IRC logs for #openttd on OFTC at 2023-08-13
00:30:01 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
00:40:14 *** Tirili has quit IRC (Quit: Leaving)
01:06:19 <kamnet> Just curious, has anybody suggested creating a global scoreboard?
01:07:42 <kamnet> I think if we had such a feature that it might help eliminate the feeling that the game is all about money or give a more visible objective to people who get bored too easily.
01:08:35 *** dale has quit IRC (Ping timeout: 480 seconds)
01:34:05 <_jgr_> What would the "score" be?
01:34:28 <_jgr_> The in-game performance score is basically meaningless
01:37:28 *** _aD has quit IRC (Ping timeout: 480 seconds)
02:34:44 <leothejewishturtle>
02:34:44 <leothejewishturtle>
02:34:44 <leothejewishturtle> Can anyone help me figure out why I can't compile the openTTD source code? I made a STL test project just to see if I could display a window and it works, so why wont it compile and launch openttd?
02:36:35 <leothejewishturtle> First image is showing my STL test project and the green debug label and the second picture shows the OpenTTD source code I imported to VS and the debug label is greyed out
02:53:59 *** D-HUND has joined #openttd
02:57:20 *** debdog has quit IRC (Ping timeout: 480 seconds)
04:45:50 *** D-HUND is now known as debdog
04:52:14 *** keikoz has joined #openttd
05:01:54 *** HerzogDeXtEr has joined #openttd
05:50:44 *** nielsm has joined #openttd
06:01:00 *** bryjen has quit IRC (Quit: Leaving)
06:09:54 <andythenorth> Max number of statues built
06:23:33 <DorpsGek> [OpenTTD/OpenTTD] Bouke opened issue #11193: [Crash]: nightly crashes after running wentbourne for ~10s
06:43:14 <debdog> OpenRCT2 sucks
06:43:30 * debdog goes back to OTTD
06:43:55 <andythenorth> When is OpenTropico?
06:44:34 <andythenorth> THH I didn’t finish the paud version yet, and it has a sandbox mode also
06:44:52 <debdog> yes
06:48:21 * debdog sings Roller Coaster suck to the tune of "Water sucks, it really really sucks...."
07:00:07 <DorpsGek> [OpenTTD/OpenTTD] Bouke opened issue #11194: [Bug]: performance regression between 13.4 and nightly
07:07:32 <DorpsGek> [OpenTTD/OpenTTD] LordAro commented on issue #11194: [Bug]: performance regression between 13.4 and nightly
07:30:06 <DorpsGek> [OpenTTD/OpenTTD] Bouke commented on issue #11194: [Bug]: performance regression between 13.4 and nightly
07:52:43 *** asymptotically2 has joined #openttd
08:16:41 <truebrain> that moment `vcpkg install python3` fails with the error: please install python3
08:16:43 <truebrain> lol
08:26:19 <_zephyris> Base graphics question for this lovely Sunday morning - what are people's opinions on my OpenGFX2? It's a bit of a style shift, so keep separate from original OpenGFX? Or try to merge/update the OpenGFX repo?
08:27:00 <alfagamma_0007> It's nice
08:27:09 <alfagamma_0007> Looks exactly like original graphics
08:27:28 <alfagamma_0007> Do you intend to redraw the default vehicles as well??
08:27:58 <truebrain> _zephyris: choices is always good. Some people will like it. Others won't. So in my opinion, call it OpenGFX2 or something completely different πŸ™‚
08:28:33 <alfagamma_0007> Though I am accustomed to opengfx 1 , so I will stick with it
08:28:44 <alfagamma_0007> Now a big fan of extrazoom
08:32:01 <ahyangyi> opengfx2 has a non-ez version as well
08:32:14 <truebrain> the irony is so lovely πŸ™‚
08:36:15 *** Wolf01 has joined #openttd
08:40:39 <DorpsGek> [OpenTTD/grfcodec] MatthewGentoo opened pull request #23: data: Fix Action5 table
08:45:23 <alfagamma_0007> truebrain: Floor is made of floor made of floor???
08:45:35 <alfagamma_0007> Badger badger badger
08:47:13 <DorpsGek> [OpenTTD/grfcodec] MatthewGentoo commented on pull request #23: data: Fix Action5 table
08:50:23 <DorpsGek> [OpenTTD/grfcodec] MatthewGentoo updated pull request #23: data: Fix Action5 table
08:55:58 <truebrain> oof .. more vcpkg problems ... "lzo" is downloaded from an URL that is no longer functional
08:56:12 <truebrain> the domain no longer exists even
08:56:55 <truebrain> means I can steal the domain, I guess πŸ˜›
08:57:04 *** Flygon has joined #openttd
08:59:10 <truebrain> ah, no, their provider is just offline πŸ˜„
09:00:41 <truebrain> it is claimed to be a cloud, the host used .. but it doesn't do anything anymore πŸ˜›
09:00:43 <truebrain> poor cloud
09:00:47 <truebrain> someone closed their MBP
09:13:14 *** Eddi|zuHause has quit IRC (Remote host closed the connection)
09:17:50 *** Eddi|zuHause has joined #openttd
09:18:31 <truebrain> hmmm ... seems my change of `InitializeWindowViewport` from yesterday now causes a crash
09:18:32 <truebrain> lovely
09:19:09 <truebrain> but I tested ittttttt 😦
09:19:42 <truebrain> ugh, what a stupid mistake
09:21:34 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain opened pull request #11195: Fix 0238a2b5: crash when news message is displayed
09:25:09 <Eddi|zuHause> one of the mod updates of the last few months seems to cause an endless loop in cities:skylines... how the hell would i isolate which one?
09:52:06 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler approved pull request #11195: Fix 0238a2b5: crash when news message is displayed
10:03:14 *** _aD has joined #openttd
10:16:40 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain merged pull request #11195: Fix 0238a2b5: crash when news message is displayed
10:59:18 <andythenorth> Is cat?
11:02:33 <DorpsGek> [OpenTTD/OpenTTD] PeterN commented on issue #11194: [Bug]: performance regression between 13.4 and nightly
11:04:08 *** gelignite has joined #openttd
11:04:47 <peter1138> Is lunch?
11:07:48 *** Flygon has quit IRC (Remote host closed the connection)
11:09:00 *** Flygon has joined #openttd
11:10:24 <truebrain> sure!
11:14:42 <truebrain> for the first time ever compiling stuff on my new server I bought .. months ago. It is scray fast in comparison to my desktop ..
11:14:57 <truebrain> AMD 5600X
11:15:12 <truebrain> versus my normal i7-8700K
11:15:29 <truebrain> weird, how a few years of CPU development makes a difference πŸ˜„
11:16:43 <peter1138> Kinda annoying. Buying the best at the time is just expensive, and buying with an eye to upgrading later is pointless as nothing is compatible with it later...
11:18:24 <truebrain> yeah .. AMD did try to break that a bit, but yeah ...
11:18:30 <truebrain> so I bought the 5600X as it was a nice balance
11:18:38 <truebrain> but I didn't want to replace my desktop .... now I am tempted πŸ˜›
11:20:12 <truebrain> wow, vcpkg is quick in fixing issue .. someone already made a PR to address the Python3 issue ..
11:20:14 <truebrain> that is nice
11:37:22 <truebrain> running OpenTTD on the other hand, is faster on my 8700K than on my 5600X πŸ˜„
11:37:25 <truebrain> shocker πŸ˜›
11:39:30 <truebrain> ` Unknown ICU component: data` ugh bah .. I was trying to build an older OpenTTD version
11:39:33 <truebrain> can I fix that .. hmmm
11:40:34 <truebrain> we did something weird, where we validated we also knew all ICU components .. and we didn't know about `data`
11:40:37 <truebrain> so totally irrelevant error πŸ˜„
11:46:16 <truebrain> `openttd-12.0-beta1~16-linux-generic-amd64` .. lol .. that is an unexpected version number πŸ˜›
11:50:07 <andythenorth> peter1138: Was lunch.
11:50:13 <andythenorth> Salad
11:50:20 <andythenorth> At Sacre Couer
11:50:29 *** WhilelM has left #openttd
12:02:53 <alfagamma_0007> What sort of salad
12:05:47 <LordAro> andythenorth: peter1138: was sausage & black pudding sandwich
12:05:52 <LordAro> is good
12:09:08 <talltyler> Two sausage sandwich?
12:09:12 <peter1138> Leafy greens with tomato/cucumber, etc...
12:09:18 <talltyler> Anything else besides meat and bread?
12:09:33 <peter1138> And then I had an almond croissant to ruin it all, because it was breakfast as well as lunch
12:09:47 <peter1138> talltyler: usually butter πŸ˜„
12:10:35 <peter1138> ketchup or brown sauce if the sausage is cheap & nasty
12:12:06 <LordAro> talltyler: there were 2 sausages in it, yes
12:12:23 <LordAro> talltyler: a boat load of butter too
12:16:10 <talltyler> Sounds tasty to me, although my girlfriend made increasingly disgusted faces the more ingredients I relayed to her
12:17:17 <LordAro> :D
12:17:40 <LordAro> peter1138: good speed yesterday, btw
12:21:54 <peter1138> My legs were telling me about it, hah!
12:43:47 <truebrain>
12:44:03 <truebrain> I am slowly filling in the points between the current points, but it seems in the last 2 years we got a bit faster
12:44:34 <truebrain> this is a lot of compiling and running πŸ™‚
12:48:01 <truebrain> (this is from just before 12.0-beta1 till now; all compiled as final release builds (so asserts disabled))
12:48:33 <truebrain> I am mostly surprised that the memory footprint went down a lot .. we will see what commit did that πŸ˜„
12:52:00 <locosage> most title games don't seem to be particularly meaningful for performance
12:52:17 <locosage> I can send you some big games if you want
12:54:20 <locosage> also, would be interesting to see from like 1.0
12:54:50 <locosage> don't have 1.0 saves though xD
12:54:53 <truebrain>
12:56:36 <DorpsGek> [OpenTTD/OpenTTD] arikover opened pull request #11196: Fix: Max road speed is doubled in buy menu dropdowns
12:57:12 <truebrain>
12:57:12 <truebrain> my bisect algorithm is working fine πŸ˜„
12:57:20 <truebrain> slowly adding points in the middle of all other points πŸ™‚
12:57:56 <truebrain> (this was with wentbourne deselected, in case that isn't all that clear)
12:58:15 <truebrain> very small games use a bit more memory, but everything else just plummeted
12:58:27 <truebrain> I suspect it is one of peter1138 's saveload vector thingy patches πŸ™‚
12:58:38 <LordAro> that is notable
12:58:49 <LordAro> truebrain: do JGRPP as well ;)
12:59:58 <truebrain> finger .. hand ...
13:03:04 *** xarothbrook has joined #openttd
13:03:04 <xarothbrook> arm
13:06:34 <locosage> that was unexpected summoning
13:12:29 <xarothbrook> summoning?
13:14:31 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain commented on issue #11194: [Bug]: performance regression between 13.4 and nightly
13:14:34 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain closed issue #11194: [Bug]: performance regression between 13.4 and nightly
13:15:22 <truebrain> that was a lot of work to know for sure there wasn't a regression as far as I can measure πŸ˜„
13:16:31 <truebrain> I wonder if I can get GitHub to run this every night, and update the statistics ... I am sure I can πŸ˜›
13:47:22 <truebrain> the 1.4 title game is really odd .. does random shit from time to time πŸ˜›
13:50:38 <truebrain>
13:50:38 <truebrain> more datapoints
14:03:35 <pickpacket> Does anyone happen to have the windows binary of 13.3 at hand?
14:05:07 <truebrain>
14:05:09 <truebrain> for all your binaries
14:06:18 <pickpacket> nice! Thanks! Which version should I pick, though? win32, win64, or amd64?
14:07:55 <talltyler> Depends on your computer πŸ™‚
14:08:15 <pickpacket> :P
14:08:33 <pickpacket> in this case my gf's computer, which I know nothing about and she knows even less
14:08:42 <pickpacket> trying the win64 and we'll see what happens
14:09:02 <xarothbrook> win64 most likely
14:09:10 <xarothbrook> if that fails, try win32
14:09:24 <pickpacket> xarothbrook: that's what I figure too, thanks :)
14:09:31 <pickpacket> what's special about amd64?
14:09:41 <xarothbrook> it's arm64
14:09:49 <xarothbrook> which is not a common cpu
14:10:41 <pickpacket> oh! I misread XD
14:10:43 <pickpacket> that makes sense
14:10:54 <pickpacket> windows on arm must be very unusual
14:11:30 <emperorjake> It's ARM not AMD, not a lot of windows devices have ARM cpus
14:26:13 <LordAro> truebrain: how odd. cargodist, perhaps?
14:26:25 <LordAro> or some other "mildly more representative of a real game"? :p
14:27:25 <truebrain> dunno .. it is the only one that is actually not all that stable
14:27:34 <truebrain> so yeah, something realtime is influencing the behaviour there
14:28:06 <truebrain> but a bit dissapointing that this channel is the same as always .. bitching about what should be different, instead of enjoying what is given πŸ˜›
14:31:52 <xarothbrook> Bitching is easy, being helpful takes effort.
14:39:21 <brickblock19280> where do trains and roadvehicles read bridge speed limits
14:41:48 <truebrain> hmm, also seems that the 1.4 title game didn't work for a month or so .. also the memory drops while the CPU usage drops .. πŸ˜„
14:50:39 <truebrain>
14:50:39 <truebrain> and another pass finished to increase the resolution of the graph .. the next one will take a lot longer πŸ˜›
14:51:34 <truebrain>
14:51:34 <truebrain> without westbourne, as it is making the title games invisible πŸ˜›
14:52:19 <truebrain> build times in 2021 were a lot shorter than now .. despite the tricks we added to make compiling faster πŸ˜„
15:11:59 <truebrain> LordAro: 1.4 doesn't have cargodist enabled πŸ˜›
15:12:38 <truebrain> setting-wise it is very default-ish
15:14:43 <truebrain> all other games have a jitter of like 0.02% .. the 1.4 games has one of over 3% πŸ˜›
15:14:47 <truebrain> something is very weird with that game
15:21:52 <peter1138> Hi.
15:21:55 <peter1138> What have I done?
15:22:09 <truebrain> good things πŸ™‚
15:22:15 <peter1138> Odd.
15:23:06 <truebrain> memory usage of OpenTTD has been going down pretty drastically πŸ˜›
15:23:17 <truebrain> somewhere in March
15:23:24 <truebrain> so I was guessing it was one of those vectorize patches πŸ˜„
15:27:02 <andythenorth> Patch probably made GS faster also?
15:28:12 <truebrain> lot of commits in the range where it changed .. so I will let me script run to narrow it down a bit more πŸ™‚
15:28:17 <truebrain> will take hours and hours .. if not days πŸ˜›
15:34:30 <peter1138> Yeah, it's fine, I'm happy to take it πŸ˜‰
15:50:07 <locosage> [2023-08-13 15:04:41] dbg: [net] Connection with Game Coordinator lost; reconnecting...
15:50:07 <locosage> [2023-08-13 15:04:41] dbg: [net] Connected to
15:50:07 <locosage> [2023-08-13 15:04:57] Another server with the same invite-code registered itself. Switching to "local" game-type.
15:50:14 <locosage> multiple servers went local for no reason
15:52:03 <truebrain> That is not for no reason πŸ™‚ seems your connection cut, so we have two incoming TCP connections claiming to be the same server .. sadly that is a hard problem to solve to know which is the right one
15:52:22 <truebrain> Clearly it picked the wrong one here
15:52:44 <truebrain> (It kicks one of the two to "local" to avoid constant battles)
15:52:44 <locosage> there are like 100plus servers missing on the list
15:53:13 <locosage> no bt-pro, no n-ice
15:58:42 <locosage> truebrain: maybe do 1 retry? dead connection won't fight back
15:58:49 <truebrain> I love the entitled done you have here πŸ™‚ really productive πŸ˜›
15:59:47 <truebrain> There is no retry possible. There is a choice to make .. a new connection comes in, claiming to be the same server
15:59:56 <truebrain> One of the two have to give
16:00:21 <locosage> server can retry
16:08:58 <locosage>
16:23:09 *** isosys has joined #openttd
16:23:09 <isosys> I mean, my plan is to do nothing and let the server restart once finished. It should be enough to fix.
16:23:09 <isosys> But you can't really tell your "clients" that or they get mad at you. Costumer service 101 πŸ˜…
16:25:21 <locosage> πŸ˜†
16:25:48 <locosage> you can run `set server_game_type public` on each server to get them backk
16:27:26 <isosys> Thanks for the tip! Let me try
16:27:50 *** iSoSySt has joined #openttd
16:29:57 <isosys> Great! That did the trick. Thanks! πŸ‘
16:30:25 *** iSoSySt has quit IRC ()
16:52:16 *** Wormnest has joined #openttd
17:23:19 <andythenorth> truebrain: why are you using a complicated cloud stack to test perf? Just use my MBP zzzzzzzzz
18:00:56 <_zephyris> Is there a place to download old OpenTTD versions? The old format links don't work 😦
18:10:22 <talltyler> Try
18:15:28 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler opened pull request #11197: Codechange: Use correct type for max timetable start years constant
18:19:55 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler commented on pull request #11196: Fix: Max road speed is doubled in buy menu dropdowns
18:24:38 <truebrain>
18:24:38 <truebrain> It is crawling on ... you start to see how accurate the performance measurements are πŸ˜„
18:38:48 <DorpsGek> [OpenTTD/OpenTTD] eints-sync[bot] pushed 1 commits to master
18:38:49 <DorpsGek> - Update: Translations from eints (by translators)
18:45:41 <Bouke> truebrain: The graph looks great, but what is it measuring? The title reads β€œcompile instructions”, but what does compilation have to do with this? Is it counting the number of instructions it takes to simulate a certain in-game period?
18:46:06 <truebrain> where do you read "compile" instructions?
18:47:09 <truebrain> owh, in that note above there. That refers to how the binary was created .. compile as per instructions for linux-generic
18:47:21 <truebrain> you have to compile the binary some time
18:47:26 <truebrain> so I am a bit confused by your question?
18:47:53 <truebrain> ah, I think you didn't read the header of each graph. That states what the graph represents
18:50:17 <Bouke> truebrain: I did read the caption and I don’t understand what is meant by it, hence the question. What is meant by β€œaverage CPU of five runs”?
18:50:33 <truebrain> The title reads "Total CPU instructions (lower is better)"
18:50:59 <truebrain> so the graph represents how many instructions the CPU had to process for that savegame
18:51:16 <truebrain> not really sure how to explain this .. hmm .. let me think
18:51:28 <truebrain> just a check question: CPU instructions you are familour with? (before I make stupid assumptions)
18:51:36 <Bouke> CPU instructions consumed to do _what_?
18:52:09 <truebrain> sometimes it is funny to have a mismatch in communication, as I am scratching my head what your question is, and you are scratching your head why I don't understand the question πŸ˜„
18:52:55 <truebrain> what I do to create these graphs, and they are relative to themselves, not to eachother btw, is measure how many CPU instructions it took to run the savegame for a predefined amount of ticks
18:53:07 <truebrain> to make sure we are not measuring noise, I run the same savegame 5 times
18:53:08 <truebrain> take the average
18:53:11 <truebrain> and plot that
18:53:13 <truebrain> for each savegame
18:53:50 <truebrain> a CPU instruction is just what the CPU reports back how many instructions it spend on the job asked, in this case, running OpenTTD for a predefined amount of ticks
18:54:05 <truebrain> worth nothing: not predefined amount of time, as time is pointless for this kind of performance test
18:54:08 <truebrain> does that help?
18:54:10 <Bouke> Right! The number of instructions it takes to simulate a certain number of ticks.
18:54:27 <truebrain> You really confused me with "Compile instructions" πŸ˜„
18:54:43 <Bouke> Yes, it confused me as well!
18:54:54 <truebrain> but different instructions
18:54:59 <truebrain> one as in "manual"
18:55:03 <truebrain> other as in "work"
18:55:04 <truebrain> πŸ˜„
18:55:07 <truebrain> I will reword πŸ˜‰
18:55:13 <Bouke> It all makes sense now
18:55:56 <truebrain> in this case btw, westbourne runs for 300 ticks, the rest 3000 ticks
18:56:04 <truebrain> running westbourne for 3000 ticks is impractical πŸ˜„
18:56:32 <truebrain>
18:56:36 <truebrain> hope that helps
18:58:14 <Bouke> truebrain: Interesting idea, maybe it could even run per PR to prevent regressions from being merged, and possible improvements be verified.
18:58:34 <truebrain> it could; but as you can see it changes little, so that is a lot of work for very little use
18:58:39 <truebrain> but by all means, knock yourself out πŸ˜„
19:00:00 <truebrain> ha, I was validating if the "switch to local" of the game-coordinator was working as intended, which made me discover a nasty bug that has been bugging me for a while πŸ˜„
19:02:14 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain opened pull request #208: Fix: don't remove a server twice when a connection is quick enough
19:02:27 <Bouke> Well seeing that performance is only ever increasing, maybe not as necessary right now.
19:02:54 <talltyler> Care to give #11197 an approval so I can include it in my rebase? πŸ™‚
19:03:15 <truebrain> Bouke: that is how I was balancing it, yes. But yeah, having a tool like I wrote really helps to confirm we are not regressing .. I might integrate it with the nightly πŸ™‚
19:03:34 <talltyler> So many little chunks of weird tech debt to find! πŸ˜„
19:03:37 <truebrain> talltyler: if you look into my TileIndex PR πŸ˜›
19:03:48 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #11197: Codechange: Use correct type for max timetable start years constant
19:03:51 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler merged pull request #11197: Codechange: Use correct type for max timetable start years constant
19:04:49 <talltyler> Thanks! Yes, I'll look into your PR this week...too much focus on timers today πŸ™‚
19:05:04 <truebrain> pffffffff πŸ˜›
19:05:08 <truebrain> (it is okay)
19:05:09 <Bouke> truebrain: Does it include actual rendering? So catching regressions in video rendering regressions?
19:05:19 <truebrain> yes and no, kinda
19:05:22 <truebrain> it does most of the work
19:05:34 <truebrain> just not the part that normally runs in the drawing-thread
19:05:38 <truebrain> which is shocking little
19:06:01 <truebrain> so viewport is rendered and everything
19:06:13 <truebrain> (also why you need OpenGFX, for example, to run dedicated server)
19:07:03 <Bouke> Ok great, so I can use that to measure the impact of that JGR patch (dirty window handling).
19:07:33 <truebrain> some video drivers also handle dirty-rect themselves
19:07:51 <truebrain> so if you want good benchmarks, you have to run it with a GUI πŸ™‚ For which we have little infra
19:08:07 <truebrain> no videodriver handles `ticks` argument .. I have been willing to fix that for a while, but never did πŸ˜„
19:08:20 <truebrain> NOTE: I am going to disconnect all multiplayer servers again in a bit; they keep running out of memory, so assigning a bit more.
19:08:38 <DorpsGek> [OpenTTD/game-coordinator] TrueBrain merged pull request #208: Fix: don't remove a server twice when a connection is quick enough
19:08:46 <truebrain> and deploying that fix while at it, so two disconnects!
19:09:08 <Bouke> truebrain: So that’s maybe a good thing to also address with this patch.
19:09:21 <truebrain> no, not with the same PR please πŸ˜„
19:11:38 <truebrain> sorry 30 people connected via TURN 😦 😦 I am so sorry ... but I forgot the TURN service will also restart 😦
19:12:30 <truebrain> that was a shitty thing to do on my part 😦 Owh well ..
19:17:04 <xarothbrook> I demand a refund!
19:22:59 *** gelignite has quit IRC (Quit: Stay safe!)
19:37:11 <truebrain> they reconnected nearly instantly; so all is good I guess πŸ™‚
20:13:05 <truebrain> between 2023-02-26 and 2023-03-03 we made a change that made westbourne very happy memory wise ... the gap is narrowing πŸ˜„
20:13:58 <truebrain> and, and the 1.4 game crashed for a few weeks .. that is why the memory and cpu was so low πŸ˜›
20:14:14 <andythenorth> Pfff….I made a GS that used about 1GB or so :p
20:14:31 <andythenorth> β€œMemory saving”
20:14:57 <andythenorth> I think that was in JGRPP though
20:15:17 <truebrain> right, now all I need is a bit more diversity in savegames .. that is always a bit tricky. I have a big archive somewhere on my disk, I should look through it πŸ˜›
20:16:43 <truebrain> and we really need a argument that downloads all needed NewGRFs etc from BaNaNaS via the command line
20:27:57 <talltyler> Bouke: If you're looking at upstreaming drawing performance improvements from JGRPP, you might look into also upstreaming two commits attached to this bug report. There's some missing pieces for a clean merge, but if you're up to speed on that stuff maybe you'll find it straightforward to make it compatible. πŸ™‚
20:27:57 <talltyler>
20:34:46 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:14:39 <truebrain>
21:14:39 <truebrain> Another sweep .. even more detail .. you can see except for 1.4 it is very stable .. so yeah πŸ˜„
21:14:54 <truebrain> now I need to figure out how I can get this in percentages ...
21:25:04 <DorpsGek> [OpenTTD/OpenTTD] 2TallTyler updated pull request #10700: Codechange: Split dates and timers into Economy and Calendar time
21:25:50 <truebrain> pff, anything more than basic graphs is really hard in grafana 😦
21:26:03 <truebrain> like: give in % against highest in serie
21:26:04 <truebrain> I have no clue πŸ˜›
21:29:08 <talltyler> I guess I can't use templates for `static constexpr TimerGameCalendar::Year ORIGINAL_BASE_YEAR = 1920;` πŸ€”
21:29:29 <truebrain> bit of a weird question .. why do you want a template for a const .. πŸ˜›
21:29:33 <truebrain> that question makes little sense πŸ˜„
21:30:23 <talltyler> Well, I need to use the original game years for both calendar and economy time
21:30:32 <talltyler> Mostly for saveload, but also NewGRF and inflation
21:30:45 <talltyler> I don't really want to make it part of the timer class
21:30:48 <truebrain> so conceptually you are thinking wrong about this πŸ™‚
21:30:57 <talltyler> Undoubtedly πŸ˜›
21:30:57 <truebrain> the idea of splitting these timers, is that they live in their own world
21:31:07 <truebrain> so you don't even would want to "template" them
21:31:10 <truebrain> they have different meaning
21:31:18 <truebrain> like one is a car, the other is a tree
21:31:30 <truebrain> saying the tree has 10 horsepower makes no sense
21:31:39 <truebrain> it is that distinct that you have to think about these timers
21:31:55 <talltyler> Well, outside Real Time Mode economy time has to follow calendar time exactly, so they're not that different
21:32:08 <truebrain> they tick the same
21:32:14 <truebrain> but they might not
21:32:16 <truebrain> one could go faster
21:32:21 <truebrain> so they are totally different entities
21:32:28 <truebrain> and as result, you also need two different constants
21:32:37 <truebrain> it makes no sense to mix them or "template" them πŸ™‚
21:32:59 <talltyler> Fair enough, I guess I'll move everything inside the proper classes then
21:33:03 <truebrain> (I go in this deep about this, as it is also why I mentioned the static_casts yesterday .. conceptually, it doesn't add up πŸ˜„ )
21:33:10 <talltyler> And get rid of `date_type.h` entirely
21:33:27 <truebrain> I think there is no other way around that, to keep the code sensible
21:33:34 <truebrain> that you either see an economy timer, or a calendar time
21:33:42 <talltyler> It will mean a lot of duplication, that is fine?
21:33:45 <truebrain> but never have to go: why is this calendar timer being added to this economy timer? πŸ˜„
21:33:56 <truebrain> you can always make a base class in top, to put the common stuff in there
21:34:12 <talltyler> Ah, didn't consider that πŸ˜„
21:34:17 <truebrain> OO yo!
21:34:57 <truebrain> (OO being Object Oriented πŸ˜› )
21:35:20 <talltyler> Uh oh, now I have to come up with a name of the parent class πŸ˜›
21:35:29 <truebrain> TimerGameCommon
21:35:29 <truebrain> there
21:35:36 <talltyler> πŸ™
21:35:47 <truebrain> now how do I normalize this graph in Grafana?
21:42:00 <truebrain> this is so weird ... sometimes Grafana transformations just don't do anything, or at least, it seems that way
21:51:47 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
21:57:54 <truebrain> had to check the network traffic to see the error .. lolz
22:10:25 <truebrain> owh, lol, you have Transformations in Grafana, with which you can modify data ... just .. you can't chain them depending on the query .. you can only chain them on all data .. that is not useful at all
22:19:21 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
22:21:47 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:25:13 *** twb has joined #openttd
22:25:39 <twb> Is there an easy way to make the openttd start screen a little less busy? i.e. remove the sort of self-playing backdrop
22:30:28 *** bornacorn has quit IRC (Quit: User went offline on Discord a while ago)
22:31:19 <truebrain> remove `opntitle.dat` from the installation folder would do it
22:31:30 <twb> Thanks, I'll try that
22:31:31 <_glx_> you'll get full water
22:32:25 <truebrain>
22:32:25 <truebrain> did find a performance regression with 3 savegames .. this is percentage based from its max, and you can see a clear uptime around 2023-06
22:32:59 <truebrain>
22:32:59 <truebrain> this doesn't happen for other savegames
22:34:14 <peter1138> Strange
22:35:06 <truebrain> a bit like: all games got a bit quicker, but some suffer just a tiny bit πŸ˜„
22:35:13 <truebrain> (realise that these games are very very small games)
22:36:40 <truebrain> during that time, the Utf8Encoder got C++ified
22:36:49 <truebrain> so it might be possible those games use that, relatively, more than the other games
22:37:01 <truebrain> for example because there are a ton of labels .. or because there are very few vehicles πŸ˜›
22:37:09 <truebrain> not sure it is actually useful, going this deep in these stats πŸ˜„
22:38:47 *** _aD has quit IRC (Ping timeout: 480 seconds)
22:43:34 <locosage> it may just be something irrelevant
22:43:40 <locosage> pick games that test actual limits
22:43:51 <locosage> lots of vehicles, lots of water, trees, houses, etc
22:44:07 <locosage> some newgrf code
22:44:21 <truebrain>
22:44:21 <truebrain> okay, this is also a nice visualisation. The most left value is always 100%, the rest is based on that
22:44:30 <truebrain> you can see the different "type" of games πŸ™‚
22:46:24 <truebrain>
22:46:24 <truebrain> even better πŸ˜„
22:46:39 <locosage> are all the slow ones < 1.4 ?
22:46:44 <locosage> do you measure saveload?
22:46:51 <locosage> there may be some slow conversion
22:47:25 <locosage> well "slow", ones that got slower
22:47:52 <truebrain>
22:47:52 <truebrain> memory is a bit more messy πŸ˜„
22:49:15 <truebrain> those that go up in CPU also have another memory pattern than the rest πŸ™‚
22:49:32 <truebrain> hihi, okay, this is fun πŸ˜› Not useful to delve that deep into stuff, but fun πŸ˜„
22:51:10 <truebrain> okay, it completely has to do with the simplicity of those 4 titles games πŸ™‚
22:51:53 <peter1138> SDL2: using driver 'KMSDRM'
22:52:01 <truebrain> wtf? Lol?
22:52:08 <truebrain> KMS .. DRM .. what does that have to do with video? πŸ˜›
22:52:40 <peter1138> direct rendering manager / kernel mode setting
22:52:45 <truebrain> hahaha
22:52:52 <truebrain> okay, I did not expcet that πŸ˜›
22:52:53 <peter1138> So... it's talking direct without X11 or Wayland.
22:53:00 <truebrain> Key Management System ...
22:53:02 <truebrain> πŸ˜„
22:53:17 <peter1138> Which is interesting, but not useful because that machine is in a closet upstairs πŸ˜„
22:53:22 <truebrain> lolz
22:53:43 *** Alkel_U3 has quit IRC (Quit: maintenance)
22:54:05 <peter1138> After I wiped my Debian WSL install I used vscode-server on a real Debian system πŸ™‚
22:54:25 *** Alkel_U3 has joined #openttd
22:54:29 <truebrain> okay, from 2021-08 up till like 2022-10 the CPU was slowly going up for all games .. and then it started to go down.
22:54:53 <peter1138> Also, it's even using OpenGL -- OpenGL driver: Intel - Mesa Intel(R) HD Graphics 4600 (HSW GT2) (4.6 (Compatibility Profile) Mesa 22.3.6)
22:55:02 <truebrain> amazing!
22:55:05 <truebrain> VNC to get a display? πŸ˜›
22:56:43 <truebrain>
22:56:43 <truebrain> I like this graph ... shows our memory usage for all titlescreens is now nearly equal πŸ˜„
22:57:34 <truebrain> owh well, time to get some zzzz .. the server will hopefully finish compiling all binaries tomorrow πŸ™‚
22:59:51 <peter1138> Hmm, long HDMI cable? πŸ˜„
23:00:34 <peter1138> USB HDMI capture card, sending captured video over the network...
23:22:22 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:51:05 *** tokai|noir has joined #openttd
23:51:05 *** ChanServ sets mode: +v tokai|noir
23:57:43 *** tokai has quit IRC (Ping timeout: 480 seconds)