IRC logs for #openttd on OFTC at 2020-01-02
⏴ go to previous day
00:27:12 *** snail_UES_ has joined #openttd
00:30:48 *** Progman_ has joined #openttd
00:36:19 *** Progman_ is now known as Progman
02:19:25 *** Wormnest_ has joined #openttd
02:47:59 *** WormnestAndroid has quit IRC
02:48:12 *** WormnestAndroid has joined #openttd
03:40:02 *** Wormnest_ has joined #openttd
04:16:48 *** Betaquarks has joined #openttd
06:55:06 *** crazystacy has joined #openttd
06:55:58 <crazystacy> i am running it from the terminal. don't have gui access
07:10:39 *** HerzogDeXtEr has joined #openttd
08:33:20 *** sla_ro|master has joined #openttd
09:06:56 *** andythenorth has joined #openttd
09:11:10 <andythenorth> Pikka: tramz next? :)
09:56:09 <andythenorth> hmm is train pathfinding broken recently?
09:56:16 * andythenorth is seeing odd things
10:01:50 <andythenorth> nope just a weird PBS artefact I think
10:15:44 <peter1138> Well, that could still be broken...?
10:20:06 <andythenorth> savegame contains unreleased grfs as usual :P
10:20:14 <andythenorth> which is why I don't report bugs
10:20:30 <andythenorth> and grfs have usually been reloaded
10:22:53 <peter1138> I wasn't going to look anyway
10:23:28 <andythenorth> was it lunch yet?
10:59:59 *** Progman has joined #openttd
11:27:21 *** andythenorth has joined #openttd
11:44:57 *** m1cr0man has joined #openttd
11:46:15 *** fonsinchen has joined #openttd
11:46:31 *** Hazzard has joined #openttd
11:46:50 *** ^Spike^ has joined #openttd
11:47:15 *** Terkhen has joined #openttd
11:47:15 *** V453000 has joined #openttd
11:47:15 *** ChanServ sets mode: +o Terkhen
11:58:50 *** snail_UES_ has joined #openttd
12:10:08 <peter1138> Is it elevenses then?
12:19:49 <peter1138> Gosh timezone comments are so new!
12:21:21 <Eddi|zuHause> i haven't seen any this whole decade
13:52:19 <peter1138> 10 minutes til I do it.
14:04:54 *** skrause has joined #openttd
14:10:30 <skrause> Hi, is a max drawing rate of around 30 fps normal? It seems that the mouse cursor is also only 30 fps which feels a bit laggy, is there any way to improve this? I'm using version 1.9.3 on Windows and OpenTTD is using less than 10% CPU of a single core (~1% in Windows task manager with 8 cores), so it doesn't look CPU-limited.
14:16:16 <FLHerne> skrause: That's normal
14:16:52 <FLHerne> AAUI there's a redraw every tick, which is exactly 30ms, so just over 30fps
14:17:18 <Pikka> andythenorth, probably not trams. rail pathfinder preprocessor next, make route building faster. :)
14:17:49 <Pikka> btw, you mentioned it seemed a little unaggressive before... there was a bug which basically stopped it building new cargo trains once it had a passenger network. should be fixed now. :)
14:18:11 <andythenorth> yes, it's building more now
14:18:16 <andythenorth> seen the 'one order' bug?
14:21:07 <FLHerne> skrause: OpenTTD's architecture is still very much a classic '90s "one big game loop" thing
14:22:06 <andythenorth> it also tickled some PBS edge case
14:22:26 <andythenorth> the one-order train somehow managed to prevent all other path reservations on that route
14:22:33 <andythenorth> so the other trains were stuck
14:22:44 <andythenorth> the AI eventually just scrapped them all :)
14:25:03 <Pikka> did this happen a few decades into the game?
14:26:18 <FLHerne> (I still think this auto-closing of "stale" issues is a mistake...)
14:27:30 <andythenorth> Pikka: 9 years in
14:27:34 <andythenorth> according to savegame
14:28:09 <Pikka> well that's one theory out ;)
14:28:11 <andythenorth> FLHerne: stalebot is gone currently afaik
14:28:29 <andythenorth> I disagree but eh
14:28:41 <Pikka> do you have the autosave immediately before it built the train? it might replicate.
14:28:56 <andythenorth> 'closing issues' seems to be the modern equivalent of tabs vs. space
14:29:21 <andythenorth> one crowd likes them 'in case they're useful' and the other crowd 'finds pointless things pointless'
14:32:02 <FLHerne> The issues don't go away because you've closed them
14:32:39 <FLHerne> Sure, close vague bugs where it's unclear what went wrong or if there's even a problem
14:33:23 <FLHerne> But closing real issues that have generated discussion just means you end up with duplicated effort
14:33:34 <FLHerne> And people repeatedly asking the same question
14:34:03 <andythenorth> it's just not an argument that's going to resolve one way or the other
14:34:29 <FLHerne> The cursors, or the bug-closing? :P
14:34:37 <andythenorth> the approach to tickets
14:34:43 <andythenorth> there are two camps, and probably a third camp who wish it was never talked about
14:35:28 <andythenorth> I 'solved' this at work
14:35:39 <andythenorth> a bunch of tickets get moved into 'not now'
14:35:53 <andythenorth> and then after 2 years or so, they all get closed, because the emotional attachment has gone
14:36:08 <andythenorth> just closing things causes way too much emotion
14:36:14 <skrause> FLHerne: Thanks. Yeah I've seen this issue (went through a lot of Google results searching for a solution...) and am mainly asking because I've seen some 60 fps YouTube videos of OpenTTD (e.g. https://youtu.be/lu0BojXU8G0) where the mouse cursor also looks like 60 fps (or at least much better), so there must be *some* improvement I don't see locally.
14:39:53 <andythenorth> turn off 'full animation'
14:40:18 <andythenorth> but if you're getting 30fps already, that shouldn't make much difference
14:40:38 <FLHerne> I don't really see it
14:41:01 <DorpsGek_III> [OpenTTD/website] auge8472 opened pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexzi
14:41:03 <FLHerne> But then I play Portal on this laptop at ~15fps, so perhaps not the most discerning viewer
14:41:19 <andythenorth> game is supposed to run at ~34fps
14:41:40 <andythenorth> nobody should be seeing 60fps afaik
14:51:56 <andythenorth> nothing to do with 30fps in OpenTTD just a good article
14:52:15 <peter1138> I've been playing Doom II WADs for a while again.
14:52:29 <peter1138> Currently working my way throug Hell Ground, which is... somewhat difficult.
14:53:25 <spnda> my game running at 32fps for me while on 165hz
14:55:16 <andythenorth> "This remains the key lesson that still matters today: there are often tradeoffs that can be made that gets you a significant advantage in exchange for limitations that you can successfully cover up with good design." – John Carmack
15:01:03 <andythenorth> Pikka: it's built some steel trains, after I got my company to feed a steel mill :)
15:11:39 <andythenorth> oops crashed it ;)
15:11:52 <andythenorth> PowerWalk L122 and L132
15:14:29 <peter1138> Oof, credit card balance... £1800 :(
15:15:03 <peter1138> Need to stop spending :p
15:18:13 <peter1138> Oh yeah I bought that UPS too.
15:18:20 <peter1138> And Christmas presents.
15:18:34 <spnda> Are NewGRF waypoints always class id 1? And Stations are just 2..255 for ids?
15:19:07 <peter1138> They're always WAYP class.
15:20:07 <Pikka> andythenorth, got a screenshot of the debug log?
15:21:09 <andythenorth> not in one screen
15:22:10 <andythenorth> there's nothing useful further up the screen, just repetition
15:22:40 <andythenorth> scroll down is just slightly different numbers
15:23:35 <Pikka> ya. such a pain when it crashes during one of those iterative walks.
15:24:09 <Pikka> can you see where it got up to with the track conversion on the map? anything interesting there?
15:26:47 <Pikka> tile 63346 is 0xF69B, if that makes locating it any easier with the ? tool :)
15:33:22 *** supermop_work has joined #openttd
15:33:33 <Samu> i'm getting crashes when loading my old savegames :(
15:34:34 <Samu> does that mean backwards compatibility is no longer assured?
15:41:20 <andythenorth> my guess was somewhere else tbh
15:42:27 <andythenorth> that's an unserviced route, partially converted to el
15:42:36 <Pikka> yeah, that looks like it
15:42:54 <Pikka> and it'll be the other player's station there that's breaking it
15:46:15 <spnda> I need some advice on a problem. Similar to waypoints, I want 2 different "types", settable by the grf coder. I thought of something like a property with a byte or word changing the type in the spec. Though then I would need to pick the ones specifically by some sort or something. Maybe define a specific range for when it is one type or another?
15:54:29 <DorpsGek_III> [OpenTTD/website] glx22 approved pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jex2y
16:16:20 <andythenorth> oof not really viable
16:17:05 <DorpsGek_III> [OpenTTD/website] LordAro commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexas
16:19:00 <DorpsGek_III> [OpenTTD/website] auge8472 commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexal
16:19:45 <DorpsGek_III> [OpenTTD/website] LordAro commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexa0
16:23:16 <supermop_work> have not actually done any work the past 90 min
16:24:03 * andythenorth is doing compliance
16:24:10 <andythenorth> I love this shit
16:24:26 <supermop_work> i do have a lot of time sheet stuff to do
16:26:13 <DorpsGek_III> [OpenTTD/website] auge8472 dismissed a review for pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jex2y
16:26:14 <DorpsGek_III> [OpenTTD/website] auge8472 updated pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexzi
16:28:34 <DorpsGek_III> [OpenTTD/website] LordAro approved pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexaP
16:29:01 <DorpsGek_III> [OpenTTD/website] auge8472 commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexaX
16:30:31 *** Wormnest_ has joined #openttd
16:31:19 <DorpsGek_III> [OpenTTD/website] LordAro commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexaS
16:38:16 <DorpsGek_III> [OpenTTD/website] auge8472 opened pull request #136: More or less complete rework of the page header https://git.io/JexVT
16:40:37 <DorpsGek_III> [OpenTTD/website] auge8472 dismissed a review for pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexaP
16:40:37 <DorpsGek_III> [OpenTTD/website] auge8472 updated pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexzi
16:44:36 <DorpsGek_III> [OpenTTD/website] auge8472 commented on pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexVG
16:45:14 <DorpsGek_III> [OpenTTD/website] LordAro approved pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/JexVn
16:45:45 <DorpsGek_III> [OpenTTD/website] LordAro merged pull request #135: Change: remove ruleset for block #openttd-logo-ipv6 from base.css https://git.io/Jexzi
16:48:36 <DorpsGek_III> [OpenTTD/website] auge8472 updated pull request #136: More or less complete rework of the page header https://git.io/JexVT
16:56:26 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on issue #7496: Crashes on start - malloc(): invalid next size (unsorted) https://git.io/fjqCX
17:02:14 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick opened issue #7891: Loading a save from 1.7.2 crashes on 1.10.0-beta2 https://git.io/JexVP
17:05:54 *** snail_UES_ has joined #openttd
17:19:48 <Samu> the crash is on CanBuildVehicleInfrastructure
17:21:27 <Samu> seems like a not road types crash
17:27:00 <Samu> i had in the config, a max number of road vehicles set to 0
17:27:29 <Samu> it was a ship only game, all vehicles = 0 except ships
17:46:03 <spnda> in the action 4 code, I don't quite understand why all features above 3 have some special id. They use another switch case with GB(id, 8, 8). What's the difference?
18:18:34 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on issue #7891: Loading a save from 1.7.2 crashes on 1.10.0-beta2 https://git.io/JexVP
18:38:31 <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on issue #7891: Loading a save from 1.7.2 crashes on 1.10.0-beta2 https://git.io/JexVP
18:41:35 <LordAro> Samu: as best as i can tell, it's related to there being an "effect vehicle" (smoke, explosions, sparks, etc) being saved. the game is then encountering this on load, and falling over
18:41:50 <LordAro> for some reason it's trying to make it a road vehicle, which isn't working so well
18:43:40 <Samu> there's only ships in the game
18:43:54 <Samu> they can breakdown if i recall
18:44:30 <LordAro> peter1138: don't suppose you remember anything about the above?
18:51:41 <DorpsGek_III> [OpenTTD/OpenTTD] michicc commented on issue #7891: Loading a save from 1.7.2 crashes on 1.10.0-beta2 https://git.io/JexVP
18:54:02 <DorpsGek_III> [OpenTTD/website] LordAro commented on pull request #136: More or less complete rework of the page header https://git.io/Jexo3
18:55:13 <spnda> Does nobody know the difference between a vehicles name string and a stations class and name strings?
18:57:22 <nielsm> spnda: many pecularities of newgrf stem from it being defined for ttdpatch
18:57:29 <nielsm> with all the constraints ttdpatch had to work within
18:58:33 <DorpsGek_III> [OpenTTD/website] auge8472 commented on pull request #136: More or less complete rework of the page header https://git.io/JexoW
18:58:55 <spnda> true. but I would assume it'd still be basically the same. A stations name seems to be loaded quite differently as opposed to a stations class id.
18:59:18 <nielsm> I think it has to do with being able to use fewer bits to refer to a string
18:59:25 <spnda> Andd I want this new feature to have a class id string and a name string aswell, so I would need something similar, I guess.
18:59:37 <nielsm> i.e. since you know you're working with a train, you look in the train strings table
18:59:42 <DorpsGek_III> [OpenTTD/website] LordAro commented on pull request #136: More or less complete rework of the page header https://git.io/Jexo4
19:05:40 <Samu> maybe the FOR_ALL vehicles isn't iterating the same thing as the new iterator?
19:07:12 <nielsm> Samu no read the code again
19:07:21 <nielsm> what's the difference between "type" and "v->type" ?
19:07:43 <nielsm> answer: the first does not reference the vehicle being iterated over
19:12:12 *** andythenorth has joined #openttd
19:17:52 <Samu> testing the last commit before NotRoadTypes
19:18:29 <Samu> 21edf67f89c60351d5a0d84625455aa296b6b950 Peter Nelson <peter1138@openttd.org> 13-04-2018 20:03:11 +00:00 Codechange: Untangle game and scenario toolbars.
19:22:17 <DorpsGek_III> [OpenTTD/website] auge8472 updated pull request #136: More or less complete rework of the page header https://git.io/JexVT
19:28:30 <Samu> c02ef3e4564b7b54d49f0827d2d7625cbc38f335 peter1138 <peter1138@openttd.org> 06-04-2019 06:46:15 +00:00 Feature: Add NotRoadTypes (NRT)
19:28:42 <Samu> that's the faulty commit
19:29:00 <nielsm> thank you for revealing nothing new
19:29:14 <Samu> uhm just wanted to make sure
19:34:58 <peter1138> nielsm, it's amazing, isn't it?
19:39:26 <Samu> what's the v->roadtype of effect vehicles supposed to be?
19:39:44 <Samu> maybe a savegame conversion is missing?
19:41:12 <peter1138> I would guess that effect vehicles don't have a road type, because they're not road vehicles.
19:42:05 <nielsm> if (type == VEH_ROAD && GetRoadTramType(RoadVehicle::From(v)->roadtype) != (RoadTramType)subtype) continue;
19:42:15 <Samu> if (type == VEH_ROAD && v->type == type && GetRoadTramType(RoadVehicle::From(v)->roadtype) != (RoadTramType)subtype) continue;
19:43:09 *** frosch123 has joined #openttd
19:43:38 <nielsm> the basic error is that it never tests whether v is a road vehicle before trying to get it as one
19:45:47 <DorpsGek_III> - Update: Translations from eints (by translators)
19:46:50 <Samu> will it crash for current savegame version?
19:47:26 <nielsm> it can probably crash for any savegame version
19:47:35 <nielsm> even without having a save
19:48:04 <nielsm> if the game needs to check whether a player is capable of build road infrastructure while an effect vehicle is active
19:48:20 <Samu> max num of road vehicles = 0
19:55:26 *** Wormnest_ has joined #openttd
20:10:31 * LordAro compiling something prior to 0.6
20:12:40 <frosch123> should we push commits to old branches to keep them compileable?
20:12:57 *** snail_UES_ has joined #openttd
20:13:30 <nielsm> I suspect that would be lots of work
20:13:57 <nielsm> easier to build some VMs with old junk software
20:14:39 <frosch123> ottd 0.4 actually compiles fine, just one line to fix
20:14:49 <frosch123> later versions need more changes for some reason
20:15:12 *** WormnestAndroid has quit IRC
20:15:26 <frosch123> LordAro: so, i recommend branch 0.4
20:15:41 <nielsm> more advanced = more hard
20:16:17 *** WormnestAndroid has joined #openttd
20:16:49 <LordAro> frosch123: i need savegame 72, unfortunately
20:17:11 <frosch123> nielsm: full compilation, 12.5 real, 34.2 user
20:17:25 <frosch123> LordAro: that's still in master. libpng works like that
20:17:41 <frosch123> c people wanting to use exceptions :)
20:18:19 <LordAro> i had to fiddle with them at work recently
20:18:31 <LordAro> seems that they don't work on mingw, but __builtin_longjmp does
20:18:40 <LordAro> (computers were a mistake, etc etc)
20:18:52 <frosch123> hmm, my 0.5 checkout is old, i need "svn upgrade" first :p
20:19:36 <frosch123> hmm, no modifies, apparently never compiled it...
20:27:04 <LordAro> error: exception cleanup for this placement new selects non-placement operator delete
20:28:27 <nielsm> gotta find an older compiler with less smarts
20:29:12 <frosch123> the problem is mostly libpng
20:29:16 <LordAro> ah, that was an easier fix than i thought
20:29:23 <frosch123> png12->14 was quite incompatible
20:29:37 <LordAro> mm, i cherry-picked the commit that fixed that
20:37:39 <andythenorth> am I the cat or the train?
20:38:51 <andythenorth> maybe I should get a Doom WAD editor
20:39:04 <andythenorth> it's been 22 years since I last made a Doom WAD
20:45:01 <nielsm> the tooling has improved a lot since then
20:45:20 <Wolf01> Think about making IH for mashinky
20:50:48 <LordAro> Error: Cannot open file 'sample.cat'
20:50:56 <LordAro> i forgot about the dangers of pre1.0
20:51:13 <Wolf01> Download it from bana...oh
20:56:33 <frosch123> get a free sample cat with every copy of ttd
20:57:03 <frosch123> now i finally understand why andy is looking for the cat all the time
20:58:25 <LordAro> frosch123: alternatively, download it from the first result on google!
21:03:16 <DorpsGek_III> [OpenTTD/website] LordAro approved pull request #136: More or less complete rework of the page header https://git.io/Jex6d
21:03:28 <DorpsGek_III> [OpenTTD/website] LordAro merged pull request #136: More or less complete rework of the page header https://git.io/JexVT
21:03:59 <andythenorth> can't remember the origin of is cat
21:04:08 <andythenorth> there was a precursor to FIRS with cats
21:04:24 <glx> [19:42:16] <Samu> if (type == VEH_ROAD && v->type == type && GetRoadTramType(RoadVehicle::From(v)->roadtype) != (RoadTramType)subtype) continue; <-- you again choose the complex path
21:04:24 <glx> if (v->type == VEH_ROAD && GetRoadTramType(RoadVehicle::From(v)->roadtype) != (RoadTramType)subtype) continue; <-- this should be enough
21:05:10 <glx> type equality is checked on the nex line anyway
21:11:41 <Samu> do i change line 1780 too?
21:11:59 <nielsm> the cause of the crash is RoadVehicle::From(v) when v is not a road vehicle
21:12:29 <nielsm> the purpose of the loop is to detect whether the player owns any vehicles requiring the specific type of infrastructure
21:12:42 <andythenorth> maybe I've got the photoshop file, let's see
21:13:03 <glx> 1780 should be safe already
21:13:16 <glx> only the right engines are selected
21:13:18 <nielsm> the condition tested for in that line is whether the infrastructure searched for is different from the required
21:13:43 <nielsm> the loop doesn't terminate when it finds a negative, it terminates when it finds a positive
21:13:57 <glx> bug looks like a "bad" copy/paste
21:15:52 <Samu> oh i see line 1789 does the check
21:17:01 <frosch123> hmm, there are some more hunks, compared to whenever i compiled it last
21:17:21 <frosch123> -rwxr-xr-x 1 frosch frosch 28880064 Jul 30 2012 bin/openttd <- ah, 7.5 years
21:19:04 <frosch123> LordAro: are all of them errors, or did yuo also fix warning?
21:19:10 <LordAro> frosch123: all warnings
21:19:58 <frosch123> anyway, a lot bigger than the 0.4 diff
21:22:50 <andythenorth> the click / double-click interaction would work exactly like cargo subtype list, which is not bad at all
21:22:57 <frosch123> hmm, that old binary claimns that libpng is the only lib with an incompatible so-version.
21:23:04 <andythenorth> there would be a new prop or cb for the group subtype string
21:23:11 <glx> andythenorth: better with the offset I'd say
21:23:38 <andythenorth> I thought so, but precedent is the subtypes refit list, no offset
21:24:28 <frosch123> "no offsets" is better
21:24:41 <frosch123> that way all the columns are real columns
21:25:21 <andythenorth> there is still too much gap in my mockup
21:25:34 <andythenorth> anway prop (cb?) for parent vehicle ID
21:25:45 <andythenorth> prop (cb?) for group substring
21:26:03 <andythenorth> adjust buy menu, available vehicles menu, and auto-refit
21:26:08 <andythenorth> not sure about AI/GS
21:26:20 <frosch123> andythenorth: also, no cb please... vehicles moving from the "hot stuff" to the "historic" group hurts my brain
21:26:39 <andythenorth> just we default to cb > property when there's a rationale
21:27:02 <glx> yeah but cb are often abused
21:27:03 <andythenorth> oh and the buy menu sort order, someone solved that (forget who)
21:27:12 <andythenorth> would always just be the properties of the group parent
21:27:36 <andythenorth> if authors want complex sort algorithms for groups within groups
21:27:40 <andythenorth> they can submit a PR
21:27:57 <andythenorth> or wait for someone to do it in JGR
21:28:09 <peter1138> The defacto standard version, eh?
21:28:20 <andythenorth> it's a useful safety valve
21:28:33 <andythenorth> I am in no way jealous of JGR's support work though
21:28:58 <andythenorth> is it in JGR yet?
21:29:23 <frosch123> no idea, i just assumed so
21:30:03 <frosch123> but yeah, maybe all the nullptr, std::vector and FOR_EACH stuff scared him
21:31:05 <nielsm> I think it's sort of in, but not in his main branch
21:31:10 <frosch123> a single person supporting N patches, which are probably mostly abandoned by their authors
21:31:24 <andythenorth> it's a remarkable job
21:31:54 <glx> each big change usually kill a patch pack
21:32:36 <andythenorth> so VehicleBuyMenuGroups or what?
21:32:59 <andythenorth> it's not specifically liveries, as arbitrary properties can be different
21:33:02 <frosch123> yeah, but the patch packs before were a single huge patch (i.e. all changes squashed), while jgr kept things separate
21:33:03 <glx> I like the idea andythenorth
21:33:34 <andythenorth> I mean we _could_ have them inherit the properties of the parent vehicle, and just use the substring as a single var which can be checked in the graphics chain only
21:33:41 <andythenorth> but that destroys the simplicity of the idea
21:33:43 <frosch123> it's definitely solves problems which were addressed by multiple people, not just andy's flipping
21:34:06 <andythenorth> latest Horse reduces what the flipping can do
21:34:16 <andythenorth> it's mostly reversing company colour, which is otherwise random
21:34:22 <andythenorth> or actual flipping
21:34:56 <andythenorth> I pushed it quite a long way past the 'BAD FEATURE' threshold
21:35:17 <LordAro> (╯°□°)╯︵ ɥʇɹouǝɥʇʎpuɐ
21:35:25 *** lpx is now known as Guest12975
21:35:50 <andythenorth> highly localised user feedback is that 'you should get the train you see in the buy menu' but the CC colour flipping is ok
21:37:32 <frosch123> you need to cut the "highly" when they stop sitting in your lap
21:38:45 <andythenorth> that hasn't happened yet, if he wants to see what's on my screen
21:39:08 <andythenorth> also all my stupid livery tricks won't work with RVs
21:39:33 * andythenorth should really do NRT Hog
21:39:39 <andythenorth> and stop doing Horse not-really-better-just-different changes
21:40:02 <frosch123> i guess a single string id is enough to group vehicles? or does it need a 4-byte "vehicle class", so you can group vehicles from different grfs or addon grfs?
21:40:44 <Samu> what do I write in the commit message?
21:40:45 <frosch123> i think a 4-byte id only works when there are "few" things to label properly. if there are many things, people won't bother naming them
21:40:56 <Samu> maybe somebody else fix it
21:41:04 <frosch123> so, i guess, add-on grfs cannot add to existing groups
21:41:12 <andythenorth> I would have grouped by just setting a parent ID in a prop
21:41:19 <andythenorth> it could be optional grfid in a register :P
21:41:26 <frosch123> what is a parent id?
21:41:42 <frosch123> oh, wait... yes you are right
21:41:47 <andythenorth> I assumed one vehicle would be parent node, rest are child nodes
21:41:50 <frosch123> i assumed the group needs a text, but it doesn't
21:42:03 <andythenorth> the parent also has to be repeated in the child nodes, like in subtype refit UI
21:42:09 <andythenorth> bit weird, but precedent is there
21:42:28 <andythenorth> then each child gets an additional string
21:43:03 <frosch123> 1) vehicles can set a group number, which is an arbitrary grf-local number. vehicles with same number are grouped
21:43:32 <frosch123> 2) you do something silly like the vehicle-ordering property, where each vehicle has to reference another vehicle to be in the same group with
21:43:51 <frosch123> i think (2) is awful :p
21:44:04 <nielsm> 2 sounds like a mess to work with
21:44:33 <andythenorth> in 1) how do you determine what's shown on the parent node?
21:44:53 <nielsm> and then you end up with bad grfs with a weird graph of related vehicles
21:45:02 <frosch123> the parent is the last-built vehicle from that group?
21:45:19 <frosch123> or the first vehicle in vehicle-order
21:45:21 <nielsm> lowest vehicle id is default parent of group
21:45:42 <frosch123> grfs already define a sorting order for vehicles, no need to invent a new one
21:45:47 <nielsm> and then sure, you could maybe have an UI option to select a different parent
21:45:47 <andythenorth> I stopped managing IDs a long time ago
21:45:56 <frosch123> the default parent is just the first in the group
21:45:59 <andythenorth> it's quite a bad habit I was in, imagining that ID had some meaning :)
21:46:20 <frosch123> ID is important for savegames :)
21:46:20 <glx> Samu: Fix #7891: really check the vehicle type before casting
21:46:35 <andythenorth> I set IDs manually, I just no longer worry about the value, except that it's unique
21:46:56 <andythenorth> neither 1) nor 2) scare me
21:47:05 <frosch123> but do you define a sort-order for the vehicles?
21:47:16 <andythenorth> but my compile sorts it out
21:47:19 <andythenorth> 2) is a disaster with auto-assigned IDs in nml
21:47:27 <frosch123> and does the parent differ from the 1st vehicle according to that sort-order?
21:47:29 <andythenorth> auto-assigned IDs are a disaster anyway
21:47:54 <andythenorth> but they're a fact :P
21:48:21 <nielsm> for NML you can make it not a property of the vehicles, but a separate top-level thing that defines ordering and grouping, or something like that
21:48:23 <andythenorth> oh yes, the sort order is a neat solution frosch123
21:48:26 <frosch123> yeah, auto-generated IDs are a misfeature of nml
21:48:28 <nielsm> and make that assign ids as well
21:48:38 <andythenorth> use the sort order, then fold them up if they're in the same group
21:49:03 <andythenorth> it's just a tree, and groups create a branch rule
21:53:59 <LordAro> hmm, 0.7 is segfaulting when loading an AI
21:55:52 <andythenorth> I can station walk I guess
21:59:33 <frosch123> LordAro: just move your .openttd folder somewhere where it cannot find it
22:00:04 <frosch123> andythenorth: why can't you add more platforms to existing stations?
22:00:41 <frosch123> oh.... that old issue
22:01:15 <andythenorth> I've also been playing with 'refit any available' in some cases
22:01:30 <andythenorth> which works great, except when cdist (rarely) creates an unwanted link
22:01:35 <andythenorth> which can then never be removed
22:01:45 <andythenorth> one mitigation is to not mix stations
22:01:46 <LordAro> frosch123: i'd rather fix whatever's causing it to crash :)
22:01:47 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick opened pull request #7892: Fix #7891: really check the vehicle type before casting https://git.io/JexPK
22:02:08 <frosch123> LordAro: somewhen you fixed and read-beyond-eof in the tar reader
22:03:08 <LordAro> frosch123: pretty sure it's not that
22:03:14 <LordAro> somehow it's segfaulting in strdup
22:03:27 <LordAro> with a pointer that i'm reasonably certain is not null, or otherwise invalid
22:04:23 <frosch123> is strdup from a different library than malloc?
22:05:18 <frosch123> is the string properly zero-terminated?
22:06:09 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 approved pull request #7892: Fix #7891: really check the vehicle type before casting https://git.io/JexP1
22:09:54 <frosch123> out of standard errors :)
22:13:17 <LordAro> frosch123: runs fine if i use valgrind
22:13:21 <LordAro> which is a bit annoying
22:14:47 <andythenorth> lol 45k population city
22:15:25 <milek7_> maybe try with -fsanitize=address
22:15:34 <frosch123> andythenorth: i have no context, but there is an ancient coop blog about how to grow cities with millions
22:16:44 <andythenorth> I had city growth on high for some reason
22:20:51 <LordAro> milek7_: works, but i'm compiling 0.7 here, there are going to be issues :)
22:22:13 <LordAro> valgrind indicates there's an invalid free in the window/widget code
22:22:22 <LordAro> which is a long way away from the script code
22:22:31 <andythenorth> town hates me, so I bulldoze 16x16 of buildings with magic bulldozer
22:22:43 <andythenorth> it's a reddit post waiting to happen
22:25:05 <frosch123> bulldoze the road piece under the town sign :)
22:25:26 <andythenorth> magic bulldozer solves all problems
22:25:46 <andythenorth> I tried faffing around with that 'remove more town owned stuff' setting
22:25:58 <andythenorth> but it doesn't seem to do anything useful, so magic bulldozer
22:26:08 <andythenorth> I just need an equivalent 'magic town ratings fixer'
22:26:19 <frosch123> with that setting turned off you get the original ttd behaviour
22:26:27 <frosch123> in ttd you can only remove dead ends
22:26:38 <frosch123> once there is a circle in a road, it is meant to stay
22:26:43 <andythenorth> such great and well balanced game mechanics :)
22:30:35 <andythenorth> I have played current OpenTTD game for > 80 game years
22:30:41 <andythenorth> quite rare to go so long, no GS either
22:38:42 <frosch123> wow, you found it :) i assumed the gui code changed so heavily before and after 0.7 that it would be impossible to find
22:39:30 <LordAro> how that causes a strdup to segfault, i'll never know
22:39:35 <LordAro> yay for stack corruption
22:39:45 <frosch123> commit message of 06d7d63216d7e5e2265353c7b45dbf6d7bafa3a7 tells you
22:39:52 <frosch123> it's the optimiser removing code
22:40:28 <LordAro> the comment in the code tells me as well :)
22:40:38 <LordAro> i feel like i've run into that before
22:40:40 <frosch123> that's too long, i did not read it
22:44:33 <milek7_> "The delete operator is overwritten to not delete it;"
22:45:09 <frosch123> delayed deallocation magic
22:45:55 <frosch123> it's used to solve issues like "close windows and nested windows while iterating over all windows"
22:47:49 <frosch123> but yes, you can probably remove all deletes and replace them with some method "MarkForDestruction" or similar
22:50:31 <andythenorth> hmm, really capacity pax vehicles for later game?
22:50:40 <andythenorth> all my stations have 3k-5k pax waiting
22:50:45 <frosch123> funnily ottd also has the reverse concept to "use data after destruction". pools initialise the "index" member before calling the constructor, and the constructor has to make sure to POD-not-initialise that member :p
22:51:20 <andythenorth> cdist network with 7 nodes connected, probably too many
22:51:32 <andythenorth> cdist is probably much better with point-to-point
22:51:56 <frosch123> can you flip those nodes with ctrl+click?
22:52:31 <andythenorth> I might be doing it wrong, but I can't
22:52:44 <andythenorth> I have a 3 node network, only 2k pax waiting per station
22:52:58 <andythenorth> so the conclusion is, cdist requires non-connected stations
22:53:24 <andythenorth> well, maybe 1 connection :P
22:56:50 <LordAro> frosch123: yeah but those use pools and placement new stuff properly, rather than the use-after-free stuff that the windows use
22:57:47 <frosch123> use-before-init is just as bad
23:06:40 <milek7_> curious if this is guaranteed to be correct
23:06:48 <milek7_> or if compiler is allowed to also remove other assigments from that destructor if it wants
23:11:43 <milek7_> "[...] after the lifetime of an object has ended and before the storage which the object occupied is reused or released, the following uses of the glvalue expression that identifies that object are undefined: [...] Access to a non-static data member or a call to a non-static member function. "
23:31:35 *** andythenorth is now known as Guest12979
23:31:35 *** andythenorth has joined #openttd
23:34:11 *** andythenorth is now known as Guest12980
23:34:11 *** andythenorth has joined #openttd
23:53:39 *** andythenorth has left #openttd
23:57:35 *** ChanServ sets mode: +v tokai
continue to next day ⏵
