IRC logs for #openttd on OFTC at 2019-09-14
⏴ go to previous day
00:46:47 *** Smedles has joined #openttd
04:06:34 *** Thedarkb-X40 has joined #openttd
07:02:48 *** WormnestAndroid has quit IRC
07:22:07 *** ChanServ sets mode: +v tokai
07:29:27 *** WormnestAndroid has joined #openttd
08:02:48 *** WormnestAndroid has quit IRC
08:18:19 *** Progman has joined #openttd
08:23:09 *** WormnestAndroid has joined #openttd
08:49:25 *** sla_ro|master has joined #openttd
09:19:46 *** andythenorth has joined #openttd
10:57:29 <Samu> achilles heel of my AI is still the pathfinder speed :|
11:07:22 <Samu> experimenting a self-regulated min_distance and max_distance for routes but...
11:07:42 <Samu> the pathfinder makes it unviable
11:09:15 *** andythenorth is now known as Guest2529
11:09:16 <Samu> @calc (DAYS_IN_TRANSIT * 2 * 3 * 74 * AIEngine.GetMaxSpeed(bestengineinfo[0]) / 4) / (192 * 16)
11:09:16 <DorpsGek> Samu: Error: There's really no reason why you should have underscores or brackets in your mathematical expression. Please remove them.
11:09:45 <Samu> @calc (45 * 2 * 3 * 74 * 127 / 4) / (192 * 16)
11:13:11 <Samu> @calc (45 * 2 * 3 * 74 * 56 / 4) / (192 * 16)
11:13:47 <Samu> @calc ((45 * 2 * 3 * 74 * 56 / 4) / (192 * 16) * 2) / 3
11:13:56 <nielsm> ever considered setting this up in excel or writing a quick program in python or something?
11:14:55 <Samu> oh, i got excel 2007 installed here
11:14:57 <nielsm> excel even has some what-if analysis features
11:15:09 <nielsm> (maybe not in older versions)
11:16:52 <Samu> terron has the best ratio between number of buses and profit
11:17:16 <Samu> or probably nonocab is better
11:17:50 *** andythenorth has joined #openttd
11:26:43 <Samu> railwai has the opposite approach, it creates as many short routes as possible
11:27:05 <Samu> but quickly reaches max veh limit
11:27:32 <Samu> profits are good, pathfinding is quick, but ...
11:28:04 <Samu> that limit makes it look weak later on
11:42:16 <Samu> nah, nonocab has a way different methodologie
11:43:54 *** gelignite has joined #openttd
11:44:10 <Samu> sometimes his routes are short, sometimes they are very distant, like 370 tiles away
11:47:25 <Samu> I better make this DAYS_IN_TRANSIT a config setting, saves me a lot of trouble
12:11:05 <Samu> @calc ((150* 2 * 3 * 74 * 56 / 4) / (192 * 16) * 2) / 3
12:15:39 *** HerzogDeXtEr has joined #openttd
12:15:42 <Samu> between 202 and 303 tiles, not as long as some of nonocab routes
12:19:44 <Samu> my poor pathfinder is so slow, it's unsuitable for this
12:20:03 <Samu> I don't know what to do about it
12:21:51 <thexa4> there are preprocessing steps you can do to speed pathfinding up
12:23:38 <thexa4> (depending on the amount of pathfinding you're doing it might be slightly more efficient)
12:25:34 <Samu> i'm not sure what steps are those
12:29:53 <thexa4> Jump point search would probably help as well (easier than hierarchical pathfinding)
12:51:35 *** andythenorth has joined #openttd
13:07:16 *** supermop_work_ has joined #openttd
13:22:12 * LordAro sitting in a park drinking coffee
13:23:14 <andythenorth> there are worse things
13:28:25 <andythenorth> got a changelog and stuff?
13:32:23 <LordAro> there's a draft PR i need to fixup
13:32:41 <LordAro> pretty sure everyone is happy with 1.9.3-rc1?
13:34:51 <andythenorth> did we do a forum post?
13:37:50 <LordAro> TB will be able to tell us how many have downloaded it
13:38:13 <LordAro> but i think the rc1 was to satisfy niels that everything was working more than anything else
13:38:23 <LordAro> the number of changes is pretty small
13:43:57 <LordAro> given it's mostly for osx related fixes, have you tried it? :p
13:44:29 <LordAro> try to connect to some server running it, assuming there is one
13:47:04 <Eddi|zuHause> maybe i should actually look at #7000 again this time...
13:51:48 <TrueBrain> LordAro: just remember, we can only have 1 RC/beta on the website. So always release 1.9.3 before you make a new beta/rc :)
13:53:46 <LordAro> what would happen if there were 2? just display the latest? or the first one alphabetically?
13:53:56 <TrueBrain> the "latest" in terms of versions
13:54:44 *** WormnestAndroid has quit IRC
13:54:57 *** WormnestAndroid has joined #openttd
13:55:50 <Eddi|zuHause> btw, we should probably merge one of the fixes for #7737 :)
14:02:24 <andythenorth> LordAro: I haven't tried it no :P
14:02:29 <andythenorth> I'll download now
14:05:40 <andythenorth> no 1.9.3-rc1 servers
14:06:59 <LordAro> but you can download the list of servers to see that there are none!
14:11:03 <andythenorth> I get a long list of servers
14:26:51 <peter1139> Might be lunch time.
14:26:57 <peter1139> Oh crap, we had that yesterday.
14:27:09 <andythenorth> bacon, it's Saturday
14:27:45 <peter1139> I don't have any. So I'll probably just have a, um, salad...
14:29:57 <andythenorth> salad is always valid, unless hungover
14:32:32 <DorpsGek_III> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Some NewGRF variables concerning railtypes https://git.io/fhI7h
14:32:47 <Eddi|zuHause> some attempt at a rebase...
14:32:56 <Eddi|zuHause> do i have to redo stuff for NRT?
14:36:40 <andythenorth> I owe you a test case for that PR also
14:36:52 <andythenorth> remind me later? ...I have to go out
14:37:29 <Eddi|zuHause> hm, i should check for NULL vs nullptr
14:43:04 <DorpsGek_III> [OpenTTD/OpenTTD] Eddi-z updated pull request #7000: Some NewGRF variables concerning railtypes https://git.io/fhI7h
14:45:45 <michi_cc> This is the other way around, i.e. instead of trying to force everything to sRGB, we just try to conform ourself to whatever the window is created with.
14:46:20 <michi_cc> As apparently according to the form thread, the current fix is even worse for at least one person than before.
14:46:23 <thexa4> I can test on mac if needed
14:46:39 <michi_cc> Patch applies to master.
14:47:38 <michi_cc> If you can test it, great (note: not tested at all, might not compile due to type or so).
14:47:51 <michi_cc> Otherwise, I'm AFK for about an hour.
14:56:57 <Eddi|zuHause> so, i'm thinking mimicing #7000 for NRT should be a separate PR, this is just railtype stuff...
15:00:36 <Eddi|zuHause> now, the big open question was speed limit, giving the raw railtype speedlimit is probably not very useful
15:01:50 <thexa4> `src/video/cocoa/wnd_quartz.mm:109:24: warning: unused function 'QZ_GetCorrectColorSpace' [-Wunused-function]`
15:03:05 <thexa4> looks ok, getting normal framerates but I haven't encountered the described issue before. I think I have a different display profile (macbook air)
15:04:54 <thexa4> hm, performance when running with fast forward is lower
15:05:38 <thexa4> compared with 1.9.2 (not master)
15:07:40 <thexa4> master is equally slow
15:07:58 <thexa4> so don't think that's due to this change
15:08:14 <thexa4> likely due to the way I built it
15:18:05 <supermop_Home_> what're you working on eddi? something with railtypes?
15:19:03 <Eddi|zuHause> supermop_Home_: for things like hybrid electric/diesel engines
15:35:37 *** snail_UES_ has joined #openttd
15:44:58 *** andythenorth has joined #openttd
15:58:48 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 opened pull request #7740: Fix: inconsistent description for 32bpp-sse4-anim blitter https://git.io/JeYr6
16:08:51 <michi_cc> Applies to current master. Might or might not do something.
16:14:42 <andythenorth> bunch of findings
16:14:45 <michi_cc> thexa4: Thanks for testing. Difference between master and 1.9.2 could very well be due to some new feature in master.
16:15:08 <andythenorth> in no particular order:
16:15:20 <andythenorth> - color is now correctly wrong for me
16:15:40 <andythenorth> - FPS with full animation on is ~same as the 7644 patch compiled locally
16:16:07 <andythenorth> - 1.9.3-RC1 is noticeably lower FPS than 7644, or the michi patch
16:16:24 <andythenorth> - all are much faster FPS than 1.9.1 with full animation on
16:17:00 <andythenorth> - all FPS is a bit unreliable over short runs, dfifferent maps etc, but we're aware of this
16:17:06 <michi_cc> My patch is basically the reverse of the 7644 patch that was commited, i.e. it just gets whatever the OS set for the window colour space and tries to use that internally for everything.
16:17:09 <glx> maybe try the patch on 1.9 branch
16:17:27 <michi_cc> Whereas 7644 tries to use sRGB internally and also for the OS window.
16:17:51 <andythenorth> I wonder if I fat fingered that
16:18:01 <andythenorth> I thought it was supposed to default to P3
16:20:42 <andythenorth> nope I didn't fat finger it, sRGB was intended it seems
16:30:37 *** Wormnest_ has joined #openttd
16:33:33 <DorpsGek_III> [OpenTTD/OpenTTD] michicc opened pull request #7741: Fix #7644: [OSX] Try to use system colour space to avoid video output… https://git.io/JeYoJ
16:37:13 <glx> I think it needs to be tested on 1.9 branch too
16:39:34 <glx> it's better if we can get feedback without releasing
16:42:48 * andythenorth trying to figure out where 1.9 branch is to checkout
16:43:36 <DorpsGek_III> [OpenTTD/OpenTTD] michicc updated pull request #7741: Fix #7644: [OSX] Try to use system colour space to avoid video output… https://git.io/JeYoJ
16:45:09 <andythenorth> if I just checkout upstream/release/1.9 is that good enough?
16:48:56 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh approved pull request #7740: Fix: inconsistent description for 32bpp-sse4-anim blitter https://git.io/JeYow
16:50:09 <andythenorth> patched applied to 1.9 branch
16:50:14 <andythenorth> - colours are correctly wrong
16:50:28 <andythenorth> - FPS is same as with patch applied to master
16:51:11 <andythenorth> apparently mine are wrong
16:51:15 <andythenorth> but I am used to them
16:51:21 <andythenorth> so I'd rather they didn't change
16:52:17 <andythenorth> more accurately, colours are same as 1.9.1
16:52:37 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 merged pull request #7740: Fix: inconsistent description for 32bpp-sse4-anim blitter https://git.io/JeYr6
16:55:15 <DorpsGek_III> [OpenTTD/OpenTTD] michicc commented on pull request #7717: Codechange: [OSX] Use std::unique_ptr with a custom deleter to simply… https://git.io/JeYo6
17:01:46 <michi_cc> andythenorth: Pure interest, can you add "NSLog(@"Window color space: %@\n", color_space);" after line wnd_quartz.mm:576 (count from the PR, not the gist).
17:02:09 <michi_cc> Would be interested to see what colour space your system is really using.
17:06:51 <andythenorth> 2019-09-14 16:06:07.725 openttd[35308:368628] Window color space: <CGColorSpace 0x7fdff9536d10> (kCGColorSpaceICCBased; kCGColorSpaceModelRGB; Color LCD)
17:07:31 <michi_cc> Okay, ICC based says basically nothing :)
17:07:45 <michi_cc> Even sRGB can be ICC based.
17:08:08 <andythenorth> switching to the Apple provided "Display P3"
17:08:22 <andythenorth> 2019-09-14 16:08:15.612 openttd[35528:371688] Window color space: <CGColorSpace 0x7f8214c376e0> (kCGColorSpaceICCBased; kCGColorSpaceModelRGB; Display P3)
17:08:50 <andythenorth> perceptually the same, my 'calibrated' profile is based on Display P3, and Apple nerfed all the user 'calibration' options a few years ago
17:09:02 <andythenorth> 'calibrated' now seems to mean I can give it a name, and nothing else
17:09:11 <andythenorth> used to be a full set of calibration controls
17:09:33 <andythenorth> oh I can mess with the white point I see
17:09:35 <Eddi|zuHause> it's apple, why would they give you options?
17:09:49 <andythenorth> there's two Apples
17:10:02 <andythenorth> there's the one that gives you no options, because they know best
17:10:09 <andythenorth> and there's another one, less known about
17:10:18 <andythenorth> where they give you what you need to get stuff done
17:10:26 <andythenorth> it's a dice roll which one you get
17:13:18 <andythenorth> hmm actually it's more fragmented than that, but I digress
17:13:35 <andythenorth> I missed all the permutations where they give you stuff, but it just doesn't work
17:14:16 <michi_cc> Also the permutation where they give you what everybody else gives you, just 20% more expensive :)
17:14:58 <andythenorth> but the corner radii are lovely
17:15:10 <andythenorth> I am still waiting for my adapters
17:15:32 <andythenorth> what are they for again?
17:15:42 <michi_cc> Sometimes though, it's just better. I've got an iPad as well, and Android tablets are just meh.
17:16:19 <andythenorth> I am surprised how bad iPads are
17:16:31 <glx> for android it really depends on the device
17:16:38 <Eddi|zuHause> makes you question how much worse android devices are
17:17:01 <andythenorth> they're fine no?
17:17:04 <michi_cc> Android tablets mostly suck because Android apps suck on tablets. It's usually not the hardware that's bad.
17:17:24 <andythenorth> whenever I pick up someone else's Android phone I find it much easier to use than iOS
17:17:32 <glx> oh I've seen android tablet with very bad tactile screen
17:17:43 <andythenorth> oh, yeah the hardware interaction sucks
17:17:45 <milek7> i have lineageos on my phone
17:17:47 <andythenorth> but the OS seems fine
17:17:47 <milek7> it gets slower and slower with every update
17:18:15 <andythenorth> like all the details of touch performance, and how nice the device is to hold and stuff is all wrong for Android devices
17:18:22 <andythenorth> but iOS is horrible
17:18:41 <andythenorth> keyboard is hideous
17:18:57 <glx> why would you want a file system ?
17:19:07 <michi_cc> iPadOS will fix all that (so they say).
17:19:08 <glx> you are not supposed to touch taht
17:19:16 <andythenorth> how does anything get done?
17:19:36 <milek7> i'm considering flashing back old 4.4, it was way faster
17:19:52 <andythenorth> I want to save a picture of a train, the browser adds it to Photos
17:20:04 <andythenorth> then I have to go to Photos and copy it to the stupid fake Files app
17:20:16 <andythenorth> then I open it from there in Pixelmator, which makes it's own copy
17:21:14 <andythenorth> also there's no shell
17:21:35 <glx> of course there's no shell
17:21:43 <glx> it can harm your device :)
17:21:44 <milek7> windows phone was similiar, no wonder it died
17:22:09 <andythenorth> it's dumb, there is a full *nix filesystem hiding under it
17:22:43 <andythenorth> but I suspect that app just plants malware
17:23:56 <andythenorth> but when your kids lose Minecraft PE savegames that they spent days on, filesystem starts to matter :P
17:24:10 <glx> anyway knowing how you can mess stuff, it seems ok to not have filesystem exploration and shell ;)
17:25:04 <andythenorth> I hate that it takes the data away
17:25:19 <andythenorth> on the other hand, it teaches that everything is ephemeral
17:25:25 <andythenorth> and we shouldn't care about Minecraft worlds
17:25:55 <glx> just use java edition on a computer then
17:26:14 <andythenorth> my kids don't get computers
17:26:18 <andythenorth> they could mess with stuff
17:26:24 <Eddi|zuHause> or java edition on the ipad :p
17:26:27 <andythenorth> they'd have a shell, and filesystem etc
17:26:37 <milek7> they won't learn otherwise ;P
17:27:04 <Eddi|zuHause> you know you can hide these things?
17:27:14 <andythenorth> there are parental controls
17:27:50 <glx> you can't sync minecraft PE saves on cloud ?
17:28:29 <andythenorth> it's a really common google search, 'my child lost their Minecraft game on iOS'
17:28:43 <andythenorth> on Android you just browse the files and it's all fine, right?
17:29:07 <glx> some stuff is hidden and require root
17:29:33 <glx> but usually accessing app data is ok
17:30:50 <LordAro> i know that there's only one GPS point within that section though (it's under some trees)
17:36:37 <DorpsGek_III> [OpenTTD/OpenTTD] michicc closed issue #7737: Crash of OpenTTD master gf81cb0a90d possibly after trying to sell the default human user https://git.io/Jemyd
17:37:04 *** snail_UES_ has joined #openttd
17:42:23 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #7738: Fix #7737: Add missing return in MoveToNextNewsItem https://git.io/JeYKC
17:42:23 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro closed pull request #7738: Fix #7737: Add missing return in MoveToNextNewsItem https://git.io/JeYUf
17:45:25 <LordAro> so last night i found a thing that analyses the code of a git repo and works out how much of it sticks around for however long - this was the result of running it over OTTD: https://imgur.com/b9HXn5t
17:55:45 <nielsm> michi_cc: I just took a look over your opengl branch, it looks pretty good but requires a long rebasing (past both 'nullptr' and 'override' global changes), I'd really like to get it back to development, though IMO it should only support OGL 3.0+
17:57:35 <DorpsGek_III> [OpenTTD/OpenTTD] michicc updated pull request #7741: Fix #7644: [OSX] Try to use system colour space to avoid video output… https://git.io/JeYoJ
17:58:03 <michi_cc> andythenorth: Can you see if you can crash that updated PR by changing colour profiles when OTTD is running?
17:58:18 <andythenorth> already tried that
17:58:48 <michi_cc> And anybody with an OSX multi-monitor setup to check moving between screen.
17:59:00 <michi_cc> andythenorth: I mean with the additional commit from just right now :)
18:01:00 <andythenorth> I checked multi-monitor yesterday with the old PR, but that's no good :)
18:04:58 <andythenorth> michi_cc: no crash when changing colour space
18:05:17 <andythenorth> there's an interesting latency while it recalculates the colour correction, which other apps don't seem to do
18:07:38 <andythenorth> no crash when dragging to iPad screen
18:08:31 <andythenorth> no crash changing colour profile on iPad screen
18:08:45 <andythenorth> the resolution on the iPad is insane
18:09:27 <andythenorth> :o 2048*1536 on a 7.9" screen
18:09:43 <andythenorth> my computer is only 1440*900
18:10:47 <Eddi|zuHause> i only have a 1920x1080 screen that i got when my previous screen broke, and that one i recently repaired with 1680x1050
18:11:10 * andythenorth puts OpenTTD to 4x UI zoom
18:13:12 <andythenorth> michi_cc: all seems fine using iPad as external display
18:14:06 <Eddi|zuHause> andythenorth: and if you run it natively on the ipad? :)
18:14:25 <andythenorth> I haven't bothered figuring out how to compile and sideload it
18:15:12 <Eddi|zuHause> that's about what i expected :p
18:18:37 <michi_cc> andythenorth: The commit is dumb and just recreates the backing store like when you resize the window.
18:21:02 <michi_cc> Doing it smarter is effort, and how often do you drag the windows between screens anyway.
18:22:45 <andythenorth> in my case never :P
18:22:57 <Eddi|zuHause> maybe someone has a dual screen setup that requires shifting the window back and forth between each refresh cycle? :p
18:26:51 <andythenorth> there's an XKCD for this
18:36:20 *** oskari89 has joined #openttd
18:59:23 <michi_cc> nielsm: I might find some time to rebase the OpenGL branch this weekend. The branch already is OGL 3.2+, I just chose to code it to use either 3.2 Core or whatever extensions are equivalent to it. Optional code paths are only for stuffer later than 3.2.
19:06:38 <milek7> LordAro: i guess that big step in 2019 is nullptr changes?
19:07:31 <LordAro> milek7: that and the container changes
19:29:01 *** Progman has joined #openttd
20:16:31 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #7728: Fix #7479: Don't close construction windows when changing client name https://git.io/JeYiC
20:16:42 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro merged pull request #7728: Fix #7479: Don't close construction windows when changing client name https://git.io/JeeIM
20:23:45 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #7086: Change #6173: Update SDL driver to use SDL 2.0 https://git.io/JeYia
20:23:54 <LordAro> anyone else want to do a final review + merge of that?
20:25:11 <DorpsGek_III> [OpenTTD/OpenTTD] Eddi-z commented on pull request #7000: Some NewGRF variables concerning railtypes https://git.io/JeYiV
20:25:23 <andythenorth> should I test it? o_O
20:26:06 <LordAro> would require a few configure flags on macos
20:26:16 *** gelignite has joined #openttd
20:29:41 <andythenorth> conflicts with master when I try a rebase though
20:29:56 <LordAro> hrm, github doesn't think so
20:30:23 <Eddi|zuHause> they might be using different rebase/merge strategies?
20:34:25 <LordAro> wouldn't have thought so...
20:35:23 <DorpsGek_III> [OpenTTD/OpenTTD] Eddi-z commented on issue #7711: #7612 causes multiple news messages to show at once and overlap. https://git.io/JeYiD
20:38:39 * andythenorth learning about petawatt lasers
20:42:46 <andythenorth> petawatt lasers are like actual rocket science
20:43:00 <andythenorth> whereas rocket science is mostly 'light the blunt end and stand away'
20:43:27 <andythenorth> except for big rockets, then it's mostly managing welding
20:43:34 <nielsm> rocket science is easy, rocket engineering is hard?
20:44:08 <nielsm> also you should play kerbal space program
20:44:36 <andythenorth> I tried KSP, but my rockets wouldn't stick together
20:58:37 <Eddi|zuHause> andythenorth: with KSP related problems there's only two possible answers: 1) more boosters, 2) more struts
20:59:12 <Eddi|zuHause> just alternate between the two, and your problem should resolve :p
20:59:53 <andythenorth> I never found the 'press go' button
21:00:30 <Eddi|zuHause> you go into space with the space bar.
21:27:24 *** andythenorth is now known as Guest2569
21:27:24 *** andythenorth has joined #openttd
23:03:50 <supermop_Home_> hmm short last mile vehicle delivery
23:03:58 <supermop_Home_> do it by trucks or a boat?
23:06:06 <supermop_Home_> i considered that
23:06:38 <supermop_Home_> its only about 15 tiles along the edge of a bay
23:18:26 <Eddi|zuHause> i wouldn't use boats if it's along the coast...
23:19:44 <Eddi|zuHause> trams are great at transporting large quantities in tight space, with trains you have all the signalling and stuff bloating up your network
23:35:26 <andythenorth> drawing intermodal container wagons is hard
23:35:33 <andythenorth> it's not just a flat shape
continue to next day ⏵
