IRC logs for #openttd on OFTC at 2021-07-27
⏴ go to previous day
00:54:17 *** APTX has quit IRC (Quit: Farewell)
01:02:54 *** ChanServ sets mode: +v tokai
01:09:45 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
01:45:00 *** Wormnest has quit IRC (Quit: Leaving)
02:26:21 *** debdog has quit IRC (Ping timeout: 480 seconds)
03:24:35 *** snail_UES_ has joined #openttd
03:57:29 *** snail_UES_ has quit IRC (Quit: snail_UES_)
04:46:45 *** tokai|noir has joined #openttd
04:46:45 *** ChanServ sets mode: +v tokai|noir
04:53:36 *** tokai has quit IRC (Ping timeout: 480 seconds)
04:57:41 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
06:04:46 *** sla_ro|master has joined #openttd
06:10:14 *** EER has quit IRC (Remote host closed the connection)
07:24:31 *** andythenorth has joined #openttd
09:08:06 *** andythenorth_ has joined #openttd
09:11:24 *** WormnestAndroid has quit IRC (Remote host closed the connection)
09:11:37 *** WormnestAndroid has joined #openttd
09:14:46 *** andythenorth has quit IRC (Ping timeout: 480 seconds)
09:20:09 *** andythenorth has joined #openttd
09:26:21 *** andythenorth_ has quit IRC (Ping timeout: 480 seconds)
09:26:21 <TrueBrain> lol .. "industry type override" doesn't work for the name, it seems .. LAME
09:33:31 <andythenorth> these are the things
09:33:44 <andythenorth> setting the string doesn't change it?
09:36:06 <TrueBrain> need to experiment a bit more :)
09:36:36 <TrueBrain> and am I right to assume that index 0 can refer to either your new industry, or the original industries, depending on the action and even the settings therein?
09:41:13 <TrueBrain> funny, if you change the name of the NewGRF, it is not updated ingame until you remove and add the GRF :P
09:41:19 <TrueBrain> guess that is also not a really common scenario :D
09:50:08 <TrueBrain> I just love how good you have to read every part of the spec .. this sentence is funny:
09:50:09 <TrueBrain> In either scheme, you can use the value 7F or FF, respectively, to define strings shown for languages that you do not provide translation for. First set all strings to a default value using this, and later override the language specific ones if they exist.
09:50:23 <TrueBrain> it is written as a suggestion, but really, it is not :P
09:50:42 <peter1138> NewGRF is horrible.
09:50:51 <peter1138> When goes it throw out, etc, etc...
09:53:02 <TrueBrain> not sure NewGRF is horrible; just the specs makes me read real good :D
10:04:05 <TrueBrain> how ever I slice it, I cannot use 09 to overwrite an industry name .. so copy/disable it is!
10:05:15 <peter1138> You're not meant to read the specs, just entered random hex numbers and see what happens...
10:06:47 <TrueBrain> AFL the fuck out of it :D
10:09:28 <LordAro> it would hit a *lot* of NOT_REACHED blocks, is my prediction
10:09:41 <TrueBrain> how to validate if it created what the user entered
10:09:43 <TrueBrain> now that is the question :)
10:16:39 <TrueBrain> that was interesting
10:18:16 <TrueBrain> funny how smooth that works .. press green button, load NewGRF, start new game, open Fund Industry, rename "Power Station" into something else .. hit the green button, and BAM, it is in-game
10:20:48 <TrueBrain> now I wonder how to do layouts ..
10:47:30 <andythenorth> but the signs are emerging :)
11:11:54 <LordAro> TrueBrain: did you get rid of the requirements.txt vs pip freeze check?
11:11:56 <LordAro> i can't seem to find it
11:57:18 <TrueBrain> Copy/pasting GPL code? :p :p
12:07:03 <andythenorth> Element got another $30m, maybe they fix ease of use and I can avoid getting Slack
12:11:10 <LordAro> TrueBrain: just confirming i was doing the right thing :p
12:11:44 <TrueBrain> if you like, I can license you to use those lines of code, no problem :P
12:12:26 <LordAro> TrueBrain: but since you're here... how would you fit a windows-only package into the whole requirements.base/requirements.txt structure? (i.e. """pywin32; sys_platform == 'win32'""") - only solutions i can think of are either duplicating requirements.txt or doing some hackery with sed
12:13:02 <TrueBrain> in setup.py you can solve that
12:13:06 <TrueBrain> but that is ofc not pinned
12:13:28 <LordAro> mm, pinning is required (build reproducibility, etc)
12:13:37 <TrueBrain> yeah, which is a good thing
12:13:43 <TrueBrain> hmm, it is an interesting problem
12:13:55 <LordAro> (we're so serious about that here that we have a caching server in front of pypi)
12:14:18 <TrueBrain> I do the same on my jobs :)
12:14:38 <TrueBrain> what you can do, is have a requirements.base/txt, and a requirements-win.base/.txt
12:14:43 <TrueBrain> in the latter, you do "-r requirements.txt"
12:14:52 <TrueBrain> that way at least you do not have duplication
12:15:50 <TrueBrain> hmm, did PEP 508 land?
12:16:53 <TrueBrain> PEP 508 supports 'sys_platform == "linux"' etc in requirements files
12:16:59 <TrueBrain> but I do not think "pip freeze" outputs them again
12:17:06 <TrueBrain> so validating correctness becomes tricky :)
12:17:20 <LordAro> the thing i'm trying to avoid is one requirements.txt file being updated and the other getting forgotten
12:18:26 <TrueBrain> doing that is easy, but doing that while keeping the check OpenTTD has in Dockerfile .. not so much :P
12:19:09 <TrueBrain> no, sorry, no suggestions there .. bit of a pita :)
12:28:32 <TrueBrain> once again I forgot to read the smaller print: "The layout consists of a list of the above tile definitions, terminated by two bytes: 0,80h " :P
12:29:28 <andythenorth> always read the smaller print :P
12:35:32 <glx> isn't there 3 different ways (at least) for requirements in python ?
12:59:38 *** snail_UES_ has joined #openttd
13:20:37 *** roadt__ has quit IRC (Read error: Connection reset by peer)
13:37:05 <TrueBrain> guess next is to make the dropdown show sprites :P
13:37:29 <andythenorth> +/- rows and cols also :D
13:37:32 <andythenorth> free-form drawing!
13:37:44 <andythenorth> grid, sprites, drag-drop, profit
13:38:19 <TrueBrain> most of my time is in finding the correct React components etc
13:38:24 <TrueBrain> not in doing the NewGRF stuff :D
13:38:52 * andythenorth wonders which is worse: current state of JS and UI, or newgrf
13:42:33 <TrueBrain> hmm .. for original industries, I guess I should embed OpenGFX or something
13:52:00 <Samu> I'm experimenting another kind of rectangle limit
13:53:11 <Samu> no search limitation vs bresenham's vs rectangle
13:55:27 <_dp_> how to use null driver? I want to advance a savegame for n ticks
13:58:29 <_dp_> hm... and how to a savegame at the end?
14:00:23 <LordAro> how to save a savegame?
14:00:43 <nielsm> I don't think that's a feature...
14:01:04 <nielsm> could maybe be a parameter to the null driver, "execute this script file at the end"
14:01:06 <_dp_> well, I guesss I can add a tiny patch...
14:01:18 <_dp_> but, yeah, having a parameter would be nice
14:01:47 <nielsm> like -v null:before=setup.script:after=cleanup.script
14:03:38 <_dp_> I think just null:save=aaa.sav would already cover the most cases
14:04:10 <_dp_> before/after can be used as extras for advanced setup but just save is better for convenience
14:08:09 <TrueBrain> or, you know, you could use autosave_on_exit :)
14:09:10 <TrueBrain> (so funny how even developers don't know all the settings :P)
14:10:21 <_dp_> I want to run them in parallel so I need a way to save into different files
14:10:47 <_dp_> well, I guess I can do schenanigans with openttd home directory but meh
14:11:00 <_dp_> I already coded save param :p
14:11:17 <TrueBrain> to be clear, I also didn't know of this setting till I was doing my massive savegame-testing-stuff :)
14:12:59 <TrueBrain> in fact, I think I made a PR to allow me to make a savegame on exit, till someone says: dude, there is a setting for that :D Funny how circles go :P
14:14:11 *** snail_UES_ has quit IRC (Quit: snail_UES_)
14:14:52 <_dp_> well, null driver param is a much better solution imo
14:16:03 <TrueBrain> the game_start.scr script was also surprisingly helpful
14:16:12 <TrueBrain> never imagined scripts having a use
14:25:12 <_dp_> scripts are made for a different purpose so not exactly convenient for testing
14:26:05 <_dp_> I'd like to specify everything in a command line, not depend on some magic globals
14:26:32 <TrueBrain> " Variational action 2 variable 43 for industry tiles can help you limiting tile consumption to 2 or 3 per industry." .. hmm ... I like to ask: how? :D I need to pick andythenorth 's brain after dinner :P
14:27:16 <andythenorth> TL;DR branch in the tile graphics chain to correct tile layout, using the industry layout as the var
14:27:30 <TrueBrain> but then you only need 1 tileID, not?
14:27:58 <andythenorth> 1 per specific tile
14:28:07 <andythenorth> tiles have some immutable properties like animation rate
14:28:16 <andythenorth> so some industries need 1 tile, some need more
14:28:23 <TrueBrain> okay, that makes sense
14:28:30 <TrueBrain> the "2 or 3" was confusing me
14:28:34 <andythenorth> so min. 1 tile per industry
14:28:35 <TrueBrain> like, what is wrong with 1 :)
14:28:45 <TrueBrain> that surely was a quick dinner somehow :P
14:28:48 <andythenorth> in principle tiles could be shared across industries, but there's no real upside
14:28:57 <andythenorth> yeah I got distracted
14:32:36 <TrueBrain> I need to figure out how to do stuff like "add to a list" in React :P
14:34:07 <TrueBrain> but first .. can I do isometric projection in React :D
14:38:15 <_dp_> I fail to understand why that grid even exists
14:39:11 <_dp_> who on earth thought that would be a good idea :p
14:57:41 <TrueBrain> hmm ... guess I need a way to filter out transparent pixels in a browser ... interesting issue :P
15:04:37 <glx> and don't forget the magic company colors and flashing ones :)
15:05:25 <glx> (well company colors are just blue shades, so not really an issue)
15:08:22 <TrueBrain> seems I have to draw it on a canvas to manipulate it
15:26:56 <TrueBrain> right, next question is to figure out how the game knows which old-tile-id matches with which basegraphics sprite id :)
15:55:40 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
15:56:09 *** WormnestAndroid has joined #openttd
15:57:15 *** ChanServ sets mode: +v tokai
16:16:00 *** Progman has joined #openttd
16:20:23 *** WormnestAndroid has quit IRC (Remote host closed the connection)
16:21:59 *** WormnestAndroid has joined #openttd
16:30:07 *** WormnestAndroid has quit IRC (Remote host closed the connection)
16:32:08 *** WormnestAndroid has joined #openttd
16:35:13 *** gelignite has joined #openttd
16:45:24 <TrueBrain> tnx glx, was looking for the first table .. hex .. didn't think of that :D
16:47:01 <glx> the joy of newgrf, sometimes it's decimal, sometimes it's hex, but it's never clearly stated ;)
16:49:24 <TrueBrain> glad it is not just me :)
16:51:22 <glx> properties, callbacks, variables, the "number" is always in hex
16:53:39 *** HerzogDeXtEr has joined #openttd
16:54:44 <TrueBrain> I loved this part: "this is a word value in little endian notation, e.g. 8134 becomes 34 81. "
16:54:52 <TrueBrain> like .... okay .... how did this help me? :D
16:55:07 <TrueBrain> some notion of what is what would have helped there :P
16:56:19 <glx> luckily NFO has the nice \w \d and other friends :)
16:56:45 <glx> but I guess you encode "by hand"
17:01:57 *** Compu has quit IRC (Remote host closed the connection)
17:04:18 *** jottyfan has joined #openttd
17:06:42 <glx> could be worse, looks cleaner than nml :)
17:11:10 <TrueBrain> meh, with OpenTTD in a browser, I keep running out of memory on my machine
17:11:17 <TrueBrain> Firefox doesn't seem to like opening the console too much
17:20:57 <TrueBrain> although I think the wiki page is incorrect
17:26:49 <TrueBrain> ah, yes .. construction vs finished
17:26:58 <TrueBrain> those tiles are just a bit weird
17:43:35 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
17:46:54 *** WormnestAndroid has joined #openttd
17:54:44 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
18:00:42 <TrueBrain> hmm, industry prop 14 .. is that for each cargo or for all combines?
18:05:29 <TrueBrain> "The industry can be exploded by a military airplane (oil refinery) " <- wait, what now? :P
18:08:49 <andythenorth> "The produced cargo will be distributed to stations only after it exceeds this amount."
18:08:54 <TrueBrain> okay, so all the other industry settings are pretty easy .. mostly about how it looks etc. The nasty stuff is in the callbacks, ofc :D
18:09:05 <TrueBrain> andythenorth: yeah .. which cargo :)
18:09:06 <andythenorth> I never understood prop 14
18:09:13 <andythenorth> industry_cmd.cpp will know
18:09:19 <andythenorth> it's the canonical source! :D
18:09:35 <andythenorth> or it will be "move cargo to station"
18:09:36 <TrueBrain> done for each cargo
18:09:47 <andythenorth> let's not ask 'why' that feature exists at all
18:09:49 <TrueBrain> so you cannot move one cargo faster than the other
18:10:12 <TrueBrain> like pax per 1, but steel per 16, you cannot do that :)
18:10:47 <andythenorth> this is example of vestigal spec that would be better rm-ed
18:11:06 <TrueBrain> what is the default value I wonder
18:11:28 <andythenorth> let me know if you find out :)
18:11:35 <andythenorth> might be 8 for reasons
18:12:21 <andythenorth> FIRS has the following
18:12:22 <andythenorth> self.min_cargo_distr = "1" # just use the most common value from default OTTD industries, this property needs set but has little use
18:13:02 <TrueBrain> correction: all are 5 but one
18:13:05 <TrueBrain> forest, that has 30
18:13:21 <andythenorth> oh because it cuts whole trees
18:13:28 <andythenorth> so we have to have 30t of tree at once
18:13:38 <andythenorth> anything else is unrealistic :)
18:13:58 <TrueBrain> worst part is that a recession is calculated after that :P
18:13:59 <andythenorth> I think the original animation was 4 trees cut in one cycle
18:14:03 <TrueBrain> so you can still have < 30t :)
18:17:23 <TrueBrain> "Do not rely on handling production cb sub-chain as default ID for the varaction 2 checking 0C, this can have unexpected results. However the ID of the varaction 2 handling the production cb itself should be the default ID for the 0C check. Makes sense? "
18:17:27 <TrueBrain> happy the author realised it does not
18:18:19 <andythenorth> pretty certain history will show that as me
18:20:05 <andythenorth> oh it was frosch
18:20:40 *** frosch123 has joined #openttd
18:21:19 <andythenorth> I am sure that edit was me, but I never understand mediawiki history
18:21:21 <andythenorth> it's so fucking odd
18:21:41 <andythenorth> yes it was me :D
18:22:34 <andythenorth> 10 years later me doesn't know what 10 years ago me was talking about
18:22:40 <andythenorth> but it was clearly important at the time
18:24:03 <TrueBrain> I think I need an example of this production cb ..
18:26:33 <TrueBrain> I am assuming that "registers" is the temporary storage varact2 talks about
18:27:11 <andythenorth> registers are temp or permanent
18:28:42 <andythenorth> dunno might be misleading you :D
18:28:55 <frosch123> looks like it worked
18:29:04 <frosch123> getting tb into newgrf :p
18:30:15 <andythenorth> this isn't newgrf
18:30:29 <TrueBrain> and yes, I blame you for the name frosch123
18:30:29 <andythenorth> it is well known that TB has nothing to do with NewGRF
18:31:50 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
18:33:16 *** WormnestAndroid has joined #openttd
18:35:05 <TrueBrain> okay, the documentation took a turn for the worse when it comes to Action2 :P
18:36:07 <TrueBrain> so there are feature-specific "type"s
18:38:02 *** Wormnest has joined #openttd
18:39:21 <andythenorth> I might misunderstand, but we have e.g. for industry, action 2, advanced / new action 2, varaction 2, advanced varaction 2, and random action 2
18:39:37 <andythenorth> and procedures, which are a varaction 2 called like a var
18:39:57 <TrueBrain> where for action0 it is pretty easy to find, when you go from left to right in the command, what it is doing
18:40:13 <TrueBrain> for industry, 00 .. 02 are industry specific, I think
18:40:17 <TrueBrain> but all of a sudden I see 89
18:40:22 <TrueBrain> which is in a completely different part of the wiki
18:40:27 <TrueBrain> bit hunting for eggs :D
18:40:35 <andythenorth> that's just a quirk, that's 'produce'
18:42:20 <frosch123> TrueBrain: 80+ are the same for all features
18:43:03 <frosch123> < 80 are different per feature, but several feature share the same logic
18:43:26 <frosch123> you can check the frontpage, which links are the same
18:47:57 <frosch123> andythenorth: how can we avoid tb implementing deprecated crap like random-action? :p
18:48:19 <andythenorth> tell TB not too?
18:48:52 <andythenorth> TB: don't implement random-action
18:49:08 <andythenorth> just get some bits in a var instead, maybe with modulo if needed
18:49:52 <TrueBrain> the piece that is not connecting for me, how a cb chain is assigned
18:50:04 <TrueBrain> I can tell an industry that it should call its production cb every 256 ticks
18:50:10 <TrueBrain> but .. what is it going to call?
18:50:40 <frosch123> action3 -> varaction2 checing var 0C -> continue with callbacks/production/whatever
18:50:57 <TrueBrain> yeah, all this talk about "graphics" is confusing :D
18:51:18 <DorpsGek> - Update: Translations from eints (by translators)
18:53:29 <TrueBrain> okay .. that just leaves the question of "wtf are those \wx0000" doing in the URL andythenorth gave me .. lets find out :D
18:53:40 <TrueBrain> (as in, I get it is a word, but I don't get why they ar ethere)
18:54:08 <frosch123> they are the labels
18:54:22 <frosch123> as it "goto <label>"
18:54:25 <TrueBrain> 12 * 11 00 0A \b2 01 FF \wx0000 21 02 22 01
18:54:47 <TrueBrain> action 0, industries, 2 props, 1 info, industry 0xff
18:54:57 <TrueBrain> prop 21 to 02, prop 22 to 01
18:55:03 <TrueBrain> I just don't get what the \wx0000 does :D
18:55:20 <andythenorth> 'everything is in the graphics chain'
18:55:31 <andythenorth> so action 3 is a dispatcher for sprites or logic :P
18:55:51 <andythenorth> but it's not in the 'graphics chain' when we talk about the 'graphics chain' meaning sprites
18:55:56 <andythenorth> it's all so simple :)
18:56:00 <frosch123> \b10 -> 0A, \w10 -> 0A 00, \d10 -> 0A 00 00 00, \wx10 -> 10 00, \dx10 -> 10 00 00 00
18:56:30 <frosch123> \wx takes the following thing as hex number and outputs it as unsigned word
18:56:45 <TrueBrain> yeah, and I don't understand what those 0s are doing there :D
18:56:57 <frosch123> "\wx0000" is the same as "\wx0"
18:57:09 <frosch123> someone liked adding leading zeros :)
18:57:13 <TrueBrain> again, I am not misunderstanding what the syntax means .. I am missing what it does :P
18:57:23 <TrueBrain> 12 * 11 00 0A 02 01 FF 00 00 21 02 22 01
18:57:36 <TrueBrain> there are 2 bytes I cannot account for :P
18:58:08 <frosch123> ah, that's the question
18:58:28 <frosch123> ok, the thing is, it's not "industry 0xff"
18:59:01 <frosch123> "B*" means extended byte: 00-FE are normal, FF means a word value follows
18:59:03 <TrueBrain> I can feel where this is going :P
18:59:11 <TrueBrain> these details ... :D
19:01:00 <frosch123> to simplify action6-usage, nml outputs all "B*" as "FF \w...", even if the short form would suffice
19:03:52 <TrueBrain> but not for action3, I am guessing?
19:04:13 <TrueBrain> B/B*, okay .. but isn't B implied with B* .. hmm
19:04:33 <TrueBrain> only B* for vehicles ..
19:05:17 <frosch123> yeah, it was a B first, but then changed to B*, but noone dared breaking things using id FF :p
19:05:32 <frosch123> didn't we change this in version8?
19:06:48 <glx> <frosch123> to simplify action6-usage, nml outputs all "B*" as "FF \w...", even if the short form would suffice <-- would have been a pain to properly handle action6 with variable size B*
19:06:51 <TrueBrain> okay, so despite the fact you can use IDs > 255 in Action0, you cannot in Action3 for industries, if I am putting 1 and 1 together :)
19:07:26 <TrueBrain> 128 per NewGRF anyway
19:08:00 <glx> are you expecting some logic in the spec ?
19:08:14 <TrueBrain> I am not having any opinion about the spec
19:08:17 <TrueBrain> just putting things together in my head
19:09:27 <TrueBrain> "limited to 255 to allow extending Action3 with an extended byte later on." <- some comments are more confusing than helping :D
19:09:29 <glx> remember newgrf spec is basically a hack on hack around original TTD/TTDp limitations
19:09:59 <frosch123> TrueBrain: limiting to 255 instead of 256 makes newgrf using if 0xFF invalid. thus you can later use 0xFF to indicate a word-sized id
19:10:01 <TrueBrain> people seem to be waiting for me to bash the specs or anything .. I really am not. Just trying to wrap my head around it
19:10:20 <TrueBrain> frosch123: the comment should mention the "instead of" part :D
19:10:33 <TrueBrain> context is everything :D
19:10:42 <frosch123> TrueBrain: anyway, "industries" are the nice part of newgrf :p
19:10:57 <glx> yeah linking spec bits together may take some time
19:11:06 <andythenorth> frosch123 not trains? :)
19:11:40 <glx> trains are not too bad, just too many things allowed in callbacks
19:11:42 <frosch123> vehicles and stations are "ttd weirdness is the default, newgrf specifiy differences to ttd behavior". everything newer is "there are some sane basics, newgrf define stuff from scratch"
19:12:43 <frosch123> so industry newgrf define how a industry shall behave. while vehicle newgrf define how their vehicles are different to default vehicles :p
19:12:47 <TrueBrain> okay, so advanced varaction2 didn't get its own type, but is a mask. now that part clicks a bit :)
19:14:35 <frosch123> TrueBrain: adv. va2 has the same thing as \bx again. you only need the 32bit variant. the 8 and 16 bit variants only add complexity to save a few bytes on disk
19:14:40 <frosch123> (no different at runtime)
19:15:41 <TrueBrain> is there, performance wise, a difference between using prop9 on industries vs prop 8 to copy it and disable the old?
19:15:52 <frosch123> ok, the "no difference at runtime" is not true. it's just "no useful difference" :p
19:16:22 <frosch123> yes, one is mandatory, the other is optional
19:16:40 <glx> and it must be the first set prop
19:16:59 <TrueBrain> I couldn't get prop9 to work in a way that was useful to me :P
19:17:29 <frosch123> the "overwrite original stuff" is rather pointless in ottd. it's only meaningful if you add newgrf to a savegame without newgrf, but we do not support that
19:17:43 <TrueBrain> okay, so just ignore prop9
19:17:48 <TrueBrain> simple == better :)
19:19:47 <_dp_> hm, weird, null driver doesn't seem to be running the game state loop...
19:20:19 <glx> it should, else regression would not work
19:20:57 <_dp_> yeah, I'm trying to figure what am I doing differently
19:21:15 <glx> TrueBrain: action 2 id can be reused in chain
19:21:16 <frosch123> there are only 256 action2 ids. you have to reuse them, otherwise you run out pretty fast
19:21:30 <TrueBrain> how I read it, you can use any set before you, and it will point to the one you expect
19:21:32 * andythenorth runs out fast anyway :P
19:21:39 <TrueBrain> anyone after you can overwrite a setid again, but that is of no interest to you?
19:22:16 <frosch123> you are always referencing the last va2 using a specific id
19:22:23 <glx> from 0xfd you can link to previous 0xfd :)
19:22:39 <TrueBrain> frosch123: "last" in what context? Runtime context? Or compile-time?
19:22:53 <TrueBrain> k, so that is what I expected
19:23:05 <glx> all this magic is nicely handled by nml
19:23:28 <frosch123> glx: let him suffer a bit with nfo. then nml looks nicer :p
19:23:42 <TrueBrain> I cannot use NML, so that is not something I can enjoy anyway :)
19:23:51 <andythenorth> I wanted to yolo approve it, but I daren't
19:26:38 <TrueBrain> okay, tnx a bunch for all this information :D Lot faster than trying out, failing, reading source, trying again, etc :D
19:27:02 <TrueBrain> I think I understand the concept of Action2/3 .. now to think of a good / useful way to put it in TrueGRF :)
19:28:12 <glx> sprites, selection of sprites, when the selection should happen
19:29:47 <andythenorth> TrueBrain location check callback is a good one to try
19:30:14 <TrueBrain> hmm .. I should extend TrueGRF to "Compile and start a new game"
19:30:17 <andythenorth> quick test is only allow the industry to locate in towns with index modulo 3
19:30:21 <TrueBrain> maybe even with same seed ..
19:37:39 <frosch123> maybe rewatch that stream :p
19:37:48 <_dp_> great, something pauses the game and null driver "runs" paused :/
19:38:43 <_dp_> also how to stop openttd from loading newgrfs from my .openttd? I'm already giving it a custom config elsewhere
19:38:45 <andythenorth> frosch123 you should re-graph them, but detecting procedure calls, and chained procedure calls :P
19:40:21 <glx> so it ignores global folders
19:41:42 <_dp_> -X too new, I'm running 1.11.2 :(
19:44:41 <andythenorth> hmm when did I run out of action 2 IDs recently?
19:45:02 <andythenorth> oh I was going to use procedures to assign names to town and industry permanent storage
19:45:33 <andythenorth> so I could just call supplies_delivered_date() or something instead of LOAD_PERM(27)
19:46:00 <DorpsGek> [OpenTTD/nml] frosch123 approved pull request #227: Fix f2cab72: Called procedures may receive wrong values in parameters https://git.io/J4hU0
19:46:04 <andythenorth> but we have 256 registers for towns and 256 for industries
19:46:21 <andythenorth> and procedures, at least the way I use them, consume 'global' action 2 IDs
19:46:26 <DorpsGek> [OpenTTD/nml] glx22 merged pull request #227: Fix f2cab72: Called procedures may receive wrong values in parameters https://git.io/JWyIz
19:47:05 <glx> well the id is freed after last use of the procedure
19:48:57 <TrueBrain> frosch123: haha, I have no clue how my trees are going to look :D
19:49:20 <andythenorth> thanks frosch123 glx
19:49:57 * andythenorth has 'tested' nml a bit recently
19:50:24 <TrueBrain> frosch123: big wishlist for v9 :D
19:50:44 <andythenorth> well at least nml was fixed quickly :D
19:50:47 <glx> at least you found bugs to fix, which is always good
19:50:55 <andythenorth> we should release nml I think
19:50:59 <andythenorth> I am not the only user :P
19:51:27 <frosch123> TrueBrain: i was just surprised that it was not yet on the list
19:52:59 <TrueBrain> I see 3 approaches for the callback stuff .. a very opinionated "you can only do these N things", which will cover the basic usage, but not andythenorth et al .. or a coding approach, or a visual approach .. hmm .. I will have to toy with it a bit ..
19:53:02 <frosch123> hmm, good point. newgrf is from 2002, grfv8 is from 2012, so we are almost at that important date
19:53:21 <andythenorth> TrueBrain you don't need to cover me :P
19:53:37 <frosch123> TrueBrain: can you add a "fork FIRS" button?
19:54:17 <andythenorth> I am off in la-la land
19:54:21 <frosch123> select firs version, enter new names for industries and cargos, compile, done
19:54:26 <frosch123> should be most usecases
19:54:29 <andythenorth> trying to work out if I can fake lists and loops in procedures
19:54:36 * andythenorth would like a list type in nfo
19:54:46 <andythenorth> and an iterator for loops :P
19:54:52 <TrueBrain> it is funny in how many ways NFO does arrays :)
19:55:00 <TrueBrain> from fixed-size, to length-prefix to nul-postfix :)
19:55:09 <TrueBrain> (for what-ever variation of nul)
19:55:25 <frosch123> is that more or less variants than in ottd savegames?
19:55:35 <TrueBrain> 1.12 savegame? Less :)
19:55:56 <TrueBrain> it is very natural :)
19:57:00 <frosch123> pretty sure we classified the acceptance callbacks under "noone uses them, so we do not need to extend them"
19:57:01 <andythenorth> really big icons
19:57:48 <andythenorth> oh it's house crap for 1F I see
19:58:02 <andythenorth> I always forget about houses
19:58:29 <frosch123> it's also industry tiles starting to accept oil only after 1960
19:58:40 <frosch123> and all that ECS stuff
19:59:20 <andythenorth> special flags bit 1 just discards the need for 2B imho
19:59:49 <andythenorth> 'my opinion' on newgrf spec isn't useful :D
20:02:03 <andythenorth> "Allow more than 256 VarAction2 ids."
20:02:11 * andythenorth signs up for that newsletter
20:02:36 <andythenorth> " Instead prefix every variable with the "scope" (parent, self, more in future), so different scopes can be mixed easily within expressions, without chaining to other Action2."
20:02:36 <frosch123> TrueBrain: how is the 3DS port going? did you include the polish translation?
20:02:53 <andythenorth> ^^ /me just found that with towns
20:02:57 <glx> B* for action 2 IDs could be a nice improvement
20:03:04 *** Gustavo6046_ has joined #openttd
20:03:08 * andythenorth tried to mix town and industry vars in expressions
20:03:09 *** Gustavo6046 has quit IRC (Remote host closed the connection)
20:03:09 *** Gustavo6046_ is now known as Gustavo6046
20:03:21 <andythenorth> procedures to the rescue!
20:03:34 <andythenorth> procedures are the best nml feature since...well...anything really
20:03:57 * andythenorth considers faking a loop iterator, but it's limited to about 255 iterations
20:04:07 <andythenorth> and using some registers to fake lists
20:04:08 <glx> the scope prefix is something for nml, with auto split in sub action 2 on scope change
20:04:51 <andythenorth> replace all nml var access with intermediate procedures :P
20:05:04 <glx> and I think most nml devs already had ideas about scope stuff
20:05:14 <glx> the current way is just weird
20:06:50 <andythenorth> it's 'fine' for industry
20:06:58 <andythenorth> scopes for trains are weird
20:07:33 <glx> "parent" is weird, it's the worst choice of naming :)
20:07:40 <andythenorth> it's so confusing what PARENT is when it's an articulated vehicle
20:07:56 <andythenorth> we could just hide it all in procedures
20:08:12 <andythenorth> that's probably bad sticking plaster idea
20:08:47 <andythenorth> I think I wrote a procedure to get the 'parent' of an articulated vehicle, based on position
20:12:14 <andythenorth> that should just be in nml :P
20:17:54 *** gelignite has quit IRC (Quit: Stay safe!)
20:18:44 <glx> andythenorth: I think you can save IDs and still use many procedures, but that implies indirect calls
20:24:34 <_dp_> finally figured it out, game is paused so null driver does nothing
20:24:52 <_dp_> and unpausing it in the driver doesn't help because it's loaded in the game loop :(
20:26:10 <_dp_> now.. how do I unpause the freaking save after load...
20:34:57 <andythenorth> glx call to a switch named use_procedure(x)? Where x is the procedure I want? :P
20:45:22 <TrueBrain> frosch123: 3DS port? What are you smoking? :D
20:46:33 <TrueBrain> lol, 1 user did a refund on GOG :) Now that is funny :)
20:46:52 <glx> refund for something free ?
20:47:19 <frosch123> TrueBrain: mail too old, you already forgot it :)
20:47:33 <TrueBrain> did you read the fantastic mass-mail to a lot of free software companies?
20:47:46 <TrueBrain> that was a treat :)
20:48:08 <frosch123> though, not sure whether i really "read" it
20:48:29 <TrueBrain> after a few lines it became "skimmed over" :P
20:48:53 <TrueBrain> still, in the top 5 of mails we received in my book :)
20:50:06 <frosch123> i think it came twice
20:50:18 <TrueBrain> I think the second time had an update of sorts
20:50:22 <TrueBrain> but couldn't be bothered :P
20:51:02 <frosch123> hmm, the ratio of steam vs gog sales is better than the ratio of steam vs gog forum posts
20:51:21 <TrueBrain> GOG is a lot more than I expected
20:51:33 <LordAro> there is a gog forum?
20:52:54 <frosch123> i felt bad for gog to put in so much effort, when i saw that forum activity
20:52:59 <frosch123> but the "sales" are much better
20:53:04 *** WormnestAndroid has quit IRC (Remote host closed the connection)
20:53:16 <TrueBrain> just not really chatty :)
20:53:23 <frosch123> maybe gog customers are smarter and know how to duckduck things
20:53:28 <TrueBrain> if you have nothing good to say, shut up? :D :D
20:54:17 *** WormnestAndroid has joined #openttd
20:54:41 <TrueBrain> NewGRF question: if you need to check 0C on callback to know if it is a production cb, why not allow on action3 to already mention that? Feels like the first chain is now always a bit silly? (asking to check if I am missing something)
20:55:08 <frosch123> action3 switches on different things :p
20:55:27 <frosch123> but yes, often you have to duplicate the 0C switch for all action3 cases
20:56:36 <TrueBrain> are these chains expensive, or is it so close to the machine it is really not that relevant?
20:57:53 <frosch123> noone knows. usually they become expensive because they are called so often, not because a single one takes too long
20:58:08 <frosch123> but that is hard to measure, because the instrumentation to measure single callbacks is so expensive :)
20:58:11 <glx> or because andy abuse them :)
20:58:32 <TrueBrain> frosch123: lol, okay, that on its own answers the question :D
20:58:46 <frosch123> we talked about adding statistical profiling for newgrf
21:00:27 <frosch123> some years ago i could speed up nuts by 8% or so, by sorting + binary search the switch ranges
21:00:30 <andythenorth> currently I only have to optimise for compile time
21:00:41 <frosch123> but that only worked beause nuts switches on like 200 cargo types :p
21:00:42 <andythenorth> I can ignore run-time performance because we are blind to it
21:01:27 <TrueBrain> and action2 chains are basically only: get props/vars/registries values, do math, store in registry, return, right?
21:02:18 <frosch123> maybe, not sure how to answer that :p
21:02:53 <glx> some vars are cpu intensive, but many are cached
21:02:54 <TrueBrain> haha, okay, that is fair :) Euhm .. it is basically an integers in / integers out
21:03:26 <frosch123> yes, but there are multiple paths for input and output
21:03:46 <LordAro> TrueBrain: are there stats on gog users anywhere?
21:03:47 <frosch123> there is not just one type of register :p
21:03:49 <TrueBrain> and all those switches I always see in NML are just a means to an end, right?
21:03:54 <TrueBrain> LordAro: in the info@ mailbox :)
21:04:17 <glx> switches are typical var action 2
21:04:32 <TrueBrain> I am really asking vague questions, sorry :P
21:05:02 <frosch123> TrueBrain: nml switches do not match 1:1 with nfo action2
21:05:18 <frosch123> nml expressions and stuff often require multiple va2 for encoding
21:06:04 <andythenorth> off-topic: is everyone here a product engineer? Or does anyone do internal tools or business intelligence stuff?
21:06:27 <TrueBrain> ha, I see I am not the only one asking vague questions :P
21:06:38 <TrueBrain> "do internal tools" .. :D I think we all do ;)
21:06:47 <frosch123> andythenorth: we have a "everyone does a bit of internal tools" policy
21:07:21 <frosch123> no idea what "business intelligence" is... sounds like something with a lot of names in different contexts
21:07:24 <andythenorth> engineers are so hard to hire in the UK, we have all engineers on product nearly all of the time, or alternatively on customer support
21:07:40 <andythenorth> the level of manual processes we run inside the business is shocking
21:07:50 <andythenorth> and we have very limited tools for data
21:07:54 <TrueBrain> devoppppsssssssssss
21:08:01 <andythenorth> it's always more pressing to do more product
21:08:05 <andythenorth> because product can be sold
21:08:20 <TrueBrain> job of a Product Owner to keep that balance healthy :)
21:08:34 <glx> you just need to find someone annoyed enough by manual processes
21:08:34 <andythenorth> that's me, and I always decide 'more product'
21:08:42 <TrueBrain> Product Manager pushes for more stuff to sell, Product Owner says: I need time to do A, B, C too
21:08:51 <TrueBrain> so you are a shitty PO :P :P
21:09:01 <andythenorth> frosch123 yeah, but there's a missing column saying 'time could be spent on things to sell' :P
21:09:34 <andythenorth> well if I hired someone else as PO and they pushed for internal tools, I'd fire them
21:09:43 <andythenorth> we make more money building product
21:09:45 <TrueBrain> hence: it is a balance ;)
21:09:52 <andythenorth> but we have batshit internal process
21:10:10 <andythenorth> nothing is automated, and we have 10s of people doing rudimentary manual tasks
21:10:12 <TrueBrain> in general you sell more if your internal process is streamlined as well :)
21:10:21 <TrueBrain> which is the same, I guess
21:10:26 <frosch123> andythenorth: anyway, my experience from my past companies: internal tools are crap if they are developed by people who do not use them themself
21:10:39 <andythenorth> generally they can be bought off the shelf
21:10:41 <TrueBrain> ddeeevvvvooooppppssssss :D
21:10:43 <frosch123> so "people exclusively doing internal tools" seem to always be a waste
21:11:05 <andythenorth> I mean...we did do all the classic mistakes, like building our own CRM, ticket system etc
21:11:23 <andythenorth> TrueBrain is that some kind of meme?
21:11:43 <TrueBrain> no; it is a solution to your problem :P
21:11:57 <andythenorth> ever tried hiring a devops engineer?
21:12:02 <andythenorth> rarer than unicorns
21:12:19 <TrueBrain> have seen plenty :)
21:12:27 <frosch123> devops is better than having devs and ops
21:12:44 <andythenorth> we don't have ops
21:12:46 <TrueBrain> lot of dev like to do some ops too, if they can automate it
21:12:49 <frosch123> pure ops usually send excel sheets via mail, and ask for input and sending them back
21:12:51 <TrueBrain> not many ops like to do dev .. that is rare :P
21:12:55 <frosch123> so they can manually merge them from all teams or so
21:13:19 <TrueBrain> andythenorth: you might not call them ops, but you sure have ops
21:13:23 <TrueBrain> someone is keeping those machines alive
21:13:30 <andythenorth> oh I don't mean hosting :)
21:13:34 <andythenorth> I mean business stuff
21:13:41 <andythenorth> hosting is a different shitshow
21:13:41 <TrueBrain> we are back to the "stuff" again :P
21:13:50 <TrueBrain> but I mean to say: you do have ops :)
21:14:09 <TrueBrain> companies I worked at, devops also took care of things like CI/CD, ticket system, and even customer support
21:14:09 <andythenorth> yeah we have the work, we just make people do it on top of the rest of their job
21:14:44 <glx> at one point someone will decide to automate it
21:14:59 <TrueBrain> okay .. so does action2 chains "return", or does it do a call in the end which publishes the return value or something .. hmm .. lets find out
21:15:39 * andythenorth fell in a devops salary checking hole
21:15:39 <glx> when set it's a return, else it's a chain to next
21:15:47 <andythenorth> devops salaries have gone weird
21:16:00 <TrueBrain> you might be underpaying the rest of your staff :P
21:16:02 <andythenorth> they were £80k-120k a few years ago here
21:16:14 <andythenorth> now they are a lot less
21:16:27 <TrueBrain> devops is more mainstream?
21:17:15 <andythenorth> or more people advertising who can't afford £80k
21:17:17 <TrueBrain> glx: doesn't seem to be the case always
21:17:22 <andythenorth> stats don't say the job was actually filled
21:17:29 <TrueBrain> this action2 industry chain I am looking at, just ends with a "format version 2"
21:17:49 <glx> ah yes it can be a spriteset too
21:18:10 <frosch123> TrueBrain: va2 ranges are "jump" not "call". you can also have real"calls" inside the chains, but that is more magic
21:18:12 <glx> but that's still the end of the chain
21:18:38 <andythenorth> action 2 returns action 1 in the sprites case?
21:18:44 <andythenorth> what's the weird loaded/loading thing
21:21:26 <frosch123> TrueBrain: anyway, va2 ranges can be both "jump to <x>" or "return y". < 0x8000 is jump, >= 0x8000 is return
21:21:40 <frosch123> all other action2 are returns
21:21:51 <TrueBrain> and per callback is defined if it expects a return value, I guess?
21:21:56 <frosch123> (the basic action2 with type < 80)
21:22:25 <glx> yup each callback tells what the return should be
21:22:30 <frosch123> uhh.. let's say yes, but the term "callback" is ambiguous in some cases :p
21:23:01 <glx> some are bools, some are more complexe
21:23:25 <frosch123> there is also a special callback result "callback failed", which is returned by linking to an ordinary action2, instead of returning an integer
21:23:58 <frosch123> some callbacks also have output values in specific registers, in addition to the callback result
21:24:25 <glx> 256+ registers (write only from NFO)
21:24:29 <andythenorth> monthly production change, for example
21:24:37 <andythenorth> puts a string in register
21:24:49 <TrueBrain> okay, so this industry production callback .. you end with a call to tell in which registers are the goodies ..
21:24:55 <TrueBrain> guess that is a solution to returning more than 1 variable
21:25:40 <frosch123> TrueBrain: "industry production callback" is one of the two ambiguous "callbacks" :) it's not really a callback
21:25:44 <andythenorth> ^ optionally returns a string in register 0x100 bits 0..15 and another value in bits 16..23
21:25:51 <frosch123> it has no integer result like the other 100 callbacks
21:25:58 <andythenorth> ho maybe I need to buy a nocode platform
21:26:00 <TrueBrain> frosch123: it is just a procedure, basically?
21:26:05 <andythenorth> then maybe hire someone to write code for it
21:27:10 *** Samu has quit IRC (Ping timeout: 480 seconds)
21:27:27 <TrueBrain> that line .. checks for callback 0x0, which doesn't exist?
21:28:54 <glx> 0 means graphics ie basic action 2
21:29:35 *** Progman has quit IRC (Remote host closed the connection)
21:29:43 <TrueBrain> owh, 00 means "not in a callback"
21:30:13 <glx> nml produce commented nfo, but not everything can be commented
21:31:17 <glx> usually to understand a block of nfo you have to follow the jumps
21:31:29 <TrueBrain> it is not NFO I am not understanding glx :)
21:32:20 <TrueBrain> frosch123: the first 3 lines are very useful :)
21:33:43 <frosch123> the meaning of var10 and var18 depend on var0C :)
21:33:59 <TrueBrain> so it mentions .. but I cannot find what it means when var0C is 0 :D
21:35:05 <TrueBrain> okay, tnx frosch123 :D This really is a nice hunt for information :)
21:36:12 <frosch123> at least we no longer have to read ttdp assembly code
21:37:07 <TrueBrain> okay, so 0C = 0, 18 = 0 means a train arrived and you can decide what to do
21:40:26 <TrueBrain> I love that you can just smell the iterative way NewGRF is created :)
21:40:39 <TrueBrain> okay, so industry production bla is the exception in it all, cool
21:41:43 <TrueBrain> "(was width of "€" character) "
21:42:14 <frosch123> i reverse engineered that at some point. it used to be "reserved"
21:42:45 *** tokai|noir has joined #openttd
21:42:45 *** ChanServ sets mode: +v tokai|noir
21:43:11 <TrueBrain> "Has always all bits set; you can use this to make unconditional jumps " <- how nice :D
21:44:24 <frosch123> haha, that is underselling it
21:44:31 <glx> heavily used in nml to create any number
21:44:35 <frosch123> you also need it for creating all constand numbers
21:44:50 <TrueBrain> lol, there is no other way to create a number? :D
21:45:06 <glx> var 1A & the number you want
21:45:11 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:45:14 <frosch123> no, constants are ((var1A >> 0) & value)
21:45:19 <glx> then you can do math with it
21:45:28 <TrueBrain> that is just awesome :D And indeed, underselling
21:45:48 <TrueBrain> var1C, result from most recent VarAction2 ... which "recent"? Yours? Of the feature you are in? Guess I should look this up :P
21:46:13 <TrueBrain> you can repeat the words, but it still lacks context :D
21:46:47 <frosch123> TrueBrain: the value used to decide that last range
21:46:56 <glx> like you do (a + b) then chain to another action 2, and it can use 1C to get (a+b) and do stuff with it
21:46:57 <frosch123> usually from the va2 that chained to the current one
21:46:59 <TrueBrain> ah, that "result" :D
21:47:22 <andythenorth> does nml have that var
21:47:37 <frosch123> andythenorth: i hope not, it wouldn't work :p
21:47:42 <TrueBrain> awesome, tnx again guys :)
21:48:05 <andythenorth> frosch123 I only just learnt I can just do 'return;' and yield the value from the switch expression
21:48:11 * andythenorth so clueless about nml
21:48:23 <glx> may be needed with procedures (unless we smartly hide it)
21:48:28 <andythenorth> I was putting those results into registers :P
21:48:30 <TrueBrain> "Manhattan distance of closest dry/land tile " <- some variables really surprise me :D
21:48:33 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:48:51 <glx> andythenorth: but return; is in the doc
21:49:03 <andythenorth> I never read those docs :)
21:49:17 <andythenorth> FIRS was ported to nml by other people, and I just copy-paste since then
21:49:40 *** tokai has quit IRC (Ping timeout: 480 seconds)
21:50:18 <andythenorth> I did quit FIRS when it was nml-ported, but PM said I was being a cry-baby and should get over myself :)
21:51:22 <frosch123> and then you took revenge and turned it into pynml
21:53:08 <andythenorth> it did have all those variadic CPP macros in
21:53:09 <glx> pynml is good to trigger nml bugs ;)
21:53:33 <andythenorth> the worst time was when I was using python to generate the variadic macros, during a migration
21:53:38 <andythenorth> that was .... unpleasant
21:54:23 <glx> 60+X var have an extra arg
21:54:24 <andythenorth> I think the peanut gallery here told me "You shouldn't be doing this variadic macro stuff" and I was like "FML, I'm trying to delete it"
21:54:47 <TrueBrain> glx: ah, tnx. That is what they mean with "size"
21:54:57 <TrueBrain> hmm, no, that is not filled in
21:55:51 <TrueBrain> I love this part: "The variables are independent of the number of cargo types and their order."
21:55:56 <TrueBrain> I have no clue what it says :D
21:56:43 <andythenorth> hmm what other bugs did I trigger recently?
21:56:43 <glx> for var 60+X the format is 60 <extra data>
21:56:52 <andythenorth> oh "town registers don't work"
21:56:56 <TrueBrain> how do you know? Besides knowing?
21:57:02 <andythenorth> but that was just because debug window isn't redrawn
21:57:39 <andythenorth> it's nice having a debugger where you have to refresh window :)
21:58:31 <andythenorth> when the register values are incremented on some time period, I have to windowshade the debug window on some precise frequency to see them change :)
21:58:45 <andythenorth> but my sense of timing is....poor
21:59:08 <andythenorth> also sometimes it just garbles the font sprite, so the number is indistinct :)
21:59:15 <andythenorth> this is like RNG in the tank game, adds fun
21:59:53 *** sla_ro|master has quit IRC ()
22:00:30 *** andythenorth has quit IRC (Quit: andythenorth)
22:05:11 *** frosch123 has quit IRC (Quit: be yourself, except: if you have the opportunity to be a unicorn, then be a unicorn)
22:07:44 <glx> that's the hardest part with newgrf, knowing stuff but not knowing where it is in the doc
22:08:30 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
continue to next day ⏵
