IRC logs for #openttd on OFTC at 2019-04-27
⏴ go to previous day
00:24:32 *** Wormnest has joined #openttd
00:45:42 <Eddi|zuHause> i vaguely remember someone did a patch so if you build a station piece before and after a bridge/tunnel, it would treat the whole tunnel as platform
00:47:20 <peter1138> This suggestion is a weird idea :p
00:48:41 <FLHerne> peter1138: Stations on bridges/tunnels are obviously useful, particularly for metros
00:49:12 <FLHerne> But putting anything on individual bridge tiles is (currently) impossible because they don't really exist
00:49:35 <FLHerne> Whereas treating the entire bridge as a station is probably relatively trivial to implement
00:50:17 <FLHerne> It's a short-termist hack, but I can see the appeal
00:51:01 <peter1138> Short-termist hacks lead to long-term maintenance nightmares.
00:51:05 <FLHerne> It's basically the same workaround as the "pretend every <n>th bridge/tunnel tile is a signal" patch
00:51:43 <FLHerne> Yes, so it's unsurprising that neither of them have ever been merged :P
00:52:44 <FLHerne> I still like cirdan's [IIRC?] idea of allocating off-map strips of maparray for bridges and tunnels
00:53:45 <FLHerne> It would solve a lot of these silly problems, although I suppose cache locality would be hurt a bit
00:54:13 <FLHerne> ...or just make the map array twice as big per tile and have done with it?
00:55:28 <peter1138> Does not solve anything.
00:57:39 <Heiki> just make the map 3-dimensional, e.g. 4096×4096×256
01:00:36 <peter1138> Good for memory and performance I'm sure.
01:03:42 <Heiki> isn’t that what they build those supercomputers for?
01:05:15 <DorpsGek_II> [OpenTTD/OpenTTD] stale[bot] commented on issue #7059: Town name language choice affects number of towns / world population https://git.io/fjsNe
01:28:55 *** Wormnest has joined #openttd
01:56:53 <Eddi|zuHause> Heiki: that would require some multicore-ness of the algorithm
01:59:49 *** Supercheese has joined #openttd
02:14:12 <Artea> towns with 90k population is insane :o
02:14:28 <Eddi|zuHause> the highest i heard about was 3M
02:15:35 <Eddi|zuHause> it probably involves optimizing the road layout
02:15:57 <Artea> there is zones with 4 town fused
02:16:18 <Eddi|zuHause> that probably won't help either :p
02:16:54 <Eddi|zuHause> it's not so much about the size of the map
02:17:15 <Eddi|zuHause> but at some point the growth speed of towns cancels out with the decay speed of houses
02:18:02 <Eddi|zuHause> so you want to max out the growth speed by avoiding that the growth algorithm gets caught in dead end roads
02:18:48 <Eddi|zuHause> (also, you can use NewGRFs who provide houses with bigger population and longer lifetime)
02:19:18 <Artea> I need to review the newgrfs in next version
02:20:48 <Eddi|zuHause> you want to avoid dead end roads that end at rails, oceans or other unbuildable points
02:21:09 <Eddi|zuHause> and sometimes it helps if you build tunnels from the town center to the outskirts
02:23:14 <Artea> names with more than 100k population: Sillyton and Tastyweed
02:25:00 <Artea> is hard to control the towns now
02:31:33 <Artea> just made Sillyton lose 8k population :(
02:33:11 *** Wormnest has joined #openttd
03:37:37 *** Wormnest has joined #openttd
04:17:45 *** Supercheese has joined #openttd
04:42:25 *** Wormnest has joined #openttd
05:20:16 *** Gustavo6056 has joined #openttd
05:24:58 *** Gustavo6056 is now known as Gustavo6046
05:46:56 *** Wormnest has joined #openttd
05:56:46 *** HerzogDeXtEr has joined #openttd
06:44:51 *** ChanServ sets mode: +v tokai
06:51:31 *** Wormnest has joined #openttd
06:55:42 *** Alberth has joined #openttd
06:55:42 *** ChanServ sets mode: +o Alberth
07:56:02 *** Wormnest has joined #openttd
08:11:35 *** andythenorth has joined #openttd
08:47:56 *** Supercheese has joined #openttd
09:00:30 *** Wormnest has joined #openttd
09:13:15 <peter1138> Oops, I broke the scenario editor.
09:14:44 *** andythenorth has joined #openttd
09:32:10 *** sla_ro|master has joined #openttd
09:33:55 *** arikover has joined #openttd
09:35:08 <LordAro> Alberth: feel like looking into 6605 a bit more? :)
09:36:38 <Alberth> I read your explorations, didn't help much in understanding what happens
09:37:25 <Alberth> hmm, maybe add a watch on the poolsize changing?
09:42:06 <LordAro> watchpoints make everything super slow
09:42:32 <LordAro> i suspect that there actually 5 i valid depots in the savegame
09:43:19 <LordAro> though i've no idea how they got there
09:46:22 <Alberth> set a break point at the afterload "thingie 5" point
09:53:22 <LordAro> isn't that too late though? the depot pool has already been initialised by that point
09:53:34 <peter1138> It's all too late cos it's already happened ;)
09:54:36 <Alberth> it's a bug in an earlier conversion?
09:55:02 <peter1138> The savegame is already broken.
09:57:00 <peter1138> Ok, should I add the show-coverage button to the scenario editor town window, or just ignore coverage there?
09:57:16 <Alberth> ie in an earlier upgrade to the current version of the savegame thus
09:57:38 <LordAro> or in how the game was originally saved
09:57:47 <peter1138> It's just a broken save as far as I know.
09:58:02 <peter1138> But one that's been reported more than once.
09:58:07 <LordAro> peter1138: seems useful to add it to the scenario editor?
09:58:47 <Alberth> adding it to the SE would be useful, people may want to pixel-perfect put down industries etc
09:59:45 <peter1138> Urgh, debug build loading slow :p
10:01:23 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGeg
10:03:41 <andythenorth> meh old Severn Bridge is closed
10:05:14 *** Wormnest has joined #openttd
10:18:42 <Wolf01> I think I got lost there once, playing geoguessr
10:32:38 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGeh
10:41:44 <LordAro> ok, can't use IsDepotTile inside Load_DEPT, it's too early
10:41:49 <LordAro> they're all invalid :)
10:44:16 *** Progman has joined #openttd
10:51:11 <LordAro> alright, that actually works, interesting
10:56:33 <LordAro> i do believe i have a fix for the invalid depots though
10:56:38 <LordAro> not sure how correct it is, but..
11:02:30 *** arikover has joined #openttd
11:09:31 *** Wormnest has joined #openttd
11:56:48 <Samu> aha, finally a round where my AI sucks
11:59:43 <Samu> looks like 100k is too little for a route
12:04:05 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGvH
12:04:27 <andythenorth> weird forums day
12:04:31 <andythenorth> is NRT merged then?
12:14:04 *** Wormnest has joined #openttd
12:30:44 <Wolf01> 7km walk, not so much tired but completely wet
12:39:28 <Artea> GoG is selling Theme Hospital for 5.29 euros
12:40:20 *** gareppa has joined #openttd
12:46:44 <andythenorth> it's big it's wood it's Hog
12:51:01 <Artea> so happy my vps can handle OTTD
12:58:49 <DorpsGek_II> [OpenTTD/OpenTTD] ldpl commented on issue #7534: GSGoal.QuestionClient sends message to wrong client https://git.io/fjGf4
13:04:18 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGfa
13:18:36 *** Wormnest has joined #openttd
13:20:47 <andythenorth> much and very sprites
13:37:27 *** alekseiytalanov has joined #openttd
13:37:33 *** alekseiytalanov has left #openttd
13:38:30 *** alekseiytalanov has joined #openttd
13:38:35 *** alekseiytalanov has left #openttd
13:39:10 *** alekseiytalanov has joined #openttd
13:40:28 <alekseiytalanov> Good classic graphic for oldschool
13:42:47 <alekseiytalanov> This is game OpenTTD Configure this game setup graphic and settings and setup game with NewGRF and start transport business
13:46:52 *** alekseiytalanov has left #openttd
13:56:41 *** alekseiytalanov has joined #openttd
13:57:25 *** alekseiytalanov has left #openttd
14:00:15 *** alekseiytalanov has joined #openttd
14:01:35 *** alekseiytalanov has left #openttd
14:07:13 <peter1138> Wolf01, rained or... sweaty? :p
14:09:07 <andythenorth> LordAro: do we use it? o_O
14:12:43 <LordAro> andythenorth: i'm not actually sure
14:12:52 <LordAro> it was the unreliable part of ...something
14:13:00 <LordAro> but i can't remember if that got fixed when azure happened
14:13:08 <DorpsGek> LordAro: Truebrain was last seen in #openttd 1 week, 5 days, 18 hours, 12 minutes, and 3 seconds ago: <TrueBrain> if that is the excuse, I am done
14:16:02 <andythenorth> not finding anything from searching
14:17:50 <LordAro> well it definitely uses docker
14:17:57 <LordAro> and that will use dockerhub by default
14:19:28 <nielsm> yeah it publishes to docker hub
14:21:30 <nielsm> doesn't look like there is any unusual activity from the public feed
14:23:07 *** Wormnest has joined #openttd
14:24:13 <peter1138> Ok so shall I add code to handle the case when that Coverage button is missing insteaD?
14:24:41 <nielsm> having it to show town boundaries could be nice too
14:24:54 <nielsm> maybe even the town zones
14:39:54 <andythenorth> so is the update to the website image expected?
14:39:57 * andythenorth has NFI how it all works
14:40:17 <nielsm> yes, the website is updated every time a new nightly is uploaded
14:40:29 <nielsm> since the links need to be fixed to point to the new nightly
14:45:25 <nielsm> I'm thinking making just the percent rating available to AI, but also making the raw numbers (cargo-tiles travelled and ditto delivered) available to GS
14:48:55 <Eddi|zuHause> wouldn't it be easier to have both with the same data access?
14:49:25 <nielsm> the difference is just adding @api -ai to the declaration or not, afaik
14:49:51 <nielsm> but the raw numbers aren't available to human players so an AI shouldn't be able to get them either
14:57:57 <Eddi|zuHause> what's the rationale for hiding it from the player?
14:57:58 <peter1138> Switched back to Mate from Gnome. At least I can resize OpenTTD properly :p
14:58:32 <peter1138> I guess it's not so much hiding rather than not displaying too much mostly irrelevant data?
14:58:46 <nielsm> Eddi|zuHause: mostly that they're very large numbers that are difficult to make sense of
14:58:53 <peter1138> (Like articulated overtaking vehicles isn't "disabled")
14:58:59 <nielsm> difficult to fit into the UI in a sensible way
14:59:35 * peter1138 ponders support for house tile layouts...
14:59:48 <Eddi|zuHause> NoOvertaking branch next?
15:00:31 *** gelignite has joined #openttd
15:00:51 <peter1138> I don't know. Maybe it's something NRT could address?
15:01:01 <peter1138> "Highways" are currently useless.
15:01:10 <Eddi|zuHause> i feel like it's out of scope for NRT
15:01:26 <Eddi|zuHause> feels very orthogonal
15:01:55 <Eddi|zuHause> but it ties in with all the state machine bullshit
15:02:19 <peter1138> Yeah, 3-lane highways might be nice.
15:02:28 <peter1138> And, perhaps, 1-lane tracks.
15:02:58 <Eddi|zuHause> 1 lane tram on 2 lane road
15:03:23 <peter1138> There's no provision for tram/road compatibility :/
15:03:34 <peter1138> Like, tram down the middle between two lanes.
15:04:27 <peter1138> Maybe the game should define... yes, that.
15:04:46 <nielsm> bool do_overtake = (my_max_speed > blocking_vehicle_max_speed + 5) && ((my_max_speed - blocking_vehicle_max_speed) * blocking_vehicle_length < distance_to_next_intersection);
15:04:51 <nielsm> there solved the problem for you
15:05:19 <peter1138> nielsm, now make it an option to satisfy the "it should be an option like 90 degrees" ... o_O
15:05:42 <nielsm> hell no make that a newgrf flag on road types
15:05:45 <peter1138> Hmm, so if NRT gets to define lines... maybe it defines the direction of a lane as well.
15:06:19 <Eddi|zuHause> parallel tram stations would be nice
15:06:20 <nielsm> well could maybe have "tram lanes"?
15:06:42 <nielsm> like boulevards with tram tracks in the middle
15:06:47 <andythenorth> all these things
15:07:05 <Eddi|zuHause> also: two-tile wide road, 2 road lanes ->, 1 tram lane ->, 1 tram lane <-, 2 road lanes <-
15:07:48 <Eddi|zuHause> there should be space for 4 lanes on a tile
15:07:58 <Eddi|zuHause> (with no sidewalk)
15:08:53 <nielsm> hmm, I wonder if it really is right to use just int32 for these cargotiles values
15:09:03 <Eddi|zuHause> problem with offset lanes is, how to make transition tiles?
15:09:24 <Eddi|zuHause> also, height offset?
15:09:44 <peter1138> Eddi|zuHause, quite.
15:11:28 <peter1138> Yeah, how to transition?
15:11:36 <peter1138> And not via hormones.
15:12:02 <Eddi|zuHause> <insert inappropriate comment here>
15:13:18 <peter1138> And... we already have one-way roads via map bits.
15:13:35 <Eddi|zuHause> that should probably be ripped out then
15:14:34 <peter1138> Maybe consider them as they are: no-entry flags
15:14:39 <Eddi|zuHause> transition tiles would work like crossing, composed of trackbits?
15:14:40 <nielsm> how would a transition from regular road with tram track to two one-way roads (opposite directions) in parallel with tram track in the middle even look?
15:14:48 *** Gustavo6046 has joined #openttd
15:15:16 <Eddi|zuHause> the transition is in the center of the tile, provided by the NewGRF
15:16:22 <Eddi|zuHause> also, bidirectional tram track, but no signalling? :)
15:19:25 <peter1138> I guess we don't want new road types for each possible lane combination?
15:19:47 <Eddi|zuHause> what i'm imagining now is: each roadtype defines valid lane configurations, and the configuration is stored with the trackbit somehow
15:19:59 <peter1138> Eddi|zuHause, that's where I was just headed.
15:20:04 <Eddi|zuHause> then the NewGRF is responsible to provide graphics for all configuration combinations
15:20:12 <peter1138> Then it can use varactions to pick the graphics based on the lanes.
15:21:07 <peter1138> There's a few spare bits lying around.
15:25:20 <nielsm> that doesn't involve a 2x2 tile prefab
15:26:53 <LordAro> peter1138: fixed the "maglev under bridges gets converted to standard rail" issue
15:26:58 <LordAro> guess what caused it :)
15:27:23 *** Wormnest has joined #openttd
15:31:17 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro opened pull request #7546: Fix #6507: Don't try to load invalid depots from older savegames https://git.io/fjGUZ
15:31:26 <LordAro> peter1138: ^ get told
15:32:51 <peter1138> Someone mentioned old saves were broken a few months ago but never provided any examples for me to work with.
15:33:01 <peter1138> And never filed a bug, so it got forgotten.
15:33:08 <peter1138> And by someone, I mean a dev.
15:33:23 <LordAro> well there were 2 examples in the previously mentioned bug reports :p
15:33:30 <LordAro> i think nielsm mentioned TTO saves were broken
15:33:44 <LordAro> and i imagine no one's tried TTD saves in a while
15:33:45 <peter1138> This was several months ago.
15:34:08 <peter1138> And it was an off the cuff remark in IRC. Not here.
15:34:28 <peter1138> So that's a backport candidate :p
15:34:45 *** frosch123 has joined #openttd
15:35:01 <peter1138> But hey, that's not why depots were wrong ;p
15:35:12 <peter1138> And those bug reports were about crashing on load, not wrong rail types.
15:35:33 <LordAro> true, but once you fixed those issues... ;)
15:35:53 <peter1138> You'd have to look around the map to notice it was wrong as well, as it wasn't specifically reported.
15:36:45 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7546: Fix #6507: Don't try to load invalid depots from older savegames https://git.io/fjGUC
15:38:06 <peter1138> Damn, I was ~27 when I started tinkering with OpenTTD.
15:38:21 <Eddi|zuHause> i was ~12 when i started playing TTO
15:38:44 <LordAro> yeah, but you're all old
15:39:02 <LordAro> i think i was introduced to TTD when i was 6 or 7
15:39:11 <LordAro> not sure i ever played TTO, though i think we had it
15:39:38 <Eddi|zuHause> i found TTD(Patch) around 2001-ish
15:40:01 <Eddi|zuHause> and OpenTTD around 2006?
15:41:08 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7546: Fix #6507: Don't try to load invalid depots from older savegames https://git.io/fjGUg
15:41:55 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGU2
15:42:01 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro merged pull request #7545: Fix 81d335b: Coverage button missing from town window in scenario editor. https://git.io/fjGeg
15:45:37 <Artea> is together with OTTD 0.3.4
15:47:25 <peter1138> Hmm, I have to remember how to start pulseaudio on Windows :/
15:47:50 <peter1138> Yes, I use it to get sound out of my Hyper-V Linux VM :p
15:48:44 <peter1138> Hyper-V assumes you're running Windows in it and using Remote Desktop, so just does audio that way.
15:49:27 <LordAro> ooh, found another save that crashes
15:49:32 <LordAro> this one looks kdtree related
15:49:39 <peter1138> Yeah, that happens.
15:49:46 <peter1138> ProZone 13 is one such save.
15:49:54 <peter1138> There's a PR to fix it
15:50:17 <peter1138> Problem is the usual dependency issues :(
15:51:05 <peter1138> Basically it depends on the map array before the map array has been converted.
15:51:31 <peter1138> Profit: £29,573 £154,587
15:51:37 <peter1138> Efficiency: 76% 73%
15:52:01 <nielsm> I did consider making it a table
15:52:04 <LordAro> peter1138: confirmed, that PR fixes it
15:52:59 <nielsm> but the profit numbers can also get lartge
15:53:12 <nielsm> especially if you use currencies like yen
15:53:35 <peter1138> Well, it can get wider?
15:53:41 <peter1138> Depends what's worse :p
15:57:04 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh updated pull request #7353: Feature: Measure vehicle capacity utilisation efficiency https://git.io/fhho4
15:57:32 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on issue #7371: Kdtree is built too early in savegame loading process https://git.io/fjGUd
15:58:20 <LordAro> except it's actually locomotion
15:58:51 <peter1138> Also, not for Windowsx.
15:58:56 <peter1138> Been out a few years.
16:00:29 <Artea> I'm so off I didn't knew
16:00:48 <peter1138> There was even a bunch of tt-forumers who met up with... someone? CS even? about it.
16:01:00 <frosch123> nah, with the marketing guy
16:02:40 <Artea> I was talking to my friend yesterday and he asked something that I'm curious
16:03:08 <Artea> is there an year cap or goes on forever ?
16:03:44 <Artea> that is one of the goals of my server
16:04:08 <frosch123> in which year did you start?
16:05:04 * LordAro looks up the maximum year
16:05:07 <LordAro> hmm, might take a while
16:05:11 <frosch123> LordAro: don't spoil it
16:05:29 <LordAro> i'm all about not spoiling things at the moment
16:05:38 <LordAro> bloody endgame and not getting to see it until tuesday
16:06:17 <Artea> towns already reach 100k
16:06:31 <frosch123> Artea: make contact with some cryosleep company
16:07:07 <frosch123> or write a letter to your grand-grand-children
16:08:55 <LordAro> in 900 years you're not sitting on silly piles of money?
16:08:55 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on pull request #7353: Feature: Measure vehicle capacity utilisation efficiency https://git.io/fjGTk
16:09:33 <Artea> and too many aircraft crashes
16:11:07 <Artea> I wanted to start AroAI yesterday
16:11:36 <Artea> seems off now after so long
16:11:51 <Artea> still have player playing
16:11:56 <Artea> also dp joined yesterday
16:12:30 <LordAro> i'm not sure there are any AIs that are properly balanced to actually play against
16:13:06 <LordAro> but my knowledge of AIs is about 5 years out of date
16:13:32 *** Maarten has joined #openttd
16:13:36 <Artea> but crashes after 10 years or so
16:14:54 <LordAro> i wonder if in the event of a crash, the game should just restart the AI and pretend nothing happened
16:15:03 <LordAro> (with appropriate protections against repeated crashing)
16:17:19 <Artea> I wonder how you bot would handle towns with 50k+ population
16:18:01 <Artea> probably is a good feature
16:18:12 <Artea> but had to filter some crashes
16:18:20 <Artea> CPU excessed usage is one of them
16:18:30 <Artea> which is what crashes AIAI
16:23:00 <peter1138> Rip quality: 100.00%
16:23:18 <peter1138> 1) Needless precision 2) Buying CDs, LOL...
16:32:07 *** Wormnest has joined #openttd
16:43:25 <nielsm> hmm, ScriptGroup::GetProfitThisYear() seems to not handle sub-groups correctly
16:43:33 *** Eddi|zuHause2 has joined #openttd
16:46:58 <Eddi|zuHause2> so, who ordered this weather?
16:48:44 *** Eddi|zuHause2 is now known as Eddi|zuHause
16:48:55 <Eddi|zuHause> it's always portugal, with their azores
16:48:55 <nielsm> hm okay maybe it is fine after all
16:49:57 <Artea> I never went to Azores or Madeira
16:51:01 <Eddi|zuHause> i went to algarve once
16:51:15 <Eddi|zuHause> that was like 20 years ago
16:55:01 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7546: Fix #6507: Don't try to load invalid depots from older savegames https://git.io/fjGTS
16:55:19 <peter1138> I went to Algarve a few years ago.
16:55:48 <peter1138> Hired a car and drove to the top of Fóia, cos... why not?
16:56:45 <Artea> did u enjoy the beach ?
16:57:41 <peter1138> Yeah, but it was a bit early in the year so not hot all the time. Had a few days of sun and some of rain.
16:58:19 <peter1138> Yeah, it was mid-May iirc.
16:59:18 <Artea> you should try September
16:59:35 <Eddi|zuHause> we went in like february and the locals thought we were crazy for going into the water :p
16:59:59 <Artea> yeah that is normal, Eddi
17:04:26 <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh updated pull request #7353: Feature: Measure vehicle capacity utilisation efficiency https://git.io/fhho4
17:35:55 <peter1138> Hmm, so, NewGRF debug window..
17:36:40 *** Wormnest has joined #openttd
17:38:42 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7104: Fix #5405: Aircraft could route to depots outside their range https://git.io/fjGko
17:42:32 <LordAro> anyone got any thoughts on #7340?
17:42:57 <LordAro> i'm increasingly of the opinion that the TGP values need changing generally, everything is far too flat
17:43:20 <peter1138> The whole thing needs redoing :p
17:44:49 <LordAro> i was under the impression that TGP was already a fairly general perlin noise generator though
17:45:41 <peter1138> Because it writes directly to the map array, it can't be used for nice extras, such as using a separate perlin map to select areas to be desert.
17:46:54 <Eddi|zuHause> ... or meandering rivers
17:48:46 <peter1138> No, git rebase, I don't want to pick ancient revisions that have svn revisions in 2000s range. Or any at all, tbh.
17:51:49 <LordAro> so we're in agreement on the TGP thing
17:51:54 <LordAro> but as an interim fix...
17:53:04 <peter1138> +typedef SmallVector<SignalReference, 4> SignalDependencyList;
17:53:14 <peter1138> Hmm, maybe JGR hasn't updated for a while? Or left SmallVector in.
17:55:42 <peter1138> Or maybe he just doesn't work with these branches any more.
17:58:45 <Eddi|zuHause> <nielsm> https://0x0.st/zbVX.png bad edit, I can't think of a good way to make the transition here <-- i fear to make that not-horrible we also need diagonal track pieces
18:01:11 <peter1138> Let's see what all the fuss is about with JGRPP anyway.
18:01:21 <peter1138> Oh. Doesn't compile.
18:02:23 <peter1138> As it's the "de-facto standard" that's somewhat unexpected.
18:02:55 <Eddi|zuHause> you mean, a loud minority?
18:03:30 <peter1138> So I was seriously going to look at programmable signals...
18:04:11 <peter1138> Maybe I'll try the progsig-rebase branch.
18:04:15 <peter1138> Oh, that doesn't compile either.
18:04:39 <peter1138> Nor progsig-sx. Hmm.
18:05:25 <peter1138> That isn't promising.
18:05:53 <peter1138> If it only works as an old-style block signal, fuck that.
18:06:32 <peter1138> So routing restrictions... is sort of similar in idea, I think?
18:18:14 <Eddi|zuHause> was that the one that works with waypoints?
18:18:27 <peter1138> I don't know. I was going to compile it to test but no luck compiling.
18:18:45 <Eddi|zuHause> considered that it might be just you? :p
18:19:34 <peter1138> error: 'UpdateWidgetSize' overrides a member function but is not marked 'override'
18:19:37 <peter1138> Probably not just me.
18:20:31 <Eddi|zuHause> that sounds like a "it probably compiles with less strict settings"
18:21:15 <peter1138> You mean not everyone uses -Werror?
18:23:00 <peter1138> Yeah, that was it. Shed loads of warnings.
18:23:29 <LordAro> i imagine the override will cause a load of them
18:30:45 <peter1138> "If entered signal of PBS block is at ..."
18:33:15 <peter1138> Yeah, so programmable signals don't with with path signals. Useless.
18:33:35 <Eddi|zuHause> i probably made that complaint like 5 years ago...
18:34:03 <Eddi|zuHause> i certainly said that about signals in tunnels
18:36:17 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7348: Fix: IniLoadFile::LoadFromDisk seems to expect filename, BaseMedia<Tbase_set>::AddFile provides fullpath https://git.io/fjGIc
18:40:30 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro updated pull request #7542: Fix #7255: Prevent crashlog corruption by only printing the 20 most recent news messages https://git.io/fjsVs
18:41:28 <peter1138> path restrictions can be on any signal but... they're just a hint, I guess.
18:41:47 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7542: Fix #7255: Prevent crashlog corruption by only printing the 32 most recent news messages https://git.io/fjGI2
18:43:50 <nielsm> thinking about PBS programmable signals, most basic just "valid path endpoints" which are tiles (which must either have a signal on them or be station platform at end of line)
18:44:26 <nielsm> and extensible with multiple groups of valid endpoints where each group has some condition on it about the train "class"
18:44:33 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7485: Add: Dropdown build menus from toolbar have icons https://git.io/fjGIK
18:44:54 *** Thedarkb-T60 has joined #openttd
18:47:22 <LordAro> i'm tempted to close #7502
18:47:23 <nielsm> could maybe with a system like that allow pointing to path endpoints that are further down the line and that way reserve a path through multiple signals
18:49:19 <Eddi|zuHause> LordAro: no objections
18:49:38 <peter1138> LordAro, is the STAT_CLASS_MAX = 256 -> 255 change no issue?
18:49:44 <LordAro> maybe it is a good idea, but without the extra setting?
18:49:48 <LordAro> peter1138: glx didn't think so
18:49:57 <LordAro> that's about all i can tell you :)
18:50:18 <peter1138> As far as I know, if it changes anything then you get 255 classes tops instead of 256t.
18:50:31 <peter1138> And... who has that many station classes?
18:55:23 <peter1138> #7542, what's the fixed buffer size that's being reached?
18:57:31 <LordAro> but regardless, hundreds of news messages is a bit ugly
19:00:39 <peter1138> Hmm, does the buffer overflow somewhere?
19:00:46 <nielsm> if there's hundreds of recent news messages that in itself might be a symptom :)
19:00:55 <nielsm> no it doesn't overflow
19:01:01 <nielsm> it just stops writing more to it
19:01:11 <nielsm> so the crashlog is truncated
19:01:53 <peter1138> Hmm, the PR says "corruption"
19:02:04 <LordAro> the issue said corruption :p
19:02:42 <LordAro> peter1138: eh, squash it and reword the commit :p
19:03:19 <nielsm> "nice but not necessary": write a line about how many messages were not writen
19:03:39 <peter1138> I'm not too worried, just that if was actual corruption and buffer overflows, then the 32 limit felt a bit arbitrary.
19:03:40 <nielsm> "+ 234 more recent news"
19:04:06 <peter1138> But if it's just truncated rather than corrupted then sure.
19:07:21 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7502: Feature: Allow or disallow large aeroplanes to land on airports with short runways https://git.io/fjGIx
19:07:22 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro closed pull request #7502: Feature: Allow or disallow large aeroplanes to land on airports with short runways https://git.io/fjqQu
19:09:56 <nielsm> "tell the player if the gun is pointed an unsafe way, but don't attempt to grasp at it if it looks like they're about to shoot themselved in the fooot"
19:13:01 *** Wormnest has joined #openttd
19:21:09 <nielsm> how "player attempting to set up an unsafe situation" should be handled
19:21:43 <nielsm> since the direction has so far always seemed to be don't disallow but, but do warn the player
19:33:04 <peter1138> Hmm, so our releases are running bad on OS X?
19:33:08 <peter1138> I guess not everyone sees that.
19:35:08 <LordAro> without someone who actually knows how to debug such things, it's really difficult to do anything about it
19:35:42 <LordAro> one person seems to think it's related to the road vehicle performance stuff
19:36:38 <nielsm> In general the game should allow anything that doesn't violate basic rules. Players should only be prevented from performing actions that are impossible, such as sending a bus to a truck stop, but always allow actions that merely are risky or "stupid", such as sending a bus to a bus stop that could never get passengers. Instead the game should try to warn players about potentially bad
19:36:38 <nielsm> situations, such as warning messages about unprofitable vehicles.
19:36:49 <_dp_> LordAro, it's related on linux but not on osx
19:38:20 <LordAro> nielsm: perhaps a different example? given afaik the game does not warn about sending buses to bus stops with no acceptance :)
19:40:59 <nielsm> no but it does warn you that the vehicle has made no profits
19:45:09 <nielsm> "For example players are not prevented from starting vehicles with no orders, but the game will send a warning about vehicles having too few orders. Because players are allowed to do so, some have made it a challenge to make working transport networks where vehicles have no orders."
19:47:01 <SpComb> I remember trying to build a road network without any orders
19:47:29 <SpComb> for trains it's easy, particularly if you do point-to-point
19:48:11 <SpComb> but building a couple bus stops in some towns, linking them together with roads and then spamming a ton of busses without any orders has a certain appeal
19:48:31 <SpComb> because iirc it does kinda work
20:02:59 <_dp_> nielsm, nice but where was that logic when overbuilding one-way roads with stations was prohibited in 1.9?
20:03:30 <nielsm> I don't know, was it intentional?
20:04:16 <_dp_> also, cough cough, overbuilding signals with rail would be nice ;)
20:04:35 <_dp_> risky and "stupid" but I like it :p
20:07:15 <peter1138> IIRC it was possible to overbuild other company's one-way roads.
20:07:59 <_dp_> peter1138, dunno about other companies but now I can't overbuild my own :(
20:09:59 <peter1138> That's okay, just remove the one-way flags first.
20:10:22 <_dp_> peter1138, if I remove it gets crowded by vehicles
20:11:07 <peter1138> Are you now saying you have a use-case for overbuilding one-way roads, rather than just saying "you can't do it any more"
20:11:16 <_dp_> peter1138, so, basically instead of just building a station now I have to stop all vehicles around, remove one-way, build station, start vehicles again
20:11:36 <peter1138> So *THAT* has never been mentioned before.
20:11:46 <peter1138> It's almost like we have to be psychic.
20:13:41 <LordAro> peter1138: haven't you been issued with your crystal ball yet?
20:14:01 <_dp_> peter1138, what that? I explained it all here the moment I noticed I couldn't extend my station like before.
20:14:26 <_dp_> peter1138, didn't notice it when the code change was being made sadly :(
20:14:31 <LordAro> also our memories are not good
20:14:33 <peter1138> Where it gets lost are a few minutes.
20:15:15 <nielsm> the issue tracker exists to keep track of issues
20:17:07 <peter1138> Jun 6th 2018, I fixed it.
20:17:15 <peter1138> We were using git back then.
20:17:47 *** Wormnest has joined #openttd
20:20:30 *** Supercheese has joined #openttd
20:20:32 <_dp_> well, when I first brought it up here I got the impressions it's pointless to open an issue
20:20:54 <_dp_> I guess it was just Eddi arguing so mb should've just ignored him :p
20:21:16 <peter1138> It's never pointless to open issues.
20:21:39 <_dp_> peter1138, say that to stalebot :p
20:21:47 <peter1138> That's not even my idea.
20:22:15 <Eddi|zuHause> i'm not sure i like stalebot either
20:22:32 <peter1138> It's definitely pointless to comment on here, without explaining your reasoning, and then of course it will be lost because time is fleeting.
20:22:52 <Eddi|zuHause> i do remember that discussion about roads
20:22:55 <dwfreed> I think your development cycle is too slow for stalebot
20:23:37 <Eddi|zuHause> stalebot is here to relief andy of the burden to tell people that their PR is bad
20:23:55 <dwfreed> honestly stalebot only makes sense in a few narrow cases
20:24:07 <peter1138> So anyway, Jun 6th is WAY before stalebot.
20:24:44 <nielsm> also stalebot ignores issues tagged "regression"
20:24:55 <nielsm> so if you have a genuine regression bug it won't get autoclosed
20:25:30 <Eddi|zuHause> but does that cover regressions of the kind "please reinstate spacebar overheating"?
20:29:50 <peter1138> Okay... so error message... should it be the "owned by ..." or the current "road is one way or blocked"
20:31:45 <Eddi|zuHause> [5. April 2019] [17:56:15 CEST] <_dp_> but whose bright idea was to forbid road stations on one-way road? :(
20:32:43 <Eddi|zuHause> that's 4 weeks ago
20:32:54 <peter1138> Somewhat later than the original commit.
20:33:13 <_dp_> peter1138, somewhat immediately after original release :p
20:34:08 <peter1138> So yeah, I see I was even there, but all I see a rant.
20:35:27 <peter1138> stalebot probably wouldn't've even closed it yet if you had made a proper report.
20:36:21 <_dp_> peter1138, ok, ok, I'll make a proper report
20:37:23 <_dp_> peter1138, I was just kinda pissed that stuff got broken
20:37:30 <peter1138> It wasn't broken, it was fixed.
20:37:47 <peter1138> You could previously remove other company's one-way bits.
20:38:14 <_dp_> peter1138, also even if there is no point for one-way issue there still may be for cloning...
20:38:24 <_dp_> peter1138, that kinda got "fixed" as well
20:38:45 <peter1138> So what the fuck are you talking about now?
20:38:47 <Eddi|zuHause> _dp_: i don't think it helps your cause by randomly merging and hopping topics like that
20:38:56 <peter1138> We still don't have crystal balls.
20:39:15 <_dp_> peter1138, I'll make a proper report when I got a clear opinion on that :p
20:40:39 <_dp_> Eddi|zuHause, ok, ok, nothing's broken ... for now :p
20:45:16 <_dp_> but basically, cloning vehicle doesn't open vehicle window anymore
20:45:29 <_dp_> which is nice sometimes but sometimes definitely not
20:45:33 <Eddi|zuHause> i'm pretty sure that used to be a setting
20:45:34 <_dp_> so I'm still experimenting with it
20:46:49 <_dp_> Eddi|zuHause, no, don't think so
20:47:05 <_dp_> actually, it's ctrl-cloning, just cloning still opens it
20:48:03 <nielsm> hmm actually, ctrl-cloning should maybe open the depot window instead
20:48:19 <Eddi|zuHause> (svn r24808) -Feature(ette) [FS#4458]: When share-cloning vehicles don't open the vehicle window of the new vehicle. The only reasonable thing to do with the new vehicle is to start it, which can also be done via the depot GUI.
20:48:20 <Eddi|zuHause> Change: Clone tool in depot window now behaves like clone button in vehicle window. Keep the tool active when share-cloning. Open the vehicle GUI when copy-cloning. (#6754)
20:48:39 <nielsm> e.g. if you've opened an in-depot rv via the vehicles list and then ctrl-clone it, you would get the depot window opened to see the new vehicle there
20:49:47 <_dp_> Eddi|zuHause, yeah, I even commented on that issue but seeing it now I'm having some doubts
20:50:08 <_dp_> basically the idea was to make cloning from depot behave the same as cloning from vehicle
20:50:36 <_dp_> but now that I think of it they're quite different
20:50:57 <Eddi|zuHause> there's also this one: (svn r24955) -Change (r24808) [FS#4458]: Revert to opening the vehicle GUI again when cloning vehicles using the clone-button from the depot GUI. This button cannot be used for sequential cloning, so the argument about opening many windows does not hold as it does for the clone button from the vehicle GUI.
20:51:07 <_dp_> as cloning from vehicle means it's in the same depot and yeah, opening that depot makes sense if it's not open for some reason
20:51:31 <_dp_> but cloning from depot may mean you're on entirely different side of track and you need to verify/skip orders
20:51:46 <Eddi|zuHause> opening the depot does sound sensible
20:58:14 <Artea> we should be able to buy a company with no clients for 5 or 10 years
20:58:59 <Eddi|zuHause> Artea: change the autoclean mode so it treats it like bankrupcy?
20:59:49 <Eddi|zuHause> that's probably not an existing feature
20:59:50 <Artea> I don't know how it work ingame
21:00:26 <Artea> well, I don't think dp company will bankrupt that soon
21:00:58 <_dp_> Artea, lol how is it still alive?
21:01:11 <_dp_> just closes the damn thing! xD
21:01:16 <Artea> 500M profit I'm not mistaken
21:03:33 <Eddi|zuHause> that should be G?
21:04:35 <Eddi|zuHause> i mean the abbreviations are k,M,G,T,...
21:05:04 *** supermop_Home has joined #openttd
21:05:13 <_dp_> I disabled autoclean on citymania recently as it crashed server
21:05:29 <_dp_> some players were quite surprised getting 1 bil goal in 48 hours xD
21:05:40 <Artea> M = Million | B = Billion
21:06:47 *** Suprcheese has joined #openttd
21:07:04 <_dp_> as for cloning after a bit of thought I think I'd prefer to have "clone last" button in depot instead of clone button in vehicle
21:07:28 <_dp_> so just clone would always open vehicle window but clone last won't
21:07:28 <Eddi|zuHause> Artea: in my world, Billion is 10^12, not 10^9
21:08:19 <_dp_> that always got me confused about english
21:08:40 <_dp_> just how do I say 1000000000 in a way that people on both sides of globe understand?
21:09:02 <Eddi|zuHause> that's why i popose 1G
21:09:18 <Artea> we have a middle between Million and Billion
21:09:49 <nielsm> million, milliard, billion, billiard, ...
21:09:55 <_dp_> Eddi|zuHause, imo it's like 10**9, only 1% will get it :p
21:09:56 <nielsm> yes "short" and "long" scales
21:10:30 <LordAro> almost no one uses long scales anymore, i thought?
21:10:53 <_dp_> peter1138, oh, right, that's what I'm using now xD
21:11:29 <LordAro> billion will be understood by the vast majority to be 10^9
21:11:38 <Eddi|zuHause> LordAro: i'm going to propose that more people worldwide use long scale than short scale
21:12:36 <Eddi|zuHause> LordAro: that doesn't help anyone
21:13:46 <LordAro> the only time i've ever known anyone use the long scale in English is to ask "is this in long or short scale?", usually just after they've read some fact site or some other silly website
21:13:53 *** Suprcheese is now known as Supercheese
21:14:23 <LordAro> i understand that directly translating the words to other languages means the values don't necessarily line up
21:14:37 <Eddi|zuHause> i've been sprinkling long scale in occasionally
21:15:22 <glx> using power of 10 is easier ;)
21:17:09 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7547: Change: Allow building road stops over self-owned one-way/blocked road. https://git.io/fjGt6
21:20:24 <nielsm> but that doesn't mean the stop will be one-way, right?
21:20:43 <nielsm> in which case it needs to have one-way roads on one or both ends
21:21:31 <peter1138> No, it gets removed.
21:21:36 <glx> pathfinder will probably take care of that
21:21:38 <nielsm> in which case it would not be a problem removing one-way-ness from the road pieces you intend to build over since the road pieces leading into them would already be one-way
21:22:04 *** Wormnest has joined #openttd
21:22:46 <peter1138> It was never really a problem to remove one-way-ness. The problem was the ability to remove another player's one-way-ness.
21:22:52 <nielsm> in other words, I don't see why this would become a problem: <_dp_> peter1138, so, basically instead of just building a station now I have to stop all vehicles around, remove one-way, build station, start vehicles again
21:23:25 <peter1138> You don't always have space.
21:25:29 <Eddi|zuHause> nielsm: when _dp_ says "one way", he actually means "blocked both ways"
21:25:44 <peter1138> Hmm, I can't remember my ./configure parameters for ARM cross-compilation.
21:25:59 <peter1138> Wait, config.log will have it :p
21:26:52 *** gareppa has joined #openttd
21:29:58 <peter1138> So my main issue is compiling with SDL.
21:32:09 <_dp_> Eddi|zuHause, that's half-truth
21:32:23 <_dp_> Eddi|zuHause, I uses one-ways as a turnaround point for vehicles coming from that side
21:32:32 <nielsm> situation B would never be a problem
21:32:47 <nielsm> situation A would be same with or without change
21:32:56 <peter1138> nielsm, I'm not sure what you are trying to analyze now.
21:33:31 <glx> nielsm: not a problem, user should know what it's doing
21:33:42 <Eddi|zuHause> i go with my initial opinion, the station should remember the oneway-ness. but that might get complicated...
21:35:15 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN opened pull request #7548: Fix #7165: Missed another Append() to push_back() replacement https://git.io/fjGtd
21:35:22 <nielsm> actually I don't see how either of those situations have it matter whether you remove the one-way-ness first and then build the stop, or can build the stop directly on top
21:35:38 <peter1138> nielsm, nobody knows what you're analyzing or why!
21:35:51 <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7548: Fix #7165: Missed another Append() to push_back() replacement https://git.io/fjGtF
21:36:02 <peter1138> nielsm, the intial issue was that you could overbuild *somebody elses* oneway/blocked roads, and it would remove them.
21:36:18 <nielsm> peter1138: I'm trying to understand why it matters that you'd have to remove one-way-ness before building a station on top
21:36:23 <peter1138> nielsm, I fixed that but preventing overbuilding in all cases.
21:36:44 <nielsm> I can't think of any scenario where removing it first would let vehicles move somewhere they wouldn't be able to anyway after building a stop
21:37:59 <peter1138> nielsm, you can't build a road-stop if there are vehicles on a tile.
21:38:10 <peter1138> nielsm, if it's blocked, you there will be no vehicles on the tile.
21:39:18 <peter1138> Hmm, I could do with an android emulator.
21:39:28 <peter1138> I don't think it'll work inside hyper-v though :/
21:39:50 <nielsm> so it's road-block used as owned land except the town can build houses out from it
21:40:13 <_dp_> nielsm, good luck extending that in 1.9
21:42:10 <_dp_> glx, pause? what's that? :p
21:42:13 <peter1138> Then they're just be paused on the tiles you want to build on :p
21:42:53 <nielsm> (in my world that kind of setup should not count towards city growth)
21:42:53 <Eddi|zuHause> i'm still thinking ctrl+click to stop the vehicle would properly handle your use case...
21:43:08 <glx> you only need to stop one vehicle I think
21:43:36 <Eddi|zuHause> yeah, stop 1 vehicle, and be done with the upgrade before the quantum tunneling triggers
21:46:46 <V453000> yay haven't seen this one in a while Unable to allocate ID for [random]switch, sprite set/layout/group or produce-block. Try reducing the number of such blocks.
21:47:30 <peter1138> Oh nice, github now lets me choose which identity to use.
21:47:31 <Eddi|zuHause> V453000: you have too many switches alive at the same time, try reordering things?
21:47:35 <DorpsGek_II> [OpenTTD/OpenTTD] PeterN merged pull request #7548: Fix #7165: Missed another Append() to push_back() replacement https://git.io/fjGtd
21:47:59 <Eddi|zuHause> V453000: a switch is "alive" between its definition, and the last use
21:48:22 <Eddi|zuHause> V453000: only 256 switches may be "alive" at the same time
21:48:23 <V453000> I know Eddi, I didn't add, just re-ordered :) moving big parts of code around
21:48:34 <V453000> ah, good to know the number at least :)
21:49:58 <V453000> I've got a convoluted system where a train can look like a wagon but a wagon can look like a train :)
21:50:19 <V453000> so of course you need all of them at the same time :D but I can split some, should work
21:50:46 <Eddi|zuHause> you might need to duplicate a few things
21:50:49 <_dp_> Eddi|zuHause, easier said than done: they're constantly moving and clipping all over each other
21:51:23 <Eddi|zuHause> _dp_: i don't see the problem, you still only need to stop one
21:51:27 <_dp_> Eddi|zuHause, and it's already dangerous enougt, one wrong move and you got a horde of trucks running amok in your town xD
21:51:32 <V453000> not sure actually, will be a mess either way
21:52:30 <V453000> actually now I read the line where the switch fails to be created, it's probably not such a nightmare
21:52:41 <peter1138> warning: comparison of integers of different signs: 'unsigned int' and 'socklen_t' (aka 'int')
21:53:09 <peter1138> Oh, we typedef int socklen_t
21:55:27 <V453000> also, last time I arrived to this problem of running out of IDs, I had extremely random code. Now I even started using includes in my code :D custom ones from python, but still :) easy to move large chunks of code around
21:56:03 <Eddi|zuHause> V453000: we'll get you to higher level metaprogramming at some point :p
21:57:43 <Eddi|zuHause> peter1138: i would expect such a type to be defined in some system library
22:00:00 <peter1138> We define it in some cases, but not in this case.
22:00:45 <Wolf01> Oh nice, they added the fields to Mashinky
22:00:49 <peter1138> ./sysroot/usr/include/sys/socket.h:typedef int socklen_t;
22:01:05 <peter1138> So we compare a size_t with socklen_t.
22:02:03 <peter1138> But not in a normal build.
22:03:21 <LordAro> doesn't really seem like it should be an int
22:03:29 <LordAro> but standard library kinda wins there
22:04:14 <peter1138> On my system headers it's typedef __socklen_t socklen_t;
22:04:32 <peter1138> (And no other reference to __socklen_t)
22:06:44 <glx> it's defined to int for windows and os2 in os_abstraction.h
22:06:50 <peter1138> Yes, as it should be.
22:10:18 <peter1138> Hmm, does that assert warn on any build?
22:10:48 <peter1138> _dp_, well, I did my bit :p
22:15:24 * peter1138 attempts to play Brutal Doom
22:17:31 *** Supercheese has joined #openttd
22:20:11 *** andythenorth has joined #openttd
22:22:11 <peter1138> Playing it on hard realism mode, heh.
22:22:47 <peter1138> Why does this assert not need a cast with my system headers?
22:24:09 <peter1138> Must be some magic, it warns if I cast it to int, as it is meant to be.
22:26:49 *** Wormnest has joined #openttd
22:28:10 <dwfreed> peter1138: what's the line? and what's the types?
22:28:23 <peter1138> src/network/core/address.cpp:271
22:28:31 <peter1138> size_t vs socklen_t
22:31:12 <glx> I have unsigne long long vs size_t here
22:33:12 <nielsm> the macos slowness has the same symptoms as what I was seeing on the dos build
22:33:21 <nielsm> i.e. no single component measures slow
22:33:47 <dwfreed> socklen_t is uint32_t equiv
22:34:09 <peter1138> dwfreed, no, it's not unsigned.
22:34:21 <dwfreed> according to my headers it is
22:34:36 <dwfreed> ./types.h:__STD_TYPE __U32_TYPE __socklen_t;
22:34:43 <dwfreed> ./socket.h:typedef __socklen_t socklen_t;
22:35:06 <_dp_> nielsm, that's probably citymania client screenshot as there are no vehicles
22:35:11 <_dp_> nielsm, they still can be slow
22:35:33 <_dp_> nielsm, though they should be part of gameloop iirc
22:35:42 <nielsm> yes and this is a blank game as he's writing
22:35:50 <nielsm> there is nothing there to make it slow
22:35:56 <nielsm> it's not simulation related
22:36:10 <nielsm> it seems like it's something with the timing of the main loop
22:36:29 <nielsm> like the timer used is wrong
22:36:44 <andythenorth> so RV overtaking is solved, in the de-facto replacement for OpenTTD?
22:36:50 <nielsm> since it's not slow simulation, it's not slow rendering, it's not slow output
22:37:04 <glx> graphics rendering is low but doesn't seem fast in average
22:37:31 <nielsm> glx: it's not 120 ms slow
22:37:58 <glx> but indeed it's fast enough for 33fps
22:38:01 <nielsm> it should be able to output 600-800 fps at that rate
22:38:51 <peter1138> I'd assume andythenorth would have reported such slowness if it was universal on OS X, though.
22:41:09 <andythenorth> eh what did I break now?
22:41:15 *** Thedarkb-T60 has joined #openttd
22:41:16 <nielsm> maybe has to do with hardware configuration, which system timers are available or something like that
22:41:32 <nielsm> you never reported getting about 7 fps
22:42:50 <andythenorth> ok so I have a busy test game open
22:43:02 <andythenorth> where FFWD performance is a dog (as a reference point)
22:43:32 <andythenorth> but I'm getting 60-70fps for both simulation and graphics
22:45:34 <peter1138> andythenorth, in this "bug" it's apparently really slow with an empty game.
22:46:23 <andythenorth> ok on a clean map, no vehicles, no AI
22:46:40 <andythenorth> simulation rate is regularly passing 2k fps
22:46:49 <nielsm> are you running fullscreen or windowed mode?
22:47:24 <glx> on the thread it says it's better with smaller window
22:47:59 <LordAro> could it be some horrifically underpowered macbook air or something?
22:48:13 <LordAro> i don't think there were any actual hardware specs in that thread
22:48:22 <peter1138> It's said that previous versions worked fine.
22:48:23 <LordAro> andythenorth: OS version?
22:48:30 <peter1138> And it'd have to be VERY unpowered to be that slow.
22:48:49 <nielsm> if it was just underpowered hardware then the slowness would probably measure to one or more of the framerate figures
22:48:50 <andythenorth> full disclosure, this is a self-compiled version of the NRT PR
22:48:57 <glx> iMac (Retina 5K, 27-inch, 2017) 3,8 GHz Intel Core i5, 24 Gb 2400 MHz DDR4, Radeon Pro 580 8 Gb Mohave 10.14.4
22:49:06 <andythenorth> also I've been running OpenTTD on macs for 12 years
22:49:06 <glx> doesn't seem underpowered to me
22:49:15 <LordAro> mm, both posts in there mention 10.14
22:49:17 <andythenorth> they used to be more underpowered than now :P
22:49:34 <andythenorth> 10.14 might have changed the graphics API to default to Metal
22:49:38 <andythenorth> among other changes
22:49:46 <andythenorth> I can't upgrade, not supported at work
22:49:55 <LordAro> when were you plan- aw
22:50:08 <glx> oh so all the deprecated stuff is now slow ?
22:50:54 <LordAro> still doesn't explain why 1.8 was fine?
22:51:03 <LordAro> unless they upgraded at the same time and didn't test both
22:51:46 <glx> or it's the way I built the previous versions
22:52:23 <glx> 1.8 used the old compile farm
22:52:28 <LordAro> we need someone with 10.14 able to test such things :)
22:52:46 <andythenorth> I believe we have one in the forums? o_O
22:53:08 <glx> yeah probably need him to build 1.8.0 to compare
22:53:25 <nielsm> can we somehow get 1.8.0 built on the new compile farm?
22:53:33 <LordAro> come up with a list of things for them to try
22:53:43 <LordAro> nielsm: with difficulty, i fear
22:55:01 <nielsm> one guess at a possible reason
22:55:21 <nielsm> though it should not give up a time slice for so long
22:55:26 <nielsm> and should not busy-wait
22:55:48 <LordAro> can we just use <chrono> stuff everywhere?
22:55:55 <peter1138> LordAro, we do now.
22:56:31 <LordAro> the video drivers aren't using it to control their loop times, is it?
22:56:38 <peter1138> Oh, I was thinking of the CSleep() stuff. Oops.
22:56:55 <peter1138> I have a patch for that, though... (lol)
22:57:33 <peter1138> Oh, didn't touch OS X though.
22:57:53 <glx> I think it should be possible to do 1.8.0 build like it's done for PRs
22:58:15 <LordAro> all the version stuff would be broken, i imagine
22:58:29 <LordAro> but it would probably serve its purpose
22:58:30 <peter1138> Yeah, it's missing the pipelines stuff.
22:58:59 <glx> using the nightly pipeline
22:59:28 <LordAro> so releases/1.8 + azure-pipelines stuff? could work
22:59:29 <nielsm> where the heck is CSleep defined in 1.9 branch?`
23:01:41 <nielsm> okay os/unix/unix.cpp it seems
23:01:46 <nielsm> and it does usleep(ms*1000)
23:07:59 <glx> ok I don't see how to queue it
23:12:04 <nielsm> it has to be possible to unify this somehow?
23:12:09 <glx> ok it's not possible because "An error occurred while loading the YAML build pipeline. Could not find /azure-pipelines-release.yml in repository self hosted on https://api.github.com using commit ebbbf0bdfb4abd849bb1f91254b60687a9660bfc. GitHub reported the error, "Not Found""
23:12:20 <nielsm> nm that that's from 1.9 branch it's basically unchanged in master
23:13:35 <nielsm> could we somehow make a SDL build for macos? or an allegro build?
23:14:07 <LordAro> nielsm: i'd like to see them merged, for sure
23:15:32 <glx> not easy LordAro, each driver use a different way to get system events
23:16:25 <nielsm> well some kind of templated stuff or whatever
23:16:30 <LordAro> sure, each driver would need a HandleEvent() function or whatever, but the loop itself and fair bit of the before, after, and contents could do it
23:16:49 <nielsm> that weird pattern where a superclass takes its subclass as a template parameter to get static polymorphism
23:17:05 <glx> yes before and after for(;;) can probably be factorised
23:17:21 <LordAro> nielsm: also pls avoid
23:17:21 <nielsm> the basic goal would be to keep the timing logic the same for all platforms
23:19:02 <glx> hmm or not, because all inter thread communication on windows uses windows specific stuff
23:19:40 <nielsm> the windows driver has received lots of tender loving care over the years
23:20:03 <glx> osx driver on the other hand ;)
23:20:41 <glx> when converting QSortT() I had to keep OSX different
23:21:19 <glx> it's still limited to 32 resolutions while all other drivers have "infinite"
23:23:16 <nielsm> I have a (by now ancient) macbook pro around but its disk died so I'd have to get a new SSD for it and somehow install OS
23:23:25 <nielsm> and don't know if it actually works anyway
23:24:03 <nielsm> or could do the hackintosh -.-
23:24:14 <nielsm> (but no guarantee it'd exhibit the bug)
23:25:12 <milek7> funny comment on emscripten build: "I am thoroughly impressed at how smooth this is running on MacOS! It seems way more smooth than the native build for some reason."
23:26:35 <glx> hmm maybe split mainloop in preloop where we may start thread, mainloop, and postloop with thread cleanup if needed
23:28:07 <glx> but won't change much as the loop itself is full of driver specific code
23:28:22 <glx> for event polling and key reading
23:29:13 <nielsm> can we make some template function that takes a bunch of function parameters and does the timing magic and calling of the game mainloop, but lets the driver fill in the other specifics?
23:31:53 <glx> oh and of course drawing on osx is not threaded
23:33:25 <glx> and I guess it was the same for dos
23:33:46 <glx> that may be the common point
23:34:15 <nielsm> yeah the allegro and macos mainloops are very similar
23:34:41 <glx> all painting is done outside main thread for sdl and windows
23:34:56 <milek7> --without-threads works fine for sdl
continue to next day ⏵
