <TrueBrain> mostly I still wonder if we kinda need that ugly typedef you had :D
22:42:47 *** Gustavo6046_ has joined #openttd
22:42:48 *** Gustavo6046 has quit IRC (Read error: Connection reset by peer)
22:42:50 *** Gustavo6046_ is now known as Gustavo6046
22:43:46 *** jottyfan has quit IRC (Quit: jottyfan)
22:46:02 <LordAro> TrueBrain: you might be right about the clocks - up to 1200% works fine
22:46:08 <LordAro> beyond that, it just seems to get slower again
22:49:04 <LordAro> couldn't possibly be an overflow somewhere, could it?
22:49:35 <TrueBrain> steady_clock overflows every 238 years or so
22:49:50 <TrueBrain> (at least, I believe)
22:50:06 <TrueBrain> but you could try doing a sed for steady_clock to .. what was the name of the high-res?
22:50:22 <TrueBrain> otherwise, you would have to TICC/TACC the sleep
22:50:30 <TrueBrain> it is also possible the sleep_for uses the wrong timer resolution
22:50:37 <TrueBrain> I do not know how that was fixed for windows
22:50:59 <LordAro> 2000% - 130fps, 5000% - 200fps, 10000% - ~330fps, 20000% - 660fps
22:51:16 <LordAro> this is a game that will do 50000fps when unlimited
22:51:23 <LordAro> (fps being simulation rate)
22:51:55 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:52:11 <TrueBrain> well, doing a TICC/TOCC around sleep_for in video_driver.cpp might be most useful, now I think about it
22:52:16 <TrueBrain> with a debug before how long it should sleep
22:53:28 <LordAro> is that the one in GameThread or SleepTillNextTick ?
22:53:42 <TrueBrain> owh, I forgot about that
22:53:45 <TrueBrain> that is already in master ofc
22:53:54 <TrueBrain> start with -vsdl-opengl:no_threads
22:54:05 <TrueBrain> sorry ... sleepy head
22:54:27 <TrueBrain> anyway, fps is not that important with the limiter, as in, an annoying value, as I need to divide it by 33.33 :P
22:54:37 <TrueBrain> the x simulation speed is more useful to quickly see what is going on
22:54:45 <DorpsGek> TrueBrain: 3.9039039039039043
22:55:00 <TrueBrain> so at 2000% it is 4x?
22:55:07 <TrueBrain> lol .. something is missing a timer :D
22:55:16 <DorpsGek> TrueBrain: 6.0060060060060065
22:55:42 <LordAro> no, it's weirder than that
22:56:00 <LordAro> and now with win32-opengl:no_threads it's even weirder
22:56:08 <TrueBrain> win32, ah, sorry, MSYS :P
22:56:18 <LordAro> 10000% & 15000% work fine, 100x & 150x
22:56:36 <TrueBrain> that sounds like a timer resolution
22:56:42 <TrueBrain> so, with no_threads we only have to debug SleepTillNextTick
22:56:45 <TrueBrain> and use TIC/TOC plz
22:56:48 <TrueBrain> not TICC/TOCC, I realise
22:56:55 <LordAro> that wasn't the case with the threads
22:56:55 <TrueBrain> as TICC/TOCC uses the same steady_clock :D
22:57:13 <TrueBrain> yeah, but with threads the two threads interact with each other
22:59:21 <TrueBrain> most unusual .. I wonder what MSYS implements as steady_clock etc
22:59:48 <TrueBrain> something worth checking btw, if a nightly has the same issue on your machine
23:00:41 <TrueBrain> "As a general rule, if you don't want surprises always use boost instead of libstdc++ on Windows." <- I love random advise from the internet :D
23:04:07 *** frosch123 has quit IRC (Quit: be yourself, except: if you have the opportunity to be a unicorn, then be a unicorn)
23:07:17 <TrueBrain> @calc 2**64 / 1000 / 1000 / 1000 / 3600 / 24 / 365
23:07:17 <DorpsGek> TrueBrain: 584.9424173550719
23:07:30 <TrueBrain> 584 years fit in std::chrono on GCC
23:07:37 <TrueBrain> so wrapping shouldn't be a problem :P
23:08:09 <LordAro> i wasn't suggesting that chrono was wrapping
23:08:20 <TrueBrain> no, I was curious if that could be a cause :D
23:08:26 <LordAro> but if something's getting coerced to a type that it shouldn't be...
23:08:53 <TrueBrain> not that I am aware; everything should be std::chrono types now
23:09:08 <TrueBrain> but the way you describe the issue, sounds to me that sleeping isn't as accurate as we would like
23:09:43 <TrueBrain> as the higher you go, at a certain point you hit a limit and it just doesn't sleep :P
23:11:26 <TrueBrain> but with threads the drawing thread is always holding that back a bit, as it will keep doing its 60fps
23:12:35 <TrueBrain> anyway, my wishlist for things to check, if you have the time for it ofc: a nightly, see if that works, and otherwise TIC/TOC around sleep_for with a debug of the duration it should be sleeping. Most interesting :D
23:13:14 <TrueBrain> another fun thing to check: change the refresh rate with threads enabled
23:14:27 <TrueBrain> also, if you haven't already: disabling hardware acceleration
23:14:36 <TrueBrain> who knows what OpenGL is doing to your poor computer :D
23:15:08 <LordAro> trouble is, linking takes over a minute because lol msys
23:15:34 <TrueBrain> double checked, all variables are either std::chrono or "auto"
23:15:42 <TrueBrain> the latter of course can be wrong, but I assume it isn't :D
23:18:13 *** nielsm has quit IRC (Ping timeout: 480 seconds)
23:19:30 <TrueBrain> owh, I forgot, TIC/TOC is in a weird unit on linux :P
23:20:07 <LordAro> yeah, i've just used TICC/TOCC anyway, just so i can actually compare numbers
23:20:17 <LordAro> since i'm checking sleep_for, rather than clock resolution
23:20:40 <LordAro> no difference with win32 vs win32-opengl btw
23:20:54 <TrueBrain> yeah, fair, and in the intro game it should be ~14ms
23:21:57 <TrueBrain> should always overshoot
23:22:20 <TrueBrain> overshoots for me with 0.1ms
23:22:53 <TrueBrain> [2021-03-18 00:22:32] dbg: [misc] 15119
23:22:53 <TrueBrain> [2021-03-18 00:22:32] dbg: [misc] [sleep] 1522101 us [avg: 15221.0 us]
23:22:57 <TrueBrain> is what I get when FFing
23:23:06 *** gelignite has quit IRC (Quit: Stay safe!)
23:24:07 <LordAro> that's pretty close to what i've got, though i abused the TICC counters instead :)
23:24:18 <TrueBrain> I have done that a few times too :P
23:24:35 <TrueBrain> okay, so not sleep_for .., and the resolution of steady_clock is fine too, clearly
23:25:28 <TrueBrain> testing a nightly would rule out your hardware in general
23:25:37 <LordAro> i think so, but i'm not sure yet
23:27:05 <TrueBrain> with no_threads, and my/your changes, at 2500%, sleep_for should be ~10ms
23:27:32 <TrueBrain> [2021-03-18 00:27:22] dbg: [misc] 924
23:27:32 <TrueBrain> [2021-03-18 00:27:22] dbg: [misc] [sleep] 102119 us [avg: 1021.2 us]
23:28:05 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
23:28:29 <TrueBrain> 825 frames per second, so yeah, 1ms give or take
23:28:51 <LordAro> ok, nightly works "better", but still not perfectly
23:29:22 <LordAro> actually, no, it's weirder again
23:29:27 <TrueBrain> as I can imagine variable CPU speeds doing weir dstuff
23:30:26 <LordAro> 5k% - 50x, 10k% - 64x, 20k% - 30x, 30k% - 31x
23:30:40 <TrueBrain> let me check what it does for me ..
23:30:55 <LordAro> possible, but it's not throttling it :p
23:30:56 <TrueBrain> disabling autosave for more reliable numbers ...
23:31:28 <TrueBrain> ugh, we should add an on-change handler to fast-forward for this kind of testing :P
23:31:52 <LordAro> yes, remember to stop FF before changing the setting :p
23:32:10 <TrueBrain> I max out at 80x simulation
23:32:39 <TrueBrain> but it doesn't do what your computer is doing :D
23:32:55 <LordAro> use a smaller map, and disable animation :p
23:34:26 <TrueBrain> okay, now I have something similar
23:34:32 <TrueBrain> with smaller map, it goes slower :D :D
23:35:16 <LordAro> there's a reason my games start at 2100 :p
23:35:18 <TrueBrain> did not know it was THAT insane :D
23:35:53 <TrueBrain> so with 10000% I get 60x
23:35:56 <TrueBrain> with 20k% I get 30x
23:36:43 <TrueBrain> it is a Windows thing
23:36:47 <TrueBrain> works flawless on Linux (well, WSL)
23:36:58 <LordAro> at a certain point it may be that calling SleepUntilNextTick at all might be what's slowing it down, nevermind what sleep is doing :p
23:37:14 <TrueBrain> but you should see that back in the deltas
23:37:15 <LordAro> but good that you can reproduce it :p
23:37:56 <TrueBrain> yeah, at least it is not MSYS / MSVC
23:38:00 <TrueBrain> or your hardware :D
23:38:19 <TrueBrain> can't believe I hit 2050 in a few seconds .. I have not been playing this with a release build :D
23:39:04 <TrueBrain> LordAro: silly enough, bigger maps seems to help for the max speed I can get while limiting :D
23:40:03 <LordAro> we did tell you it was ludicrous :p
23:40:19 <LordAro> i managed 1million fps on a tiny map
23:40:27 <TrueBrain> yeah ... on debug it already was .. did not know it was THIS ludicrous ... SpaceBalls would be proud
23:41:42 <TrueBrain> at Linux, it caps off at ~500x
23:42:05 <TrueBrain> that is the max value
23:42:46 <TrueBrain> would it have to do with thread-switching?
23:42:51 <TrueBrain> hmm, no, no_threads had the same issue
23:43:48 <LordAro> you'll be pleased to know that s/steady_clock/high_resolution_clock/ makes no difference
23:45:07 <TrueBrain> would have been the easy solution
23:45:41 <TrueBrain> you can disable the sleep_for btw; in both threads
23:45:48 <TrueBrain> might be worth a try
23:46:12 <TrueBrain> euh, no, in game-thread it will break
23:46:19 <TrueBrain> so only works with no_threads :P
23:46:41 *** spnda has quit IRC (Quit: Page closed)
23:47:20 <TrueBrain> okay, can reproduce in debug settings too, so I can debug this myself now too :D
23:47:40 <TrueBrain> holy crap, sleep is overshooting
23:47:45 <TrueBrain> dbg: [misc] [sleep] 189150 us [avg: 1891.5 us]
23:47:58 <TrueBrain> yeah .. sleep_for is an issue on Windows :D
23:48:57 <TrueBrain> it should also show you that, honestly
23:48:57 <LordAro> did that timePeriod thing get added at all?
23:50:53 <TrueBrain> timeBeginPeriod(1);
23:51:01 <TrueBrain> sets it to 1ms, comment says
23:52:05 <TrueBrain> that limits the fps for Windows heavily
23:56:31 <TrueBrain> very hard to hit deadlines that way, damn
23:58:50 <LordAro> no real way around it either
23:58:56 <TrueBrain> that mitigates it a bit
23:59:00 <TrueBrain> by making it spinlock more
23:59:09 <TrueBrain> I have another idea, but that requires more code
23:59:41 <LordAro> but ultimately not a huge issue, since you're fastforwarding anyway
continue to next day ⏵