IRC logs for #openttd on OFTC at 2019-12-22
⏴ go to previous day
00:07:08 <Samu> i just noticed I'm ranked #27 and #40 at the same time
00:10:12 *** supermop_work has joined #openttd
00:28:13 <nielsm> I'm not sure I want to PR it, as it can probably kill performance on certain games
00:28:36 <glx> and it won't measure per grf I guess
00:29:06 <nielsm> yeah that would require some extra logic in the framerate code
00:29:26 <glx> and more performance killing :)
00:29:47 *** andythenorth is now known as Guest12121
00:29:47 *** andythenorth has joined #openttd
00:30:11 <milek7_> performance measuring is so expensive?
00:30:42 <nielsm> but it might be worth it have a separate performance capture system for newgrf callbacks, that logs to a file and only for a specific newgrf
00:30:45 <glx> it's a small overload, but everything adds up
00:31:08 <nielsm> and could possibly even separate out by toplevel spritegroup or some such
00:31:53 *** andythenorth has joined #openttd
00:32:59 <nielsm> e.g. a console command you give a loaded grf index and a duration in seconds (or maybe number of ticks or game days) to capture for, and it adds up stuff and writes a csv file to the screenshots directory
00:38:46 *** peter1138 has joined #openttd
00:38:46 *** ChanServ sets mode: +o peter1138
00:40:13 <Samu> i'm off to bed, take care
00:40:51 *** supermop_work has joined #openttd
01:02:12 <Xunie> Can I create a 'ping pong' order list?
01:02:29 <Eddi|zuHause> how do you mean?
01:02:35 <Xunie> That is: An order list will be executed in sequence and then in reverse sequence?
01:02:42 <nielsm> you mean going A-B-C-D-C-B-(A)
01:03:04 <nielsm> only by putting all the orders and the reverse manually
01:03:17 <nielsm> there isn't any feature to automatically "mirror" an order list
01:03:19 <Xunie> Ouch, I don't wanna do that.
01:03:30 <Xunie> I know implicit orders are a thing, but you can't control road geometry.
01:03:51 <Xunie> Damn shame if you ask me, nielsm.
01:03:55 <Eddi|zuHause> no, you can't do that
01:04:08 <Eddi|zuHause> i agree that it would be nice
01:04:26 <Eddi|zuHause> but the order list is already overloaded with features, and needs some cleaning up
01:04:38 <Xunie> Yeah, that window was really confusing.
01:04:44 <Xunie> And moving orders around is a little iffy too.
01:04:51 <Xunie> You can't move an order around if you have it selected.
01:11:33 *** supermop_work has joined #openttd
01:40:29 <Pikka> Xunie: the easiest way to create those kind of orders (eg for ships with a long list of bouys) is to add each one twice, and click on the second one before adding the next one so the next bouy goes in above it. If that makes sense. :)
01:40:54 <Xunie> You just insert in the middle.
01:42:10 *** supermop_work has joined #openttd
01:43:00 <Xunie> I agree that while the OpenTTD GUI does look absolutely splendid, actually learning to use it is a lot of pain.
01:50:05 *** Wormnest has joined #openttd
02:12:58 *** supermop_work has joined #openttd
02:18:50 <Xunie> Are timetables even worth it?
02:19:05 <Xunie> It seems like so much hassle for little gain.
02:22:25 <Eddi|zuHause> depends how far you want to push it
02:22:43 <Eddi|zuHause> they're very useful for trams, to keep them separated on the line, instead of clumping up
02:24:54 <Xunie> Yeah, well. I'm the kind of guy that ctrl-clicks "clone vehicle" in a road depot about 20 times and let the traffic jams sort it out.
02:25:20 <Xunie> If your competitors have traffic jams? So will you! :^)
02:27:14 <Eddi|zuHause> the problem is, if tram 1 is at the station, it loads full, and leaves, tram 2 loads too, but has less to load, so leaves earlier
02:27:47 <Xunie> Yeah, there's a bit of a 'cascading' effect, but for some reason, I rarely see that in my gameplay.
02:27:54 <Xunie> It just 'averages' out eventually?
02:28:16 <Xunie> Maybe I'm playing wrong then?
02:31:09 <glx> in late game your stations are overcroweded
02:32:35 <glx> whatever you try it's almost impossible to fix because usually there's no space left around
02:35:58 <Xunie> This is the part where you all call me a scrub for never reaching late game in the first place. :^)
02:43:31 <Eddi|zuHause> we don't discriminate against different play styles here
02:43:35 *** supermop_work has joined #openttd
02:44:30 <Xunie> Man, I've never played the original TTD, but from what I can tell, OpenTTD is a well designed game...
02:45:00 <Xunie> Eddi|zuHause: regardless of your play style or competence level. OpenTTD is mighty fun.
02:54:07 *** Wormnest has joined #openttd
03:15:46 *** supermop_work has joined #openttd
03:19:21 <floodious> i suspect it isn't too fun if you prefer growing shrubberies and building fountains and park paths
03:19:43 <floodious> you need sim park garden for that
03:25:05 <FLHerne> Xunie: From most concepts of "balance", OpenTTD is an awfully-designed game
03:27:01 <FLHerne> With default settings, any halfway-good player has practically infinite money for most of the game
03:27:14 <FLHerne> There's no sane way to spend it all
03:28:57 <nielsm> maybe I should measure in nanoseconds instead of microseconds...
03:28:57 <floodious> it's not difficult, that's a very old complaint against TT from the 90s, but it's not supposed to be difficult... it's supposed to be like playing with lego
03:29:17 <floodious> the hard part is using your imagination and engineering without needing any goal or purpose
03:30:14 <floodious> realism actually often ruins the game... like i'm working currently on a realistic scenario with accurate costs, setting and industry from the 1850s in the pacific northwest
03:30:28 <floodious> 90% of the time i go bankrupt after an hour
03:30:58 <floodious> there is no option for where to place tracks... they can only go in the historic locations and minimizing cost is incredibly difficult
03:31:52 <floodious> changes the game from fun playtime in the sandbox to insanely stressful "there is no way we can afford this, let's give up"
03:32:13 <FLHerne> Interesting, I've never quite managed to achieve that in OTTD
03:32:44 <floodious> well, "working" AKA "no way we can afford this, let's give up" :)
03:32:56 <floodious> a real life version of the scenario is already playing out on my PC
03:33:02 <FLHerne> The mechanics really don't work well to create stable economics
03:33:31 <FLHerne> You're either making or losing money, and that rapidly turns into a feedback cycle
03:34:13 <floodious> nope, i'm still working on modifying the heightmap import (using color layers = forests + rivers/lakes) and configuring accurate newgrfs will probably require a ton of customization, some might require code mods
03:34:39 <floodious> the game isn't set up to allow the kind of feedback processes that make such realism possible at all
03:35:25 <floodious> good example is to create flooded areas and use barges in the past they simply tossed gravel and rock in a narrow gap and waited
03:35:46 <floodious> or to "clear" a valley the opposite, build a dam, flood it, then use explosives to crack it and wash out the valley
03:37:04 <floodious> the "f*** the environment!" oldschool method
03:39:30 <nielsm> (sprite number of the root action 2 in the grf file in rows, how many recursive calls were required to resolve the action 2 in columns, count of instances that sprite took that many recursive steps to solve as data)
03:46:26 *** supermop_work has joined #openttd
04:01:10 *** Beerbelott has left #openttd
04:01:22 <nielsm> oh, should maybe have made that as a draft PR
04:17:08 *** supermop_work has joined #openttd
04:28:25 <Eddi|zuHause> hm, if astroneer doesn't win the steam awards for best visual style, i don't know anymore
04:47:49 *** supermop_work has joined #openttd
05:04:21 <floodious> lots of noise, it would probably look better if all the rivers creeks were plotted (i manually drew the top-left river system) and the combined computed river + slope watertable data were used to plot forest density
05:04:42 <floodious> i added the noise to approximate it without having data for all the watercourses
05:05:10 <floodious> obviously the "empty" areas at the bottom of valleys would get filled by the saturation from the riversystems
05:06:11 <floodious> i'm thinking 127 levels = 127 gray, 127 gray+green, 1 black (sea), 1 blue (river/lake)
05:06:51 <floodious> but that requires filling the lake height from the edges, so makes more sense to have separate water + forest maps
05:08:06 <floodious> funny thing is the modern heightmap data includes all the man-made structures :)
05:08:24 <floodious> like "that island wasn't there" or "that bay wasn't there"
05:15:18 *** ToBeCloud is now known as ToBeFree
05:15:22 *** ToBeFree is now known as ToBeCloud
05:18:26 *** supermop_work has joined #openttd
05:49:04 *** supermop_work has joined #openttd
05:49:53 *** ToBeFree has joined #openttd
06:19:40 *** supermop_work has joined #openttd
06:50:21 *** supermop_work has joined #openttd
07:17:17 *** HerzogDeXtEr has joined #openttd
07:21:03 *** supermop_work has joined #openttd
07:51:42 *** supermop_work has joined #openttd
08:22:19 *** supermop_work has joined #openttd
08:52:57 *** supermop_work has joined #openttd
08:55:03 *** Andrew350 has joined #openttd
09:23:42 *** supermop_work has joined #openttd
09:23:45 *** andythenorth has joined #openttd
09:54:18 *** supermop_work has joined #openttd
10:07:46 *** andythenorth is now known as Guest12145
10:07:46 *** andythenorth has joined #openttd
10:20:49 <DorpsGek_III> [OpenTTD/OpenTTD] planetmaker approved pull request #7867: Update: Prepare for 1.10.0-beta2 release https://git.io/Jed7f
10:24:57 *** supermop_work has joined #openttd
10:30:58 <andythenorth> I got myself a 3m line csv :)
10:31:08 <andythenorth> cb ID 0 is the graphics chain?
10:31:31 <nielsm> yeah that's generic graphics
10:34:18 <andythenorth> Pikka: ok, moar!
10:36:30 <nielsm> there's the callback ids
10:38:48 <andythenorth> ^ I have to fit 4 more pickup stations into that area :D
10:38:54 <andythenorth> without blocking
10:39:56 <nielsm> move the cargo tram underground?
10:40:34 <nielsm> if it isn't against your aesthetics
10:42:56 <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/Jed7q
10:44:39 <nielsm> hm, excel 2013 (and later) on windows has no trouble opening gigantic files, it handles up to 2G or 4G rows
10:45:08 <nielsm> but yeah a serious analysis tool (numpy, R, whatever) is probably the best
10:46:07 <andythenorth> yeah mac excel stops at 104857
10:47:22 <andythenorth> you have been busy :D
10:48:00 <nielsm> oh... right excel 2013 also does max out at 1048576
10:49:18 <Andrew350> andythenorth: I do now :)
10:49:55 <Andrew350> not sure what it all means though
10:51:11 <nielsm> would a tool to read those profiling csvs be appropriate as part of NML? so it could perhaps also resolve sprite numbers back to names
10:52:01 <nielsm> and/or nmlc should have a mode to output an additional file with symbol mappings
10:54:47 <andythenorth> nielsm: it would be appropriate, I'm not sure it's essential :)
10:55:35 *** supermop_work has joined #openttd
10:55:39 <andythenorth> it's good to shine light into dark corners :)
10:56:19 <andythenorth> Andrew350: frosch made the spec status to keep track of all the recent newgrf changes which aren't properly finished
10:56:43 <andythenorth> we have been filling it in a few items at a time
10:57:01 <andythenorth> most of it is NRT, so your wiki page is really helpful thanks
11:00:45 <DorpsGek_III> [OpenTTD/OpenTTD] planetmaker commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/Jed78
11:14:06 <Andrew350> andythenorth: most everything on that page is a copy of railtypes with irrelevant stuff removed, plus info transcribed from frosch's draft specs from like 2 years ago
11:14:52 <Andrew350> I've tried reading the nml source to gain some info on the updated specs, but most of it is alien language to me
11:15:33 <planetmaker> nice. I'd say: just edit the "real" wiki page(s) :)
11:15:59 <Andrew350> yeah, but I don't know the specs, so don't want to put lies out there :)
11:16:38 <Andrew350> hopefully someone with knowledge can fill in the blanks
11:17:24 <Andrew350> I did update the other docs to link everything properly :)
11:23:23 <Andrew350> a few specific things I could use clarification on:
11:23:49 <Andrew350> - how ID allocatiion works (does my description make sense?)
11:24:17 <Andrew350> - any flags that were added and what they do
11:24:58 <Andrew350> - anything else like callbacks or whatnot which have changed since the initial draft specs
11:26:15 *** supermop_work has joined #openttd
11:29:21 <Andrew350> on a related note I did notice in the NML source in action0.py it says roadtypes and tramtypes get 15 IDs each, which sounds wrong since I'm pretty sure I remember seeing something that both share 64 IDs between them
11:38:06 <andythenorth> the nml patch was not very good quality
11:38:15 * andythenorth apologises to all
11:42:03 <Andrew350> lol not trying to make you feel bad :P just wondering if it's right or not
11:47:47 <nielsm> andythenorth: do you think adding the sprite number of the callback result would be useful too?
11:48:30 <planetmaker> Andrew350, I didn't check it now extensively. But even a wiki can be edited and amended/corrected :)
11:48:45 <andythenorth> nielsm: I don't know that I'd have a use for it :)
11:49:34 <nielsm> andythenorth, possibly whether certain results have longer resolution chains than others
11:50:07 <nielsm> or some results are more common and could maybe be special-cased earlier on so they resolve faster than the unusual cases
11:50:18 <andythenorth> it could be useful
11:50:41 <andythenorth> my first question though is, "are callbacks even slow"
11:51:01 <andythenorth> the last profiling info I recall reading was about sprite resolution
11:51:16 <andythenorth> getimage or something like that was relatively expensive
11:51:16 <nielsm> it's the same thing actually
11:51:28 <andythenorth> there's a forum post
11:51:41 <DorpsGek_III> [OpenTTD/OpenTTD] James103 opened issue #7869: The human-readable name of a cargo type can't be queried by AI/GS. https://git.io/Jed7d
11:51:43 <nielsm> getting a sprite always involves an action 2
11:52:08 <nielsm> and if it resolves directly to a realsprite, then that's a zero-depth callback chain
11:55:39 <Andrew350> planetmaker: that assumes someone is willing to check whether it is right and correct it ;)
11:57:03 <planetmaker> we assume that it is correct unless proven wrong. In which case... who finds a mistake is obliged to correct it ;)
11:57:12 <planetmaker> better than no documentation :P
11:57:21 *** supermop_work has joined #openttd
12:00:57 *** snail_UES_ has joined #openttd
12:04:17 *** Progman has joined #openttd
12:23:25 *** sla_ro|master has joined #openttd
12:28:03 *** supermop_work has joined #openttd
12:33:12 <Andrew350> ok, as per planetmaker's request i've just gone ahead and made my drafts the 'official' pages for both roadtypes and tramtypes. no going back now :P
12:43:52 *** frosch123 has joined #openttd
12:46:14 <frosch123> planetmaker: the goal of the api review was to make sure that the docs are actually correct
12:46:20 <frosch123> wrong docs is the worst you can have
12:47:04 <frosch123> esp. if noone knows what is actually correct
12:50:58 <Andrew350> oh i just found the flags available for nrt in the nml files. neat :) anyone know how the ROADTYPE_FLAG_TOWN_BUILD works? is it based on speed or something else?
12:57:26 <nielsm> I want to tell andy that in e.g. FIRS 3, the lookup for sprites can go as much as 11 deep and take more than 50 microseconds to resolve in really bad cases
12:57:40 <nielsm> (in this debug build, obviously would be faster in release builds)
12:58:41 *** supermop_work has joined #openttd
13:01:55 <DorpsGek_III> [OpenTTD/website] TrueBrain opened pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
13:05:04 <frosch123> nielsm: is the goal explicitly to only measure one grf?
13:08:59 <nielsm> well I suppose it could be set up to profile more than one, it's just making an interface for it
13:09:17 <nielsm> thing is it collects a ton of data that needs to be crunched too :)
13:10:40 <frosch123> yes, externally crunching is certainly better than complex ingame guis
13:11:24 <nielsm> slightly annoying for analysis that industry production uses cb id 0
13:11:49 <nielsm> and I'm not sure if it's safe to change to invent a new cb id
13:11:51 <frosch123> add the feature number then
13:12:03 <frosch123> no, changing the cbid is definitely wrong :)
13:12:17 <frosch123> but adding columns for feature and item id is probably good
13:12:36 <nielsm> I don't think I have the feature and item available in this context
13:13:13 <nielsm> at least not without special casing for the different ResolverObject subclasses
13:15:47 <nielsm> ah no wait, it's the ScopeResolver I need to look at isn't it
13:16:35 <frosch123> VehicleResolverObject::VehicleResolverObject constructor has the data, but it does not store stuff in REsolverObject
13:17:34 <DorpsGek_III> [OpenTTD/website] James103 opened issue #117: There's a couple of layouting/responsiveness issues in the OpenTTD website. https://git.io/Jed55
13:19:22 <frosch123> i should look at the old diff, how it was done
13:22:47 <frosch123> it's even the global item id, not the grf-local item id
13:23:45 <nielsm> so it's kind of useless isn't it
13:24:16 <nielsm> but feature ID is probably the most important to include
13:24:37 <frosch123> local-id is also available
13:24:55 <frosch123> all engines/industries... have a GRFFilePropsBase<> member
13:25:23 <frosch123> that has the grffile (i.e. grfid) and local_id
13:29:02 <DorpsGek_III> [OpenTTD/website] glx22 commented on pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5h
13:29:22 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh dismissed a review for pull request #7868: Feature: NewGRF callback profiling https://git.io/Jed7v
13:29:38 *** supermop_work has joined #openttd
13:37:27 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 requested changes for pull request #7868: Feature: NewGRF callback profiling https://git.io/JeddG
13:38:01 <frosch123> oh, there was an update meanwhile.. i wonder what gh did in that case
13:45:12 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/Jeddl
13:45:48 <DorpsGek_III> [OpenTTD/website] auge8472 commented on issue #117: There's a couple of layouting/responsiveness issues in the OpenTTD website. https://git.io/Jed55
13:58:20 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
13:59:58 *** supermop_work has joined #openttd
14:00:43 *** andythenorth has joined #openttd
14:25:05 <Wolf01> Yeah, crashed TF2 with a truck station
14:30:36 *** supermop_work has joined #openttd
14:40:02 *** snail_UES_ has joined #openttd
14:41:56 *** andythenorth has joined #openttd
14:43:51 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick dismissed a review for pull request #7424: Add: Kdtree for AirportGetNearestTown https://git.io/Jei0p
14:44:21 <Samu> FOR_ALL_TOWNS disappeared
14:53:27 <andythenorth> Eddi|zuHause: ultimate rick roll thanks :|
14:58:38 <Samu> #7094 is not yet approved :(
15:01:19 *** supermop_work has joined #openttd
15:01:42 *** andythenorth has joined #openttd
15:05:41 <Samu> if it's approved, it makes #7515 irrelevant
15:07:33 *** hythlodaeus has joined #openttd
15:07:45 <Samu> it also makes commit 2 and 3 of #7486 irrelevant
15:07:54 <hythlodaeus> question: is there a tag for inserting non-breaking spaces on text strings?
15:09:19 <frosch123> easy question, easy answer :)
15:10:00 <glx> I use it a lot in french translation :)
15:11:13 <Samu> #7376 has conflicts, grr :( brb
15:14:33 <glx> I think I'll close #7515 in favor of #7661
15:22:12 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on pull request #7515: Fix: Really check for start_date once an AI company slot becomes available. https://git.io/Jedbv
15:22:12 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 closed pull request #7515: Fix: Really check for start_date once an AI company slot becomes available. https://git.io/fjYeA
15:23:45 <Samu> FOR_EACH_SET_BIT - was this one also removed?
15:24:01 * andythenorth mistakes were made
15:24:12 * andythenorth has to drive 90 miles due to forgotten things :P
15:24:34 <nielsm> hope they play something good on the radio
15:24:51 <andythenorth> @calc 1.6034 * 90
15:24:51 <DorpsGek> andythenorth: 144.306
15:25:09 <andythenorth> eh glx I found some actual redundant switches you could optimise away
15:25:49 <glx> like same expr, same results ?
15:26:06 <glx> and of course same feature/scope ?
15:26:16 <andythenorth> I'll find an example
15:26:32 <glx> but checking that won't be easy I think
15:26:34 <andythenorth> it's something I could solve in the compile, just makes the compile a bit stringy
15:27:20 <glx> or maybe in the pre_process, comparing with already declared switches
15:27:51 <andythenorth> so all my code is generated from templates / macros
15:27:55 <andythenorth> often with loops
15:28:11 <andythenorth> sometimes the loop is over an iterable with only have one member
15:28:18 <andythenorth> which results in a switch with one result
15:29:12 <glx> ah yes I didn't try to optimise random switches yet
15:29:41 *** supermop_Home has joined #openttd
15:29:47 <andythenorth> obviously there are cases with > 1 result too
15:30:23 <glx> if all results are the same the first step is to optimise to 1 result
15:30:33 <nielsm> 800 milliseconds spent across 30 game days on looking up industry graphics
15:30:39 <glx> then on reference resolution it can be replaced
15:30:56 <nielsm> 1.5 milliseconds spent running production and production change callbacks
15:31:08 <andythenorth> I *can* fix it in templating, but it causes the templates to have a lot of logic
15:31:16 <andythenorth> or I can abstract it, but then there's a lot of abstraction
15:31:30 <glx> optimisation is the role of the compiler
15:31:46 <glx> it will warn a lot, but heh
15:31:51 <nielsm> glx: doesn't absolve the developer from writing better code ;)
15:31:55 <andythenorth> oh the paste misses the switch that is the entry point to lemon_switch_graphics_vehicle
15:31:59 *** supermop_work has joined #openttd
15:32:10 <glx> the warnings are there for that nielsm :)
15:32:32 <andythenorth> nielsm: so are those numbers good or bad? :)
15:32:44 <nielsm> andythenorth I'd say not good
15:33:06 <andythenorth> the 1.5ms seems ok
15:33:27 <supermop_Home> should a 1.10 title game include a roadtype?
15:33:30 <nielsm> set up by number of chained action 2's required for resolution
15:33:48 <glx> hmm aren't duplicate spritesets already handled ?
15:33:50 <nielsm> most industry tiles require 10 levels of resolution
15:34:01 <supermop_Home> that is, if nrt is in, should there be some vanilla use of it?
15:34:17 <nielsm> the title game has traditionally need newgrf free
15:34:23 <glx> oh they are not duplicates
15:34:30 <nielsm> and NRT only does something with newgrfs present
15:35:40 <glx> handling of duplicates will need a mapping table I guess
15:36:35 <supermop_Home> nielsm it always bothered be that there were base set tramway but no baseset trams
15:36:37 *** andythenorth has joined #openttd
15:37:05 <supermop_Home> don't think I see you here that much
15:37:19 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick updated pull request #7376: More than max_no_competitors could be created in network games https://git.io/fjvY8
15:37:25 <supermop_Home> though must be early morning for you
15:39:23 <supermop_Home> i guess i better brush off my nml and get my nrt stuff polished before regular people can see it
15:42:21 <Andrew350> supermop_Home: yeah I'm almost never here, didn't feel much like sleeping tonight though so I'm around working on NewGRF docs through the night :)
15:45:33 <andythenorth> 'refit any available' works surprisingly well
15:46:12 * andythenorth has trains carrying 5 cargos to one big transfer
15:46:25 <supermop_Home> andythenorth i used it a lot my last steeltown game
15:46:46 <andythenorth> it's helpful with all these output cargos eh
15:47:04 <supermop_Home> about to start a new game
15:48:46 <glx> andythenorth: if you can provide me a small working example to be optimised it would be nice
15:50:52 <andythenorth> I enjoy the bold colous
15:50:56 <glx> and if it includes things I already optimised in my PR it would be a nice regression test
15:51:34 <Pikka> I'll draw those houses properly one of these days...
15:53:57 <andythenorth> glx: does your local nml include the procedures?
15:54:53 <glx> that will be something to optimise too I guess
15:55:25 <glx> but switch optimisation should handle them
15:57:53 <andythenorth> contains two engines
15:58:02 <andythenorth> one has redundant random switches, one doesn't
15:58:33 <andythenorth> I've left all the other global stuff in place, so it's not a small nml :P
15:59:04 <andythenorth> also the code is generated, so the whitespace is horrible
16:01:15 <andythenorth> you can observe all the spriteset template bollocks for flipped / not flipped (depot flip) :D
16:01:34 <glx> hmm ok small example 1500+ lines :)
16:02:02 <andythenorth> 1000 lines of spriteset templates :|
16:02:30 <andythenorth> no only 918 lines of spriteset templates
16:02:35 <andythenorth> calm down Andrew350
16:02:40 *** supermop_work has joined #openttd
16:03:02 <andythenorth> yes, I should have run make clean :)
16:04:10 <andythenorth> oops I highlighted the wrong person, normally 'an' gets a completion on my name :P
16:04:48 <Andrew350> andythenorth: was about to ask wtf :D
16:05:13 <andythenorth> glx: if you sort the nml dir by filesize, you can see where the bloat is :)
16:05:28 <andythenorth> supporting visible cargos is very bloat
16:07:35 <Andrew350> andythenorth: did some more docs updates though, working through that newgrf specs list you linked too. finished the railtype stuff :)
16:07:47 <andythenorth> this really helps
16:08:38 <andythenorth> if you're not 100% maybe put 'draft', otherwise 'done'
16:08:38 <Andrew350> already been doing it :)
16:08:50 <glx> andythenorth: I'll remove all the stuff I don't need from the zip ;)
16:09:08 <andythenorth> I gave you a free bonus nfo
16:09:55 <glx> yeah useful for before/after comparison
16:11:13 <glx> but seems I will regenerate it without all the unneeded translations
16:12:11 <andythenorth> I can strip it down more, I made a branch
16:12:32 *** HerzogDeXtEr has joined #openttd
16:13:51 <glx> I'll do the tests on this branch then
16:14:54 <andythenorth> it's pretty easy to turn vehicles on and off, but cleaning the translations is a manual task :P
16:15:22 <glx> only relevant translations are included in nfo I think
16:16:05 <Samu> void OnQueryTextFinished(char* str) override
16:16:24 <Samu> that * was auto-moved, how do i stop visual studio from doing that?
16:16:30 <Eddi|zuHause> yeah, nmlc filters the strings that are actually used
16:17:13 <glx> Samu: I always fix that bu hand
16:17:15 <andythenorth> the compile only includes the vehicles that are enabled
16:17:27 <glx> but yes there are settings available
16:22:18 <nielsm> in fact, go through all the formatting settings for C++ and you can get the ottd guidelines implemented almost fully automatic :)
16:22:53 <glx> ok first run render graphics fail (division by 0), second run render docs fail (IndexError: list index out of range), third run render nml fail (ValueError: min() arg is an empty sequence)
16:24:16 *** andythenorth has joined #openttd
16:30:11 <glx> andythenorth: I need a working version ;)
16:30:29 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick updated pull request #7661: Codechange: Rework 'start_date' parameter for AIs as a game setting https://git.io/fjMsZ
16:30:45 <glx> <+glx> ok first run render graphics fail (division by 0), second run render docs fail (IndexError: list index out of range), third run render nml fail (ValueError: min() arg is an empty sequence) <-- in case you missed it
16:31:10 <andythenorth> oh that's interesting
16:31:21 <andythenorth> I did run make before posting it :)
16:31:27 <glx> oh and fourth run fails too
16:31:42 <glx> nmlc ERROR: Empty input file
16:33:18 *** supermop_work has joined #openttd
16:36:16 <andythenorth> yeah that repros
16:36:23 <glx> there should be 2 based on the diff
16:38:05 <andythenorth> does master build?
16:39:48 <andythenorth> platform specific issues?
16:39:58 <andythenorth> bad use of __main__ ?
16:40:27 <glx> same trace, division by zero
16:41:41 <andythenorth> I have to go for a couple of hours
16:41:51 * andythenorth had logistical error involving christmas presents
16:43:29 *** andythenorth has joined #openttd
16:45:52 <TrueBrain> GitHub Deployments API is pretty nice .. took some effort to understand, but I can automate that now too via GitHub Actions :)
16:46:10 <TrueBrain> except for the real deployment, that now works :D
16:46:34 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
16:48:33 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
16:49:12 <TrueBrain> if this really works, it means we no longer need Azure Pipelines for it, and everything is in a single place :)
16:50:44 <TrueBrain> well, not sure about that tbh :P It is as complicated as Azure is :)
16:50:57 <TrueBrain> I might rewrite a few things into custom actions, just to make it appear a bit easier
16:54:52 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
16:57:38 *** snail_UES_ has joined #openttd
17:00:34 <hythlodaeus> guys would anyone like to test a 1st version of my amended english tooltips?
17:01:38 <glx> hythlodaeus: maybe just submit a draft PR, will be easier to check your WIP
17:02:11 <hythlodaeus> i'm not so good with github. can you tell me how can I do that after cloning the repo?
17:03:42 <glx> first step is to fork openttd on github, then clone your fork on your machine
17:03:56 <glx> then you add a branch and switch to it
17:03:57 *** supermop_work has joined #openttd
17:04:25 <glx> do your changes, commit, push to your fork
17:04:57 <glx> then you can create a PR on openttd and select draft as the type
17:05:28 <nielsm> it's a lot of initial set-up, it gets faster for your second PR ;)
17:05:58 <nielsm> and if you bake it into your development workflow everything gets easier
17:07:17 <hythlodaeus> I changed a bunch of tooltips on the main toolbar I might add. I standardized it to most modern gaming convention
17:07:50 <hythlodaeus> e.g. the icon previous named "save game, load game, abandon game" is now simply named "Game Menu"
17:09:28 <Samu> i really hope #7670 gets fixed before 1.10 release
17:14:43 <Samu> because it's a common occurrence with AIs
17:34:34 *** supermop_work has joined #openttd
17:38:23 *** WormnestAndroid has quit IRC
17:39:31 *** WormnestAndroid has joined #openttd
18:00:12 *** tokai|noir has joined #openttd
18:00:12 *** ChanServ sets mode: +v tokai|noir
18:05:13 *** supermop_work has joined #openttd
18:16:20 <supermop_Home> can towns prefer / upgrade roads now?
18:35:54 *** supermop_work has joined #openttd
18:56:06 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/JedpU
19:04:32 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/Jedpq
19:06:30 *** supermop_work has joined #openttd
19:07:23 <nielsm> frosch123: GetDebugId is okay, I came up with GetItemIdentifier myself, not sure either is spot on
19:09:05 <frosch123> i fixed my spelling to "ID", apparently that's what ottd does
19:11:09 <nielsm> any idea on how to figure out whether a callback is for roadtype or tramtype?
19:11:12 *** andythenorth has joined #openttd
19:11:20 <andythenorth> mystery continues
19:11:47 <glx> nielsm: both use the same feature I think
19:12:26 <nielsm> andythenorth: you make it sound like you just drove 90 miles and did not find the presents you were looking for
19:12:57 <frosch123> glx: they share code, but the features are separate
19:13:15 <frosch123> if there were docs, there would not be so many myths :p
19:14:51 <andythenorth> nielsm: mistakes have been corrected :)
19:24:38 <nielsm> hm, should I hack to make the label for cargo/rail/road/tram types print as ASCII fourCC in the CSV, or leave them decimal?
19:25:20 <frosch123> i think decimal is easier to parse for the tools reading that stuff
19:25:21 <glx> can be handled by parser tools
19:26:04 <frosch123> oh.... but maybe it needs a BSWAP
19:26:06 <nielsm> grfid often contains non-printable characters too, definitely wouldn't ascii-fy those
19:26:25 <frosch123> for some reason ottd saves labels so that they are readible in a debugger, so it converts them to big endian
19:26:56 <frosch123> so maybe all labels needs a BSWAP32 before printing
19:28:25 <nielsm> hm but that swaps unconfitionally doesn't it
19:31:03 <frosch123> you can do it inside GetDebugId
19:31:06 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
19:31:11 <frosch123> so it only affects labels and not ids
19:31:52 <frosch123> glx: what is the status of cmake? are there things to complete? does it need testing?
19:32:07 <glx> it needs 1.10 branching :)
19:32:18 <glx> then the testing will be live :)
19:32:30 <frosch123> oh, you want to target it for 1.11
19:32:41 <frosch123> well, i don't think nrt will be finished before 2030
19:33:32 <glx> will be easier to find broken/missing stuff once it's merged
19:35:20 <TrueBrain> well, 1.10 was scheduled for release ... how many days ago? :D
19:35:28 <glx> I just rebased and added a little thing to force MinGW builds to be static
19:36:26 <TrueBrain> that moment you called it Staging, and GitHub calls it staging .. and things are case sensitive
19:37:08 <glx> windows doesn't care about case :)
19:37:13 *** supermop_work has joined #openttd
19:37:29 <TrueBrain> has nothing to do with OS
19:37:34 <TrueBrain> GitHub just decided to call it one way ..
19:37:40 <TrueBrain> AWS is the case sensitive in this case
19:39:19 <glx> btw I think there's a setting somewhere in windows to activate case sensitivity in NTFS, but nobody ever used it because legacy apps
19:39:48 <TrueBrain> it is used if you use WSL
19:39:59 <TrueBrain> when ever you create a folder via WSL, it is case sensitive
19:40:17 <TrueBrain> it leads to ... unexpected results
19:40:18 <milek7_> meanwhile, it is now possible to have case-insensitive ext4 on linux
19:40:19 <TrueBrain> lets keep it that way :P
19:40:40 <TrueBrain> but okay, how am I going to make the capitals nice here .. ugh ..
19:42:19 <frosch123> case is important for your personal lifestyle. old people use UPPERCASE, middle aged people use lowercase, kids use StuPIDcasE
19:45:49 <DorpsGek_III> - Update: Translations from eints (by translators)
19:45:52 <hythlodaeus> glx: I've just uploaded the english.txt file to my own repo. I'm having trouble finding a the PR draft category when attempting to submit a PR to openTTD main repo
19:46:29 <glx> for the draft you open the create PR window and there's a dropdown
19:49:52 <glx> commit checker won't be happy ;)
19:51:17 <glx> but that can be fixed later
19:51:23 <hythlodaeus> well i wish you told me about it
19:52:08 <glx> oh and you didn't create a branch (not a bug issue if you have only one PR)
19:52:31 <glx> yes you can fix the commit messages with git
19:54:08 <glx> drop the 2 first commits, reword the 3rd
19:55:39 <hythlodaeus> bro i have like no idea wth im doing
20:00:01 <milek7_> or reset --soft to commit before changes and commit once again with correct message
20:00:55 <nielsm> or git commit --amend to edit the last commit message (and add any further changes staged since)
20:03:19 <hythlodaeus> i think i fixed it via github
20:03:39 <hythlodaeus> guys i'm not a programmer, most of the things you're mentioning are greek to me
20:03:50 <hythlodaeus> i'm primarily a tester/linguist
20:04:12 <nielsm> well, someone else can fix the commit history since github is a special beast
20:05:08 <milek7_> PR title is entirely separate from commit message
20:06:02 <nielsm> yeah... and git can be downright user hostile
20:06:49 <nielsm> (it requires a lot of understanding and theory to be capable of using it well past the most basic things)
20:07:51 *** supermop_work has joined #openttd
20:16:07 <andythenorth> "it's just a hash tree"
20:18:47 <nielsm> as a general comment on that PR, I must say I don't like the general change of everything to Title Case, it makes things more difficult to read
20:19:53 <nielsm> I have never had good experiences reading title cased text, only neutral or bad, in particular I have seen many news headlines that were difficult to understand because title case obscured what was a proper noun and what was not
20:22:13 <hythlodaeus> nielsm: it's a standard
20:22:54 <hythlodaeus> you already have it on many parts of the UI
20:23:08 <hythlodaeus> e.g. the finances window
20:23:20 <hythlodaeus> I'm just making it consistent with standard
20:23:40 <hythlodaeus> but I won't apply it for instance in full sentences or non-names
20:24:55 <hythlodaeus> for example if a button tooltip has a verb as a title such as "Select all trains", then you can theoretically leave it non capped
20:27:20 <hythlodaeus> but it's a standard rule on english menus and dropdown buttons
20:27:31 <hythlodaeus> if you open office/libreoffice and open any dropdown
20:27:41 <hythlodaeus> you'll notice all tool/function names are capped
20:28:20 <milek7_> for consistency maybe it would better to lowercase finances window instead?
20:28:22 <hythlodaeus> I wouldn't do it on Spanish or Portuguese, for instance, since in these languages, the rule is not to cap them
20:28:53 <hythlodaeus> milek7_ I'd rather standardize it along with actual grammar rules and conventions
20:29:24 <hythlodaeus> i understand some people might not enjoy it, but I'm just fixing grammar here
20:29:42 <milek7_> i don't think there is grammar rule to capitalize everything
20:30:18 <hythlodaeus> in english it kind of is, at least on software applications
20:30:59 <hythlodaeus> just open the dropdown on your browser for instance
20:31:05 *** WormnestAndroid has quit IRC
20:32:17 *** WormnestAndroid has joined #openttd
20:33:47 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
20:34:19 <TrueBrain> This ... should work. Difficult to test without just doing it, but ... I feel good about it. Tomorrow some final tweaking and deploying on AWS in production (instead of the Dev version currently) .. should be good ..
20:36:38 <TrueBrain> you just can't tell when the new deployment is active .. that is a bit sad I guess
20:37:00 <TrueBrain> I wonder if I can make it wait till it is really deployed .. but "really deployed" is such a vague definition .. (it slowly rolls in traffic, and out if something goes wrong)
20:37:26 <TrueBrain> well, I think this alone would be a good start, so I guess I should worry about details later :P
20:38:37 *** supermop_work has joined #openttd
20:39:09 *** Wormnest has joined #openttd
20:42:21 <TrueBrain> okay, that was an easy fix \o/ :D
20:42:21 <DorpsGek_III> [OpenTTD/website] TrueBrain updated pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jed5i
20:45:33 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh requested changes for pull request #7870: Doc: improved english tooltips draft https://git.io/Jedhp
20:46:58 <DorpsGek_III> [OpenTTD/website] TrueBrain commented on pull request #116: Various of fixes in preparation of running this Docker on AWS https://git.io/Jedhh
21:09:17 *** supermop_work has joined #openttd
21:10:54 <andythenorth> Capitalisation Divides People
21:11:06 <andythenorth> conventional style on headings is capitals in UK English
21:11:15 <andythenorth> but I know lots of people who dislike it
21:13:16 <frosch123> are people who capitalise similar to people who use thousand-separators other than space?
21:18:47 <frosch123> i notice a lot of android screenshots on reddit. is that always the same guy, or does reddit favour that userbase?
21:19:15 <frosch123> nvm, it's always the same guy :p
21:19:56 <andythenorth> I like the reddit trend of taking bad phone pictures of the screen
21:20:10 <andythenorth> eh also, can anyone else compile Horse? o_O
21:20:16 <andythenorth> I broke it for glx somehow
21:20:22 <frosch123> i am used to that. customers at my old company did the same
21:20:52 <frosch123> older customer otoh print them, comment them by hand, and scan them again
21:21:19 <andythenorth> there is a meme about it
21:21:37 <andythenorth> "can you send me the pdf in a powerpoint?"
21:21:45 <frosch123> i think i used a fax machine once in my life
21:22:30 <frosch123> or rather, i watched my father explaining it to me
21:23:37 <andythenorth> in the purchase menu, they're horizontally stacked
21:25:16 <frosch123> andythenorth: if i look at horse, it may end up with rewriting the makefile
21:25:29 <andythenorth> the makefile is a car crash
21:25:37 <andythenorth> alberth started it, but I 'finished' it
21:25:45 <frosch123> make is a write-only language. when it is broken, usually you can only delete and rewrite it.
21:26:02 <hythlodaeus> I just learned microsoft changed capitalization of UI items to sentence-style
21:26:11 <andythenorth> Make seems to be an append-only type approach
21:26:18 <hythlodaeus> meaning "Update now" instead of "Update Now"
21:26:22 <andythenorth> whenever anything works, the solution seems to always be 'add more'
21:26:30 <andythenorth> works / doesn't work /s
21:27:03 <andythenorth> obviously I much prefer deleting to adding, so I really don't get on with Make
21:27:10 <hythlodaeus> So I would like to call a vote on what to do. I can make it sentence style and follow microsoft standards, or follow old standards with all capitals on on UI button titles
21:27:29 <frosch123> hythlodaeus: the documentation team at my old company decided the same last year. since then i know the term "sentence style" :p
21:28:09 <andythenorth> where do the titles show?
21:28:11 <LordAro> frosch123: make is lovely, don't dis
21:28:11 <frosch123> hythlodaeus: i would vote for only giving native people a vote
21:28:13 <hythlodaeus> Yeah, although I think most games are still using Title Style
21:28:51 <frosch123> LordAro: imo main problem is that make ignores too many errors. when there are circular dependencies or stuff it just tries to do something
21:28:57 <andythenorth> we seem to just mix freely title and sentence, in global menus
21:29:17 <andythenorth> window titles are correctly title case afaict
21:29:53 <hythlodaeus> yeah it's a bit messy
21:30:05 <andythenorth> yeah other menu items elsewhere are mixing case
21:30:08 <hythlodaeus> but it's only fair you guys make the decision
21:30:23 <frosch123> hmm, i named the folder with all the grf checkouts "devzone"...
21:30:25 <hythlodaeus> either Title Style or Sentence style
21:30:29 <frosch123> is it ok to checkout gh stuff there?
21:31:35 <hythlodaeus> IMO I prefer Title Style bc it's a bit more formal, but as long as it is uniform and goes along with a major convention, I do not mind :)
21:31:40 <frosch123> ah, i found a usecase for "following"
21:31:48 <frosch123> i can use it to remember andy's gh name
21:32:11 *** ChanServ sets mode: +v tokai
21:32:26 <andythenorth> shockingly, it's 'andythenorth'
21:32:50 <andythenorth> hythlodaeus: this is for tooltips only?
21:32:51 <frosch123> albert had some really complicated one
21:33:11 <hythlodaeus> andythenorth: eventually for the whole interface
21:33:32 <hythlodaeus> I'm only improving tooltips AFAIK
21:33:43 <andythenorth> Tooltips seem to be sentence case, and many of them are full sentences so I'd stick with that
21:34:19 <hythlodaeus> for the rest of you guys, Title Style or Sentence style?
21:34:38 <andythenorth> getting consensus here is a disaster :)
21:35:08 <frosch123> i would always capitalise "Ctrl" within a sentence, no idea whether that is already title style
21:36:06 <hythlodaeus> Oh no, key names and object names are always capital
21:36:10 <hythlodaeus> regardless of style
21:36:12 <frosch123> same for Click when used a name
21:36:14 <andythenorth> FIRS Steeltown breaks houses, I forgot about that
21:37:11 <hythlodaeus> Click, Shift, Enter, etc are always first letter capital. The difference between title style is that on strings that are not full sentences or descriptions, all first letters are capitals
21:37:34 <hythlodaeus> Build Electrified Railway - Title Style
21:37:49 <hythlodaeus> Build electrified railway - Sentence style
21:37:52 <frosch123> so, if "Use the mouse to click on stuff. Use Ctrl+Click to activate magic" does not conflict with sentence style, then i am for sentence style
21:38:04 <frosch123> mostly because sentency style is easier to explain
21:38:24 <hythlodaeus> Ctrl key - Sentence Style
21:38:31 <hythlodaeus> Ctrl Key - Title Style
21:39:10 <Samu> 3>LINK : fatal error C1047: The object or library file 'D:\OpenTTD\OpenTTD GitHub\OpenTTD\projects\..\objs\x64\Release\sdl_v.obj' was created with an older compiler than other objects; rebuild old objects and libraries
21:39:30 <frosch123> haha, i bet if you change "Build electrified railway", some NewGRF authors will complain that it was changed, no matter what style it was
21:39:49 <hythlodaeus> it was sentence style previously I think
21:40:01 *** supermop_work has joined #openttd
21:40:05 <frosch123> well, it's no good reason either way :)
21:40:17 <hythlodaeus> but I think most people seem to prefer Sentence style here, so I think I'll go with that
21:40:32 <frosch123> it's just triggered the "every change breaks someone's workflow"
21:41:06 <hythlodaeus> it would help if we were to simply guide users to a manual of style
21:41:07 <frosch123> andythenorth: "master" or some more specific branch?
21:41:31 <hythlodaeus> we can just link them to the M$ manual of style and be done with it
21:41:39 <andythenorth> frosch123: master will do
21:41:46 <frosch123> oh, there is a gui style guide... i wonder whether it says something about that
21:41:59 <hythlodaeus> can you link me frosch123 ?
21:42:23 <andythenorth> nothing about case
21:43:00 <andythenorth> frosch123: sorry, was your question about FIRS, or Horse?
21:43:01 <hythlodaeus> it's just about graphical style
21:43:07 * andythenorth assumed Horse and the build failure
21:43:13 <frosch123> andythenorth: yes, horse
21:44:34 <andythenorth> there is a more interesting branch 'procedures-are-faster-in-some-cases' which needs nml PR 66, and is very fast with pypy
21:44:57 <andythenorth> where 'very fast' means about 22 seconds, not that fast :D
21:45:42 <frosch123> compile from clean checkout worked
21:47:40 <frosch123> what is "polar_fox", and why is "git_info.py" in that folder?
21:50:40 <frosch123> it uses directories as dependencies, that's certainly troublesome
21:56:01 <andythenorth> it's a 'vendored in' library, i.e. copy-paste :P
21:56:05 <andythenorth> I could use git remotes, but eh
21:56:34 <andythenorth> it provides all the shared stuff between grfs
21:57:11 <andythenorth> polar fox should be in root, but I couldn't make the python imports work from there :P
21:57:43 <frosch123> andythenorth: if i should make a guess, i would blame "using directories as dependencies" for stuff not working
21:58:16 <hythlodaeus> I'm gonna write a small manual of style for in game and interface
21:58:31 <frosch123> add it to the gui style page in the wiki then
21:58:38 <hythlodaeus> I think the best convention should be:
21:59:41 <andythenorth> frosch123: which clown writes my stuff anyway :)
22:01:56 <hythlodaeus> Use Title Style in: Given names (e.g. companies), resource names (Iron Ore), titles (Chairman of the Board, Tycoon of the Year), nationalities and coins (North Korean, Japanese Yen),
22:03:30 <glx> frosch123: indeed dependency issues may explain the failures
22:04:10 <glx> the weird thing is each run after 'make clean' fails in a different place
22:04:20 <hythlodaeus> use Sentence style in: menu buttons (New game), tooltips ("Saves the game quick"), field names ("Profits this year"), colors ("Light blue"), objects and strucutre ("Electrified railway")
22:04:43 <glx> first in render graphics, then in render docs, then render nml, then the grf
22:05:11 <glx> so even failed deps are considered ok on next run
22:05:39 <andythenorth> glx: do the python scripts work in isolation, e.g. src/render_nml.py etc?
22:06:02 <glx> hmm I need to read the makefile to check that by hand ;)
22:06:48 *** andythenorth has joined #openttd
22:07:46 <frosch123> andythenorth: is "generated/graphics/_graphics_files_here_are_generated.txt" any hint that generation was successful?
22:08:20 <frosch123> hmm, probably not, it has an older timestamp than mose of the files
22:10:40 *** supermop_work has joined #openttd
22:12:46 <andythenorth> it's not a reliable hint
22:13:31 <andythenorth> on master head, I get 595 items in generated/graphics/
22:14:04 <frosch123> hmm, cooppaste still broken
22:14:19 <andythenorth> I am banned from it
22:19:23 <frosch123> i added another commit, that may be useful
22:24:51 <frosch123> it's english.lng, not english.txt
22:24:52 <andythenorth> I think graphics changes trigger nml now
22:26:13 <andythenorth> the speed increases are reducing the benefit of 'grfcodec + overhead'
22:26:20 <andythenorth> it's only 8 seconds faster now :P
22:26:34 <andythenorth> it was 55s or so
22:27:13 <frosch123> glx: andythenorth: pull again, i amended the english.lng bug
22:28:53 <andythenorth> frosch123: now just grfcodec again for graphics changes :)
22:32:30 <Samu> can I post a PR about changing defaults of some settings?
22:34:18 <glx> still division by zero when rendering graphics
22:35:49 <Samu> "reviewed defaults for 2020"
22:37:09 <Samu> I know some settings are there to mimic the original ttd
22:37:13 <frosch123> is that a depenency issue?
22:37:50 <glx> I don't know, but consists is always an empty list for me
22:38:52 <frosch123> render_graphics is actually the first thing that is called, it does not depend on intermediate things
22:40:08 <glx> at least with your changes it always fails there :)
22:41:00 <glx> without changes it fails in render on first run, then docs on second run, then nml...
22:41:19 *** supermop_work has joined #openttd
22:41:22 <glx> no need to make clean each time
22:41:27 <andythenorth> glx: my suspicion is that it is not rendering any graphics
22:41:35 <andythenorth> anything in generated/graphics?
22:41:37 <frosch123> andythenorth: are the ARGS used?
22:42:07 <glx> yes many pngs andythenorth
22:42:07 <frosch123> i am worried about "ROSTER = *", maybe make expands the "*" to filenames :p
22:42:27 <andythenorth> that could be changed
22:42:34 <andythenorth> it's just a string match in python
22:42:43 <andythenorth> let me check that's still used
22:42:50 <glx> python3 src/render_graphics.py '4141' '2.3.0-17-g118f7c81' '0' '*' ''False''
22:43:19 <frosch123> andythenorth: i see no difference when i run render_graphics with no args at all
22:43:46 <andythenorth> the args pass the git stuff, and the number of pool workers
22:43:54 <andythenorth> but if they're not passed, defaults are substituted
22:44:27 <andythenorth> ok roster=* can be changed trivially
22:44:37 <andythenorth> it is just a string match
22:44:45 <frosch123> glx: at least that's the same as for me
22:44:53 <frosch123> andythenorth: does not seem to be the issue
22:46:07 <glx> let's try to trace using breakpoint
22:46:16 <andythenorth> it will be something lolz :)
22:47:30 <andythenorth> maybe I should set up my azure to run it :)
22:48:19 <frosch123> when i use roster "frosch", i get the division by zero error
22:48:55 <frosch123> so, indeed, the issue is with passing the roster to render_graphics
22:49:13 <andythenorth> yes it will come up with zero rosters in that case
22:49:26 <glx> if makefile_args.get('roster', '*') == '*': <-- this is false in my case
22:49:57 <frosch123> can you print the result of the "get"?
22:50:34 <andythenorth> I could also put an exception in for no rosters
22:50:42 <frosch123> or maybe add a print to get_makefile_args
22:51:21 <frosch123> ha, so it did expand it :p
22:51:43 <frosch123> the ' are supposed to suppress the expansion of *
22:51:43 <glx> shell expanded it I guess
22:52:12 <frosch123> then using ALL instead of * should fix it
22:52:41 <glx> using * in anything command line related is asking for trouble
22:54:33 <glx> {'repo_revision': '4141', 'repo_version': '2.3.0-17-g118f7c81', 'num_pool_workers': 0, 'roster': '.devzone', 'suppress_cargo_sprites': False}
22:54:55 <frosch123> i am adding a commit
22:55:03 <andythenorth> not sure why I'm even specifying roster in the args
22:55:11 <andythenorth> I could just default to 'all' if it's missing
22:55:29 <andythenorth> oh, no, the python arg parser I'm using relies on positions
22:55:38 <andythenorth> I should probably get a better one?
22:56:04 *** andythenorth has joined #openttd
22:56:25 <glx> better use a parser similar to nmlc I guess
22:56:39 <andythenorth> there are better parses I'm sure
22:59:29 <andythenorth> now glx can make nml faster
22:59:47 <andythenorth> and soon we can profile how badly my grfs run in ottd
23:00:24 <glx> optimising doesn't make nml faster
23:00:59 <glx> but it should improve output speed if a lot of stuff is "removed"
23:01:37 <andythenorth> some of the work I've done to make it faster by removing switches is very marginal gain
23:01:38 <glx> ok now it works, I can remove the print()
23:01:55 <andythenorth> I'm definitely on negative time benefit for some of the cases :)
23:02:03 <andythenorth> better if the compiler does it
23:03:27 <frosch123> glx: btw. ottd already merges consecutive switch ranges
23:03:49 <frosch123> if they map the the same result
23:04:23 <glx> yes that makes sense, but reducing it at the source is not a bad idea
23:06:39 <glx> so many warnings, the screen buffer is too short :)
23:06:58 <glx> oh of course I don't have grfcodec in the path
23:07:10 *** andythenorth has joined #openttd
23:11:13 *** andythenorth has joined #openttd
23:12:00 *** supermop_work has joined #openttd
23:15:22 <glx> but I failed to read parsing timing :)
23:15:45 <frosch123> master has only 19.4s for preprocessing for me
23:18:17 <glx> retrying with master nmlc
23:19:06 <andythenorth> those ^^ timings are with procedures branch in PR 66
23:19:33 <andythenorth> and flherne's output patch
23:20:37 <glx> output was 8.7s with my switch PR
23:22:19 <andythenorth> the timing pastes are just to illustrate the benefit of pypy + procedures + output patch :)
23:23:32 <glx> it was faster in the other branch
23:24:13 <DorpsGek_III> [OpenTTD/OpenTTD] frosch123 commented on pull request #7868: Feature: NewGRF callback profiling https://git.io/JeFv6
23:24:14 <glx> so yes master is slower than my branch
23:24:26 <andythenorth> those timing are so painful
23:24:34 <andythenorth> I hope you have some YouTube to watch :P
23:25:12 <glx> but the timings may be affected by minwg
23:25:44 <glx> I should try the nmlc step in "native"
23:29:23 <glx> ahah 105.2s parsing in "native"
23:37:28 <glx> 88.5s preprocessing with my branch, but there's so many warnings to print :)
23:39:03 <Samu> I found a "not really a bug" with snowline interval
23:39:25 <Samu> it changes in steps of 5 units with the arrows in the settings window, but only of 1 unit in the world gen window
23:42:38 *** supermop_work has joined #openttd
23:45:41 <glx> -a---- 22/12/2019 23:43 18227767 iron-horse-master.nfo
23:45:41 <glx> -a---- 22/12/2019 23:36 17548411 iron-horse.nfo
23:46:24 <Andrew350> hey where would someone look to find info on bits used for newgrf features? trying to confirm info for the nfo newgrf specs
23:46:30 <glx> basic optimistation seems to have little effect in nfo size
23:47:04 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick opened pull request #7871: Fix: Snowline height interval is now 1, in both World Generation and Settings windows https://git.io/JeFfI
23:48:16 <Samu> thought I don't agree with the default of 15 :(
23:48:34 <Samu> I think I'm creating an Issue instead of a PR
23:48:44 <Samu> but it feels more like a suggestion
23:51:58 <frosch123> glx: 10MB of the nfo are the realsprite (long "generate/..." paths), so you reduced by 7%
23:52:43 <andythenorth> I could do something about that
23:53:11 <frosch123> i don't think there is a point in doing something about that
23:53:14 <glx> removing duplicates may help to reduce realsprite I guess
23:53:48 <frosch123> hmm, duplicate sprites are not listed when using grfcodec :p
23:53:52 <andythenorth> do we know why the grf size is stubbornly 19.x MB, even though I've removed multiple MB from nfo and nml
23:55:27 <frosch123> oh, lol, grfcodec uses container 1 by default... i was confused why the grf header looks so weird
23:57:20 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick opened issue #7872: Default openttd.cfg world gen settings don't allow Forests in Sub-Arctic https://git.io/JeFfG
23:57:42 <frosch123> container2 is 2MB smaller
23:58:16 <frosch123> oh, i enabled clipping
23:59:54 <frosch123> - $(GRFCODEC) -s -e $(PROJECT_NAME).grf generated
23:59:56 <frosch123> + $(GRFCODEC) -s -e -c -n -g 2 $(PROJECT_NAME).grf generated
continue to next day ⏵
