IRC logs for #openttd on OFTC at 2024-02-20
⏴ go to previous day
00:05:57 <_glx_> the command is so broken
00:40:31 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
01:13:43 *** Wormnest has joined #openttd
02:33:26 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
02:46:52 *** Wormnest has joined #openttd
03:02:11 *** herms has quit IRC (Ping timeout: 480 seconds)
03:10:07 *** Wormnest has quit IRC (Quit: Leaving)
03:12:00 *** gnu_jj_ has quit IRC (Ping timeout: 480 seconds)
03:38:08 <ajmiles> This isn't a reasonable looking AnimBuffer I assume?
03:38:30 <ajmiles> It looks like I probably have an off-by-4 error somewhere, but I'm not sure where
03:39:35 <ajmiles> Although I would probably have expected data in just the top left corner not 4 repeated copies across the top if that were the case
04:03:11 *** debdog has quit IRC (Ping timeout: 480 seconds)
05:34:10 <ajmiles> In answer to mine own question, yes, the format was supposed to be R8 (not R8G8B8A8) and combined with the quadruple pitch, that was all it was
06:16:14 <ajmiles> _glados_: OK, couple of false starts. But the 12 backend works, minus the mouse cursor which I haven't done yet. Once I tidy it up I might start to give some thought to what could be done beyond that.
06:33:11 *** Ox7C5 has quit IRC (Ping timeout: 480 seconds)
06:59:11 *** nielsm has quit IRC (Ping timeout: 480 seconds)
07:04:38 <j_n> glx22viaGitHub: I only just woke back and got to school so I'm not able to respond on github
07:04:42 <j_n> might be able to respond here
07:09:05 <j_n> before button press, `font` outputs
07:09:05 <j_n> ```medium: "Arial" 10 aa ["" 0 noaa]
07:09:05 <j_n> small: "Arial" 6 aa ["" 0 noaa]
07:09:05 <j_n> large: "Arial" 18 aa ["" 0 noaa]
07:09:05 <j_n> mono: "sprite" 10 aa ["" 0 noaa]
07:09:11 <j_n> might be typos because i can't ctrl + c
07:10:32 <j_n> after button press it outputs
07:10:32 <j_n> ```medium: "sprite" 10 aa ["" 0 noaa]
07:10:32 <j_n> small: "sprite" 6 aa ["" 0 noaa]
07:10:32 <j_n> large: "sprite" 18 aa ["" 0 noaa]
07:10:32 <j_n> mono: "sprite" 10 aa ["" 0 noaa]```
07:10:48 <j_n> after second button press it does you know what
09:17:06 <merni> The font command is rather confusing
09:25:43 <_zephyris> Hmm, the font output isn't even correct, right?
09:28:10 <_zephyris> Ohhh, `font medium ""` gives different results depending on whether sprite font is enabled/disabled
09:28:52 <_zephyris> Wow, super cofusing.
09:29:19 <peter1138[d]> I guess I'm weird as I never found it confusing.
09:29:51 <peter1138[d]> (And it's not even my doing :))
09:29:54 <merni> Did you implement it :P
09:30:27 <merni> Didn't you have a patch for a font picker GUI?
09:30:55 <peter1138[d]> I did but I am not going to work on that any more.
09:32:12 <peter1138[d]> Does not make sense to have have our own custom font, and then a whole lot of code to select different fonts.
09:32:45 <merni> peter1138[d]: Isn't there already?
09:33:38 <LordAro> i don't follow that logic
09:33:55 <LordAro> we have our own custom base set, and a whole lot of code to select different graphics
09:34:22 <LordAro> like, a *whole lot* of code
09:34:37 <peter1138[d]> Well. I'm not stopping anyone else developing a font chooser.
09:34:40 <merni> Honestly I would appreciate if the usage of `""` and `"sprite"` was made clearer, as well as the way the font size you type interacts with interface scaling to give the font size that actually appears
09:35:16 <LordAro> (yes, you are, of course, free to not work on whatever you like :) )
09:42:15 <peter1138[d]> Nah we're just lazy devs not doing anything the community wants.
09:44:00 <peter1138[d]> Okay well *I* am.
09:44:29 <LordAro> reldred: you're just a lazy community not doing anything the devs wants
09:47:00 <peter1138[d]> Except Xarick of course.
09:47:25 <peter1138[d]> Xarick works on exactly what the community wants being part of the community.
09:48:43 <reldred> yes, Xarick is pure and free from sin in this
09:48:53 <reldred> i however am condemned to forever seek penance
09:49:06 <reldred> my punishment must be severe
09:54:10 <Eddi|zuHause> i'm just a lazy community not doing anything.
10:19:33 <_glados_> ajmiles: I believe my problem is being to much of a perfectionist with the code and trying to remove the cpu rendering at the same time as adding the vulkan code making my attempts a mess
10:23:02 <peter1138[d]> Commits are cheap 🙂
10:39:38 <j_n> > remove the cpu rendering
10:40:17 <peter1138[d]> Yeah, that comes later to be honest 🙂
10:42:06 <Eddi|zuHause> unless you get the vulkan rendering to run on legacy hardware, probably never :p
10:42:55 <peter1138[d]> I'm assuming "remove the cpu rendering" here means only for the vulkan blitter...
10:50:46 <reldred> oh dear, looks like broadcom are going to knife us in the back with our very much vmware based datacentre business model :cryingbloodbutstillcool:
10:58:26 <peter1138[d]> libvirtd is slightly okay, but not brilliant.
10:58:49 <peter1138[d]> I was going to try something like Proxmox (or maybe there was another similar solution that I've forgotten the name of) but haven't yet.
10:59:03 <reldred> we can probably pivot our horizon business to hp anyware since that appears to hook up to kvm as an alternative to esxi
10:59:47 <peter1138[d]> "Not going to work on this"
10:59:56 <peter1138[d]> > Rebases anyway
10:59:58 <reldred> question is; will the new owners just go 'fuck that risk/noise' and just kill the dc hosting business and roll our customers into their existing datacentre locations
11:00:33 <merni> peter1138[d]: Well, I expect more tabs in future when someone merges the other kinds of settings into one window :p
11:00:55 <peter1138[d]> I don't think that's useful.
11:01:39 <j_n> I often forget the UI was refactored because I usually play JGRPP
11:02:02 <reldred> j_n: i mean its in jgrpp as well
11:02:14 <xarick> how do I share a 70 MB savegame on tt.forums
11:02:30 <peter1138[d]> Upload it to your FTP server and provide a link.
11:02:30 <j_n> reldred: oh wait maybe I just didn't update
11:02:47 <reldred> we just don't have the social tab yet
11:04:54 <peter1138[d]> The social tab does upset my muscle memory as it moved the other tabs over...
11:05:10 <peter1138[d]> Not sure why the game options window should have muscle memory...
11:06:02 <merni> It's not useful to merge game otpions with settings?
11:07:22 <peter1138[d]> Where do you stop?
11:07:46 <peter1138[d]> Advanced Settings. AI Settings. GS Settings, NewGRF Settings, Transparency Settings...
11:08:49 <xarick> this marks the end of my glorious test
11:09:17 <xarick> RIP AAAHogEx, saveload fails
11:09:38 <peter1138[d]> Advanced settings should first be categories into game-play settings and 'interface' settings.
11:10:11 <peter1138[d]> e.g. "cargo distribution" vs "viewport scrolling"
11:10:32 <peter1138[d]> Then you have "Advanced Game Options"
11:10:44 <peter1138[d]> And "Gameplay Settings"
11:11:12 <peter1138[d]> Gameplay settings includes AI, GS, NewGRF, etc.
11:11:16 <reldred> throw it all into one of those html5 tag cloud clusters that were so popular and weight the settings in the tag cloud by their popularity in the survey 😛
11:11:31 <merni> peter1138[d]: Well, we already have an "Interface" category
11:11:45 <merni> But yes some thought should go into the organisation
11:12:01 <peter1138[d]> What about "Sound"
11:12:23 <peter1138[d]> Should that go under the "Sound" tab in game options? There's no room for it there...
11:12:58 <peter1138[d]> Or do we shove everything into the settings tree...
11:12:59 <merni> Well, in general if it's all going to be in one window it should be scrollable, otherwise you couldn't fit the crazy amount of settings there are
11:13:41 <reldred> just start removing settings
11:13:43 <xarick> Oh, my AI assumes MAIL exists and will crash if it doesn't
11:13:55 <merni> peter1138[d]: No real point to taking the settings tree wholesale and shoving it into a different window, IMO.
11:13:56 <xarick> what newgrf do i need to test the crash?
11:14:22 <merni> I think AXIS has no mail? Could be wrong
11:14:42 <peter1138[d]> Improved Town Industries
11:18:22 <peter1138[d]> xarick: one author fixed their AI on the same day 😉
11:20:46 <xarick> I need to look at my next version, what happens
11:25:27 <xarick> dang it, ima need to upload version 21 😦
11:34:30 <xarick> is it safe to assume CC_PASSENGERS exist?
11:43:41 <xarick> does this look right to you?
11:44:08 <xarick> how is waste CC_COVERED and I'm using a truck that doesn't cover it?
11:44:27 <xarick> or is something I'm missing?
11:45:46 <xarick> Factories produce passengers 🙂
12:14:01 <merni> xarick: Yes, the game knows nothing about whether the graphics show a roof or not
12:14:36 <merni> Though you are probably creating a big stink by carrying waste in an uncovered vehicle :p
12:15:24 <peter1138[d]> If you can avoid assuming any particular cargo exists, that would be good 🙂
12:16:28 <peter1138[d]> Check for the classes and if they don't return anything do something else.
12:17:24 <peter1138[d]> Going by tile acceptance/production is probably most correct but likely tricky for towns, I dunno.
13:17:25 <xarick> fixing this will not be easy... the whole AI was coded on the assumption towns can produce and accept passengers and mail
13:19:11 <xarick> maybe I can fix something
13:20:56 <xarick> is it possible to know what cargo the town produces and accepts?
13:21:23 <xarick> or is acceptance dependant on houses?
13:22:31 <xarick> meh, I'll figure out something...
13:23:10 <peter1138[d]> Step 1 is to not deliberately assert when the cargo type isn't found 🙂
13:25:12 <xarick> it shouldn't even start planning if there's no CC_MAIL,
13:26:27 <xarick> so it was already wrong all the way until there
13:52:57 <xarick> wondering if cargoid 255 will be sufficient
14:01:31 <xarick> someone changed the error returned?
14:01:47 <xarick> I was expecting ERR_ALREADY_BUILT
14:07:14 <peter1138[d]> 12133 does not crash for me. Hmm.
14:09:13 <xarick> changing error return messages are very sensitive for AIs...
14:18:42 <_glx_> _current_company is 0 when the AI dies
14:21:06 <_glx_> and the AI dies while loading the .nut
14:23:29 <_glx_> it's a hidden tinyurl one
14:23:46 <_glx_> tubeit: Discord Moderator
14:26:43 <_glx_> discord desktop shows me the target url when hovering, but I'm not sure it's that visible on mobile
14:27:29 <emperorjake> Compromised account, removed links and kicked
14:27:45 <_glx_> mobile mail app have a similar issue, it's a pain to see the read address
14:28:42 <xarick> 12134 savegame is of some master, not RC1
14:29:15 <_glx_> they should load in RC1 (same version)
14:30:03 <xarick> oh, you're right, it loaded for me
14:31:08 <xarick> maybe he doesn't have the Game Script?
14:31:16 <xarick> League Tables is not on bananas
14:33:03 <xarick> asserts disabled in rc1?
14:33:18 <_glx_> and the crash dump is not related to the assert
14:34:00 <xarick> let me try with his linked AI
14:34:06 <_glx_> but it's a valid crash, and I think I know the cause for this one
14:34:07 <xarick> instead of what I have installed
14:35:20 <_glx_> using a debug build is probably not the best idea 🙂
14:37:44 <peter1138[d]> So the script fails to compile, so it dies, and confuses things.
14:39:17 <_glx_> that's definitely a bug
14:40:06 <xarick> ah, with his linked script, I get a crash
14:40:50 <peter1138[d]> Yeah, the bad script is not handle properly when loading
14:43:12 <_glx_> oh I know why, we are in the start loop, so not all AIs are initialised but the GUI assumes they are
14:46:11 <xarick> what was the commit that changed building around roads and bridges?
14:46:41 <_glx_> just check the open issues, it's already reported
14:48:50 <xarick> not exactly that, but a derivation
14:49:17 <_glx_> it's the same root cause
14:50:19 <_glx_> oops my fix doesn't work, I wanted to be smart but it breaks savegame loading
14:51:33 <peter1138[d]> Why are all the bugs reported in stupidly large savegames... this one could have a 64x64 1 AI test case!
14:52:42 <_glx_> but I also know how the assert could trigger
14:53:02 <_glx_> running game, edit the script, reload ai in a slot
14:53:23 <xarick> I reported the bug to rei-artist, then he found another bug and created a report with my save
14:53:45 <_glx_> right when it's autosave time
14:54:01 <_glx_> at least that's the order of events I can imagine
14:55:34 <xarick> _glx_: So it was Kuhnovic's fault!
14:56:40 <xarick> then LordAro tried to fix Kuhnovic
14:57:01 <xarick> and in the end it ruined my AIs
14:57:15 <merni> Seems a lot of things ruin your AIs :p
14:58:58 <LordAro> sounds like someone should add extra regression AI testing
14:59:26 <LordAro> this is not my terminal
14:59:59 <merni> Shame! Do you not use a terminal IRC client?
15:00:21 <LordAro> but this is not my main terminal :p
15:01:52 <xarick> when an ai builds a road bridge, an extra road bit is additionally built before and after
15:05:31 <xarick> i wonder how many other AIs the bridge change will affect
15:05:54 <xarick> i already saw AAAHogEx building some weird roads
15:05:55 <merni> except those that never build bridges I guess :p
15:05:59 <_glx_> as it affects players too
15:06:01 <xarick> so probably he's affected
15:11:11 <xarick> ah, it's not, as far as I can see
15:11:25 <xarick> a new version got released and probably fixed it
15:12:16 <xarick> or... 14.0-rc1 is not affected
15:15:27 <peter1138[d]> Or it doesn't care what the specific error message is.
15:17:32 <xarick> yes, my complain exactly
15:17:43 <merni> Regarding #12119, is some kind of a warning (when starting the game, say) needed when setting max loan to zero, or is it obvious enough when starting a game that you have no money?
15:18:05 <merni> Anyway I plan to add a note in the setting description
15:19:18 <_glx_> I hate debugging with 4k² saves
15:20:02 <_glx_> 2 minutes to load a game is not fun
15:20:56 <_jgr_> Presumably you can make MSVC build with symbols and optimisations at the same time?
15:21:04 <xarick> Admiral AI is affected
15:21:55 <_glx_> and it will never be fixed because it's very old
15:23:03 <_jgr_> Also 2 minutes is luxury compared to testing with valgrind 😛
15:23:23 <_glx_> oh and this time I don't have conditional breakpoints
15:25:32 <peter1138[d]> Can we undo the error message change?
15:26:50 *** D-HUND is now known as debdog
15:27:44 <merni> peter1138[d]: which change would that be?
15:28:41 <kuhnovic> Did I break everything again?
15:28:57 <xarick> you broke... 70% of road AIs 🙂 kinda
15:29:19 <peter1138[d]> But maybe it's not?
15:29:30 <merni> Ah, I am looking at a different error message related bug and assumed somehow you were talking about that :p
15:30:26 <xarick> BorkAI is also affected, I won't bother with more screenshots
15:35:46 <xarick> CivilAI is not affected 🙂
15:37:17 <_glx_> #11606 doesn't seem to change error behaviour, but it's possibly a side effect of the change itself
15:37:51 <xarick> speaking of CivilAI, this AI has some hidden potential with infinite money
15:38:04 <xarick> never seen it expand soooo much before
15:39:13 <merni> That's to be expected :)
15:42:11 <merni> station_cmd.cpp contains almost 5000 lines
15:42:25 <merni> is that the longest source file (other than the 3rd party headers etc)?
15:42:57 <merni> No, newgrf.cpp has over 10000
15:43:32 <_glx_> (don't check settings before we split the files)
15:43:59 <_glx_> it was not compilable on OSX
15:44:05 <merni> lang files are also pretty long but nobody edits those apart from english.txt I guess
15:46:05 <xarick> CluelessPlus not affected, Convoy not affected, the tables are turning...
15:46:36 <_glx_> no need to list all affected AIs
15:46:53 <merni> or indeed unaffected ones
15:47:30 <_glx_> would be simpler for you to test the commit before #11606 merge, then the commit with #11606 merge
15:47:57 <_glx_> and it would also be more helpful 🙂
15:49:05 <_glx_> it's always better to known when something broke than it's broken
16:00:52 <xarick> oh, it was LordAro's fix, not Kuhnovics
16:01:46 *** Flygon_ has joined #openttd
16:03:53 <merni> I was getting confused as to why I couldn't bomb a town into having a bad rating of me
16:04:07 <merni> THen I realised my new game config still has Eternal Love GS 😆
16:06:07 <merni> What's an easy way to quickly tank your local authority rating?
16:08:18 <merni> _glx_: Hm I guess I should have more trees in my map then
16:08:30 <merni> Anyway I got a save with appalling rating, that should be enough to test repeatedly :p
16:08:41 *** Flygon has quit IRC (Ping timeout: 480 seconds)
16:09:10 <xarick> yep, 11606 is the cause
16:10:01 <xarick> what are you gonna do about the error message returned? revert?
16:12:28 *** virtualrandomnumber has joined #openttd
16:12:31 *** virtualrandomnumber has quit IRC ()
16:15:27 <xarick> I fixed my AI meanwhile, but if it's gonna be reverted...
16:24:11 *** Flygon_ has quit IRC (Ping timeout: 480 seconds)
16:25:37 *** HerzogDeXtEr has joined #openttd
16:28:01 <xarick> talltyler: I'm in the process of adding unbunch to AIs, and I notice they have a lot of flag combination freedom, even some no one would think about.
16:28:33 <xarick> some can be enforced from the script side, but others are better enfoced from the Cmd side of things
16:29:26 <xarick> for example, HALT with unbunch
16:33:10 *** gelignite has joined #openttd
16:56:27 *** gelignite has quit IRC (Read error: Connection reset by peer)
17:19:28 <xarick> i could rant so much about unbunching...
17:19:54 <xarick> what about service at nearest non-unbunch depot
17:20:53 <xarick> or just service at specified non-unbunch depot
17:21:47 <xarick> what about adding an unbunch order when a service at nearest order exists, etc...
17:23:22 <xarick> or combining fast with slow vehicles... or implicit orders being added/removed triggering unbunch resets
17:24:10 <_glx_> it's impossible to handle all possible cases
17:24:40 <xarick> or even the routes themselves, pathfinder does not always return the same exact path, vehicle does not always go on the same side of the road
17:24:58 <merni> _glx_: ... has NOT_REACHED() also been removed?
17:25:31 <xarick> skewing average round trips
17:25:48 <merni> what is `[[fallthrough]]` actually? Macro?
17:26:10 <_glx_> FALLTHROUGH macro is not needed anymore because there's a c++ attribute common to all compiler
17:26:11 *** geli has quit IRC (Quit: Stay safe!)
17:26:27 <_glx_> they used to have their own version before
17:26:36 <merni> so it's called "attribute"?
17:27:16 <merni> just wondering the right word to use in codingstyle.md to replace "macro"
17:27:47 <merni> Also, is it still correct to say that "All fall throughs must be documented"?
17:28:10 <LordAro> it's somewhat redundant, compilers will complain if you don't
17:29:10 <_glx_> it's the PR removing some macros
17:30:03 <merni> Hm I don't think any others are mentioned in codingstyle
17:30:34 <_glx_> I think when it's obvious (like in your PR) the fallthrough is optional, but it's important after a huge block of code in a `case`
17:30:58 <merni> How about "Non-trivial fall throughs must be documented, using a `[[fallthrough]]` attribute." ?
17:35:59 <merni> Lol it does an entire compile even for doc changes
17:49:25 <peter1138[d]> Tubeless is sometimes a faff.
17:53:08 <merni> can I ask for backport-requested on 12139 and 12141?
17:56:03 <_glx_> can't you set the label ?
17:56:29 <merni> I guess only devs can do it
17:57:04 <LordAro> theoretically we could use milestones to automatically determine what gets backported
17:57:19 <merni> well, then, someone needs to add it to the milestone
18:06:35 <merni> Oops, under the new commit message rules 12140 should not be a Doc:
18:06:48 <merni> but a Codefix: or something
18:08:06 *** Xaroth9232 has joined #openttd
18:08:08 <merni> Is changing PR title enough or does the actual commit also need to be amended?
18:10:35 *** Wormnest has joined #openttd
18:35:44 <LordAro> merni: looks like a Doc to me
18:36:04 <DorpsGek> - Update: Translations from eints (by translators)
18:52:07 *** Flygon has quit IRC (Read error: Connection reset by peer)
18:58:00 <xarick> I didn't explain myself clearer 😦
18:58:24 <xarick> or I don't understand 2TallTyler
19:27:21 <wensimehrp> I also disagree with this change
19:27:21 <wensimehrp> Say, a player has unbunching enabled while using fast locomotives and a large fleet. Vehicles are always forced to enter one depot under a limited speed. This might cause vehicle bunching.
19:28:14 <ajmiles> _glados_: I'm taking a baby steps approach. Maybe I can find a way to offload some blitting a bit at a time
19:29:15 <_glados_> Maybe use a compute shader for it?
19:32:18 <_zephyris> wensimehrp: Try it 😉 It doesn't cause bunching, but it does limit the number of trains out running the route. If you have enough vehicles on a route to make that a problem then you probably don't need unbunching. The bigger solution is multi tile depots!
19:37:07 <talltyler> Yep, if vehicles entering the depot are interfering with the next trip, you don’t need unbunching at all!
19:43:03 <xarick> it's only going to work properly in a very limited number of scenarios
19:43:44 <_jgr_> Those limited number of scenarios are where most of the users are
19:44:11 <andythenorth> I recently gravedug an unhelpful post about why OpenTTD development is basically impossible in the core game
19:44:22 <andythenorth> we're pretty great at the infra, bananas etc
19:44:52 <andythenorth> but anything in the game...lol
19:45:28 <andythenorth> seems the only thing worse than not developing the game is....developing the game
19:45:58 <andythenorth> I suspect JGRPP makes so much more progress, because there are no other contributors to stop it
19:46:19 <andythenorth> much as I appreciated all the help with FIRS, it became easier to make progress solo
19:46:45 <_jgr_> I find this all a bit strange, given the unprecedentedly high rate of features in vanilla these days
19:47:35 <andythenorth> I am not in a mood forgiving of difficult people today 😛
19:47:50 <andythenorth> unfortunately that tends to make me act like one of the marvin-bots 😦
19:47:53 <ajmiles> _glados_: Yeah. I need to look into how much consistency there is in sprite sizes. If they're arbitrary sized, that could be tricky to do without a DIspatch per sprite - which is too slow
19:48:02 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
19:48:10 <peter1138[d]> They are arbitrary sizes.
19:48:10 <ajmiles> ajmiles: I'm wondering if a RasterizerOrderedView might help
19:49:00 <ajmiles> No powers of two, just a big soup of 17 x 9 sprites in all odd sizes?
19:49:27 <ajmiles> I think it's still going to be a job for raster
19:52:24 <xarick> do players even know how to share orders?
19:53:25 <peter1138[d]> Any powers of 2 are purely coincidental 🙂
19:54:05 <locosage> after bonky I wonder if there could be any less consistency in sprites xD
19:55:07 <ajmiles> All texture dimensions are either a power of two, or sum of powers or two.
19:55:59 <kuhnovic> IMO unbunching has limited use for trains since it takes so long for them to enter/exit depots. But for ships and road vehicles it's a real gamechanger.
19:56:59 <locosage> dunno, probably aint gonna change my game...
19:58:35 <talltyler> For what it’s worth, if anyone finds a bug that I have created and wants me to fix it, they should open an issue on GitHub. I’ve blocked some unhelpful commenters here on Discord (so if you ping me with a screenshot of my code, I will not see it) and am not reading scroll back . 🙂
20:00:59 <peter1138[d]> Ah, some stashes to delete.
20:04:46 <xarick> I tried it on ships. It works okayish
20:05:06 <xarick> but as soon as I add one more ship... bam...
20:05:13 <xarick> I add one more order... bam...
20:08:17 <_glx_> every time you update it needs at least a full loop to update, it's expected
20:08:34 <xarick> a full loop on every vehicle!
20:09:41 <xarick> if autorenew/autoreplace is enabled... bam, you might as well forget about it
20:09:55 <xarick> I wonder about auto refit
20:10:39 <_glx_> first vehicle about to leave the depot will set a base delay, and each next vehicle will improve it
20:11:27 <_glx_> it needs to gather data
20:18:02 <wensimehrp> talltyler: Makes sense🤔
20:18:02 <wensimehrp> But what if the player wants to build say a terminus station and for managing purposes they just use orders like go via waypoint and go to nearest depot to unbunch?
20:18:27 *** Wormnest has joined #openttd
20:22:22 <xarick> let's see what happens
20:26:09 <_glx_> as town stores the grfid used to generate its name it's not easy to change an existing town, but you should be able to change `game_creation.town_name` setting in the console
20:26:19 <xarick> wow, it works with refit!
20:26:53 <_zephyris> Of course it works with refit...
20:27:09 <_zephyris> Have you read the blog post on how it works?
20:27:30 <_zephyris> wensimehrp: If your orders are this complex then unbunching probably isn't for you 🙂
20:28:23 <_zephyris> Bashing new features if you haven't read how they work is not polite
20:30:46 <_glx_> it will work for this vehicle
20:32:02 <_glx_> it's useless, but it works
20:33:10 <_zephyris> Oops, well, proof reading of the post would have been useful
20:35:24 <_glx_> yeah town generation window needs some love 🙂
20:35:41 <xarick> > if new vehicles get added to the shared order route, the interval takes a few cycles to stabilize, but it will work itself out shortly.
20:35:41 <xarick> a few cycles, well... yeah, not really
20:35:45 <_glx_> like many editor windows
20:36:23 <xarick> unless the entire itinerary is very short
20:36:44 <_glx_> a few cycle means full trip
20:36:57 <_glx_> if your trips are very long it takes longer
20:37:04 <michi_cc[d]> xarick: Did I mess the memo that you are now forced by law to use unbunching?
20:37:20 <_glx_> but it will fix the interval after some time
20:37:51 <_glx_> unless you constantly change the number of vehicles or the orders
20:39:40 <xarick> giving the example of ships at oil rigs... ships are slow, an entire full trip can take 2 years
20:39:45 <_glx_> but as soon as one of the vehicles to unbunch finish a full trip it's pretty accurate
20:42:43 <xarick> can you make autorenew not reset unbunching?
20:42:54 <xarick> since it's renewing to the same engine type
20:43:22 <_glx_> it's a new vehicle, but I guess it should be possible to backup more data
20:44:59 <xarick> won't be a trivial task
20:45:04 <_jgr_> The unbunch data is in BaseConsist
20:45:31 <_jgr_> So is (or at least can be) carried across same as the other data in there
20:45:54 <_glx_> sinas128: using console command works, but it needs closing/reopen of the window
20:47:30 <xarick> autoreplace is that magic function that seems to do a lot of Cmds, like unsharing, stopping, starting, adding order, sharing order...
20:48:18 <_glx_> oups, it's a rocksmith stream
20:49:35 <_glx_> found the setting, won't happen again
20:51:36 <xarick> testing 12137: So you can't use go to nearest depot on an unbunch order, but you can on a non-unbunch order
20:53:22 <_glx_> that's the kind of trip I would never use unbunch for, as typically I want full load for carg
20:53:50 *** nielsm has quit IRC (Ping timeout: 480 seconds)
20:55:14 <_glx_> a ship without full load makes no sense to me (unless for passengers)
20:55:55 <xarick> Just increase number of ships, unbunch will help with station ratings, which will help the industry produce more cargo, hopefully
20:56:30 <_glx_> I just always have a ship loading
20:56:58 <_jgr_> It's sometimes possible to move cargo in both directions with refits
20:57:49 <_glx_> but the example is the same route switching cargo for each trip
20:58:46 <_glx_> would be better to use 2 groups
20:59:09 <locosage> xarick: how is it gonna help with ratings if they're not waiting on station?
20:59:35 <_jgr_> You can just turn ratings off with the new "sandbox mode" anyway 😛
21:00:53 <_glx_> I'm not sure about the number, I know the first townname is near 21
21:01:48 <_glx_> yeah I guess a first improvement could be to tell the window the setting changed 🙂
21:02:06 <_glx_> (next step would be to add a dropdown in the window)
21:12:16 <xarick> hmm, thinking about the 2 different depots waiting to unbunch and how it affects time
21:14:05 <_glx_> you can't have 2 different unbunching depot for an orderlist
21:15:18 <xarick> with nearest depot, vehicles could wait to unbunch at different depots
21:15:35 <xarick> what's the worse that could happen?
21:17:23 <_glx_> PF should always find the same depot as nearest
21:17:45 <xarick> nope, think of a bus on the left lane vs right lane
21:18:05 <xarick> after leaving from drivethrough
21:18:58 <talltyler> Or a multi-tile station
21:19:09 <_glx_> I think it won't matter as the delay is based on the vehicle trip
21:19:10 <andythenorth> /me doesn't understand
21:19:50 <talltyler> Yeah, it should be fine in 99% of cases and just not work as intended (but not deadlock) in the 1%.
21:19:58 <_glx_> some will wait in one depot, some in another, but in the end they will wait
21:20:52 <_glx_> the depot itself doesn't matter, it's only the current order when entering/exiting the depot
21:21:06 <_glx_> and the shared order list
21:22:13 <xarick> pathfinder could route the vehicle to enter the drivethrough from the opposite side
21:22:44 <_glx_> not a problem, unbunching use the average total time
21:22:45 <talltyler> Theoretically it’s possible to get vehicles on two disconnected networks if they don’t have to use the same depot
21:23:14 <talltyler> You would need to intentionally try and break things this way, but people do that sometimes 🙂
21:23:35 <talltyler> The worst that would happen would be that maybe the average wouldn’t be accurate. Nothing would deadlock.
21:24:10 <ajmiles> peter1138[d]: Are there any save games that act as a benchmark / stress test for the blitter?
21:24:19 <talltyler> To be honest, I created the issue and the PR because I didn’t intend for that to be possible. Not because there’s really anything wrong with allowing it, I think.
21:25:37 <xarick> I guess if that is the case, yeah, exactly what I was thinking... but now I feel guilty
21:26:27 <talltyler> Eh, the bug is closed and no longer requires my attention, I don’t really care how we got here 😛
21:28:05 <xarick> I prefer less restritions
21:28:52 <peter1138[d]> ajmiles: Not really. Add some "real 32bpp"-named NewGRFs if you want lots of large extra-zoom 32bpp sprites that generally slow things a bit.
21:29:05 <peter1138[d]> (They are pretty badly made and do lots of over-drawing)
21:29:09 <xarick> Xarick's 4kx4k, is that hosted somewhere?
21:29:28 <xarick> I may have a more up-to-date version of it 😛
21:30:06 <peter1138[d]> OpenGFX2 is quite sane 🙂
21:30:59 <_glx_> xarick: nah it's not using EZ or 32bpp grf I think
21:31:20 <_zephyris> peter1138[d]: The creator, however, is less so
21:32:21 <_glx_> most likely not the drawing but the action2
21:32:29 <peter1138[d]> Animated tiles is gameloop resources, not blitter intensive 🙂
21:33:04 <locosage> peter1138[d]: idk about real 32bpp stuff but auz industries that are similarly made mostly rendered slow because they didn't fit sprite cache
21:33:08 <locosage> not because of overdrawing
21:33:56 <peter1138[d]> As far as blitter stress tests go, overdrawing is about all there is.
21:34:17 <peter1138[d]> Drawing more pixels than necessary wastes time.
21:34:24 <locosage> 4k monitor + max zoom out xD
21:39:17 <xarick> 12136, it's going to iterate over the same OrderList 3 times
21:41:09 <peter1138[d]> As long as it's not iterating over the OrderListPool 3 times.
21:41:41 <xarick> how many orders can there be in OrderList?
21:41:49 <_zephyris> talltyler: FWIW I agree both ways. Not allowing nearest depot is "cleaner", but I can't think of a pathological case from allowing it.
21:43:46 <xarick> all the 3 checks are required, so I wonder if that couldn't be done more elegantly :
21:44:31 <xarick> Maybe use CommandCost instead of bool
21:44:38 <peter1138[d]> Find a performance issue before trying to optimise one away.
21:53:08 <_zephyris> ajmiles: I guess your worry is arranging many small sprites in 2^n-based texture patches? OpenGFX2 HighDef is good for large sprites, but loading lots of town building NewGRFs on a 4kx4k map would probably be better for testing loading of many sprites.
21:55:36 <peter1138[d]> Xarick I need a save with more orders than road vehicles :p
21:56:46 <_glx_> so clone without sharing
21:58:11 <xarick> I have certainly reached order limit before
21:58:14 <_glx_> worse case I think I can easily write a test GS for that
21:59:15 <peter1138[d]> `if (_vehicle_pool.items > _order_pool.items)` heh
22:00:10 <xarick> so that means AAAHogEx was using vehicles with no orders?
22:00:44 <_glx_> and articulated engines
22:04:48 *** m1cr0man has quit IRC (Quit: G'luck)
22:05:06 *** m1cr0man has joined #openttd
22:09:07 <peter1138[d]> Btw Wentbourne does not use shared orders, and that still sees an improvement.
22:09:11 <xarick> loading old savegames seems to enable infinite money, that setting didn't exist at that time
22:10:12 <xarick> is this one good enough?
22:10:13 <peter1138[d]> Yes, seems to go by current setting.
22:16:05 <xarick> this is a only ships savegame
22:16:32 <xarick> it has more orders than road vehicles, this one being 0
22:26:15 <_glx_> ah buoys can be a test case
22:27:01 <xarick> ok, 57 MB savegame, need to host on OneDrive again
22:46:39 *** Wolf01 is now known as Guest371
22:52:47 *** Guest371 has quit IRC (Ping timeout: 480 seconds)
23:05:06 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:14:55 <ajmiles> _zephyris: I think I'm going to try and see if there's a way of building it without resorting to atlases
23:26:30 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
23:28:55 *** keikoz has quit IRC (Ping timeout: 480 seconds)
23:44:18 *** tokai|noir has joined #openttd
23:44:18 *** ChanServ sets mode: +v tokai|noir
23:51:07 *** tokai has quit IRC (Ping timeout: 480 seconds)
continue to next day ⏵
