IRC logs for #openttd on OFTC at 2020-05-25
⏴ go to previous day
00:09:11 *** ChanServ sets mode: +v tokai
00:22:04 <Speeder> andythenorth, why FIRS 3 seemly... totally ignores the ratio variables?
00:22:17 <Speeder> it even checks if they are valid and all but never writes that to the nml
00:29:49 <FLHerne> Speeder: Which 'ratio variables' ?
00:30:26 <Speeder> FLHerne, for example: processed_cargos_and_output_ratios=[('BEAN', 6), ('FRUT', 6)],
00:30:35 <Speeder> code elsewhere complains if the sum is not 8
00:30:47 <Speeder> but seemly this never ends in the NML anywhere
00:34:23 <FLHerne> It should get written into the produce() blocks
00:34:38 <FLHerne> See src/templates/produce_secondary.pynml
00:35:46 <FLHerne> (I think, andythenorth would know better)
00:38:34 <FLHerne> Bleh, FIRS' stupidly huge nml file makes my editor sad
00:40:39 <andythenorth> Speeder is that the brewery?
00:40:53 <Speeder> in that one I pasted yes
00:41:17 <andythenorth> brewery isn't combinatorial
00:41:25 <andythenorth> so it doesn't check the sum is 8
00:42:06 <Speeder> no idea what this means
00:42:13 <Speeder> when something is combinatorial?
00:42:35 <andythenorth> brewery will give 6 out for every 8 in
00:43:01 <andythenorth> but blast furnace
00:43:24 <andythenorth> will give 3 steel out for 8 iron ore in
00:43:42 <andythenorth> or 2 steel out for 8 coal in
00:43:56 <andythenorth> or 3 steel out for 8 scrap metal in
00:44:05 <Speeder> combined_cargos_boost_prod <<< THIS?
00:44:10 <andythenorth> if both iron and coal are delivered, there will be 5 out
00:44:27 <andythenorth> if iron ore, coal and scrap metal are delivered, there will be 8 out for 8 in
00:45:12 <andythenorth> there is a 3 month delivery window
00:45:21 <andythenorth> they don't all have to be simultaneous
00:45:22 <Speeder> so combined industries will always produce 8, assuming fully stocked?
00:45:28 <Speeder> and non-combined can produce less than 8?
00:45:43 <andythenorth> probably yes, I haven't read the actual code
00:46:01 <andythenorth> but that check sounds like I enforced 8
00:48:39 <Speeder> so you didn't enforce 8 for non-combined because when fully stocked they won't produce more than 8?
00:50:31 <Speeder> so the brewery, if you supply 8 of each item, will produce 6 in total, and consume 4 of each item?
00:54:01 <andythenorth> if you supply 8 of each, it will produce 12 and consume 16
01:27:40 <Speeder> what cluster does on FIRS?
07:00:48 <Speeder> is it normal for openttd to spawn more passengers than the town have of population???
07:39:01 <Speeder> so... how I stop towns from erasing rivers?
07:55:53 <Eddi|zuHause> they shouldn't be doing that
08:02:24 *** andythenorth has joined #openttd
08:14:30 *** sla_ro|master has joined #openttd
08:18:03 *** Wormnest has joined #openttd
08:18:36 <Speeder> but they are doing that D:
08:19:28 <Speeder> my map is HORRIBLY unbalanced
08:19:44 <Speeder> one diamond mine is enough to earn me ludicrous amounts of money
08:19:53 <Speeder> and I can't even keep up with its production Oo
08:21:14 <andythenorth> OpenTTD town growth is very confusing
08:21:17 <andythenorth> "Town growth can be accelerated by loading and unloading at least one item of cargo at up to five stations within town influence within a 50 day period."
08:21:23 <andythenorth> that can't be literally true
09:20:53 <Speeder> I think I just saw a buggy AI
09:21:00 <Speeder> building airports in 1780 O.o
10:07:28 *** Wormnest has joined #openttd
10:09:07 <Speeder> I think this is a bug: my cargo want to go to my competitor dock Oo
11:42:06 <planetmaker> Hehe. Is it oil and do you both service the same oil rig?
12:22:38 *** WormnestAndroid has quit IRC
12:22:51 *** WormnestAndroid has joined #openttd
12:27:34 *** WormnestAndroid has quit IRC
12:27:47 *** WormnestAndroid has joined #openttd
12:33:04 *** gelignite has joined #openttd
12:51:18 *** andythenorth has joined #openttd
13:03:48 <FLHerne> andythenorth: Why can't that be literally true?
13:04:07 <FLHerne> I think it should be 'or', but otherwise?
13:04:48 <FLHerne> Oh, that's what you changed :
13:06:39 * andythenorth currently reading town_cmd.cpp
13:07:41 <andythenorth> looking what a newgrf spec for towns might be
13:09:00 <DorpsGek_III> [OpenTTD/OpenTTD] James103 commented on issue #7970: Crash in crash handler - Assertion failed at line 2981 of window.cpp https://git.io/JvsUh
13:27:26 <Eddi|zuHause> andythenorth: that reads suspiciously pre-GS
13:28:55 <Eddi|zuHause> you should probably cut out any part of that that can be achieved by GS, and you're left with something along the lines of "decide which house gets built"
13:29:47 <andythenorth> nah that's not the goal
13:30:01 <andythenorth> the objective is control of growth
13:30:58 <Eddi|zuHause> i mean something like non-circular town zones
13:31:13 <andythenorth> that would be interesting
13:31:17 <FLHerne> So why do you want to do it in a grf?
13:31:43 <andythenorth> how many GS can be packaged in a grf?
13:31:51 <andythenorth> it's not a trick question
13:32:18 <Eddi|zuHause> why would you want to package a GS in a grf?
13:32:24 <andythenorth> we have an accidental API, and we crippled newgrf because of it
13:32:32 <FLHerne> andythenorth: Hm, there's already a dependency mechanism for GS/AI
13:32:59 <andythenorth> this argument gets had every 2 weeks TBH :)
13:33:10 <FLHerne> You install a GS and OTTD automagically downloads/installs the relevant squirrel libraries
13:33:11 <andythenorth> on one side: everybody who likes the purity of GS
13:33:20 <andythenorth> on the other: the authors actually making content
13:33:31 <FLHerne> So maybe the answer is to let NewGRFs depend on a GS and vice versa
13:33:56 <andythenorth> it could be an answer
13:34:03 <andythenorth> nobody's devised a spec for it ever :)
13:34:17 <andythenorth> whereas we have almost total control of the game in newgrf
13:34:26 <Eddi|zuHause> there is no meaningful interface between GS and NewGRF to base a dependency on
13:34:29 <andythenorth> but we decided to abandon the proven content API because now we have GS
13:34:48 <Eddi|zuHause> plus, dependencies like this quickly conlfict with "only one GS per game"
13:35:00 <andythenorth> I have probably discussed 50 times how GS and NewGRF communication could be established
13:35:06 <andythenorth> there has never been a spec
13:35:18 <andythenorth> whereas frosch wrote a spec for newgrf town with no drama
13:35:36 <FLHerne> Eddi|zuHause: Dependency for game-mechanics purposes doesn't imply or require some direct interface
13:36:23 <andythenorth> specific to town control?
13:36:30 <andythenorth> GS is blind to the newgrf
13:36:30 <FLHerne> And "only one GS per game" is clearly a bad idea already
13:38:58 <Eddi|zuHause> i get why you would want tighter integration between economy GS and industry GRF, but why do you need integration between town GS and house GRF all of a sudden?
13:39:40 <andythenorth> I think it's more a case of 'why not?'
13:40:00 <andythenorth> ignore the ideas I have for FIRS, why should ~everything else have a newgrf interface, but not town?
13:40:06 <Eddi|zuHause> "why not?" -> dependency hell
13:40:21 <andythenorth> we managed the rest of newgrf without the sky falling
13:40:34 <andythenorth> give or take that refitting diagram frosch made :P
13:41:01 <Eddi|zuHause> newgrf is a very bad programming language for complex simulation stuff, which is part of the reason why state machines never happened
13:41:43 <andythenorth> it doesn't need a complex simulation, it just runs a couple of callbacks, and checks a few conditions
13:41:59 <Eddi|zuHause> there's no way to program and debug state machines in NFO-style assembler and stay sane...
13:42:00 <andythenorth> I like that it has its own page
13:42:36 <andythenorth> what callbacks would a town run besides growth?
13:43:29 <andythenorth> 2 of those are just text display though
13:44:16 <andythenorth> I did wonder about implement town control as a GS library
13:44:28 <andythenorth> then patching all my favourite GS to use it and re-releasing them
13:44:58 <andythenorth> I think templating squirrel from the FIRS compile might be an interesting project
13:46:37 <andythenorth> only one GS per game, but how many libs? o_O
13:48:36 <andythenorth> we'd need a new global newgrf var: is gamescript lib in current game? (bool)
13:49:10 <andythenorth> GS libs seem to have UIDs
13:51:08 <andythenorth> can a GS end the game?
13:51:44 *** arikover has joined #openttd
13:53:03 <andythenorth> there's no check for a newgrf, but can probably use industry cargo in / out as a fingerprint
13:54:10 <andythenorth> eh, FIRS wouldn't need a newgrf global var to find the GS
13:54:27 <andythenorth> just don't build any industry at map gen
13:54:34 <andythenorth> depend on the GS to do it
13:55:05 <Eddi|zuHause> ... i'm going to stop reading here, or i'll get nightmares
13:55:16 <andythenorth> I think this might be plausible
13:55:29 <andythenorth> it reminds me of the old days, trying to make broken things work
13:55:35 <andythenorth> it was more fun, less productive
13:56:03 <andythenorth> I think I can force the player to install both the GS and the newgrf
13:58:07 <Eddi|zuHause> you can try that, but the amount of players who will successfully manage to install FIRS will be 0
13:58:12 <andythenorth> ok so (1) generate a GS lib from the FIRS compile (2) patch the existing well known GS to use it, then re-release them(2) make FIRS depend on the GS lib
13:59:19 <FLHerne> What does this FIRS GS (lib) actually do?
13:59:20 <andythenorth> probably (4) script uploading them both to bananas when releasing, because I'll forget
13:59:26 <FLHerne> Why does FIRS need to care about towns?
13:59:46 <andythenorth> I want control of town growth
13:59:53 <andythenorth> there's no rationale, I just want it
14:00:53 <andythenorth> the disconnect between towns and cargos has always bothered me
14:01:05 <andythenorth> it's like two parallel games going on
14:01:28 <andythenorth> I have towns being served for pax and mail, purely so I can build stations there
14:01:45 <andythenorth> and I have town-destination cargos which I never bother transporting as they're useless
14:01:56 <andythenorth> just an odd odd game :)
14:04:42 * andythenorth has alternative ideas
14:04:48 <andythenorth> industry production could depend on the amount transported
14:05:08 <andythenorth> if there are 4 cargo types produced, all must be transported
14:05:12 <andythenorth> no transport, no production
14:05:59 <andythenorth> I'd need to patch FIRS to reject incoming cargo also
14:07:17 <FLHerne> That would be interesting, but annoying :p
14:07:48 <FLHerne> I'd enjoy playing at least a few games like that
14:20:22 <andythenorth> I'd need an output stockpile
14:20:29 <andythenorth> capacity maybe 128t per cargo
14:20:37 <andythenorth> that can be done inside newgrf
14:21:16 <andythenorth> but we'd still need GS<->newgrf communication
14:21:22 <andythenorth> only GS can enforce that the cargo is delivered
14:23:40 <andythenorth> hmm maybe I don't care if player wants to just dump the cargo at a station
14:23:48 <andythenorth> not my concern :)
14:28:53 <Borg> andythenorth: played a bit more BSPI?
14:30:38 <andythenorth> I like the approach, I went back to working on FIRS though :P
14:30:52 <andythenorth> it was nice to have something else to look at
14:32:25 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh reopened issue #7970: Crash in crash handler - Assertion failed at line 2981 of window.cpp https://git.io/JvsUh
14:50:32 <milek7> maybe there should be some 'standard' for modular GS?
14:54:18 <andythenorth> all GS as libs, and then just really thin wrappers?
15:18:10 <FLHerne> What does the line `0 * 4 \d8509` at the top of an NFO file mean?
15:20:39 <FLHerne> I built both FIRS and OGFX+ Industries with your patch, compared the output to 0.4.5
15:21:01 <FLHerne> With FIRS it's identical except for using prop 13 instead of 0A/B/C
15:22:29 <_dp_> biggest advantage of GS api is that it is somewhat isolated and doesn't incur that much of a maintenance cost
15:22:52 <_dp_> though as a modding mechanisms both newgrfs and gs suck when it comes to altering gameplay mechanics
15:23:12 <_dp_> I rly wish someone explored the WASM options
15:23:33 <FLHerne> (search for "forest_tile_1_animation"
15:24:02 <FLHerne> I doubt those are directly relevant to your current change, but it would be interesting to know what's happening
15:25:55 <Borg> FLHerne: 0 * is sprite identifier... next is number of bytes (4 means dword) and next is number of sprites
15:26:01 * glx was wondering why notepad++ takes so much time to start
15:26:28 <glx> and I remembered firs.nml was open
15:26:30 <Borg> gvim here start in a blink of an eye ;)
15:31:30 <glx> FLHerne: it's #118 in effect
15:38:04 <andythenorth> _dp_ write a spec :)
15:38:44 <milek7> what API aspect would wasm address, aside from performance?
15:39:45 <andythenorth> it would allow us to incorporate more coin miners
15:39:59 <_dp_> milek7, I'd start with replacing newgrf stuff
15:41:10 <milek7> I suspect that native->wasm transition cost would be non-negligible
15:41:54 <_dp_> milek7, shouldn't be slower than newgrf callbacks I'd imagine
15:43:39 <milek7> dunno about wasm, but luajit didn't recommend such use, I guess it is similiar
15:45:00 <_dp_> milek7, I don't know much about either lua or wasm but having "jit" there already suggest it's entirely different matter
15:45:22 <FLHerne> glx: Confirmed that without #118 it's only the 0A/B/C -> 13 changes
15:49:11 <_dp_> milek7, from what I know wasm is a near-native code sandbox and lua is a embedable scripting language
15:49:15 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on issue #7970: Crash in crash handler - Assertion failed at line 2981 of window.cpp https://git.io/JvsUh
15:50:10 <glx> FLHerne: now the question is "are the animations the same in both versions"
15:54:31 <andythenorth> so is the main problem with newgrf performance?
16:00:16 <_dp_> andythenorth, imo performance is the only thing about newgrf that isn't much of a problem :p
16:00:20 <andythenorth> if so, have we considered moving it to squirrel?
16:02:08 <DorpsGek_III> [OpenTTD/OpenTTD] James103 opened issue #8168: It takes two actions to move your company HQ by one tile. https://git.io/JfViG
16:03:15 <andythenorth> sometimes I wonder if we should turn dorpsgek off for issues
16:03:19 <andythenorth> it's very demoralising
16:09:38 <milek7> it seems call cost is 160ns
16:20:35 <DorpsGek_III> [OpenTTD/nml] FLHerne commented on pull request #152: Codechange: Reintroduce 0.4 syntax for compatibility https://git.io/JfVi7
16:32:07 <DorpsGek_III> [OpenTTD/nml] glx22 commented on pull request #152: Codechange: Reintroduce 0.4 syntax for compatibility https://git.io/JfVPl
16:42:16 *** gelignite has joined #openttd
17:18:44 *** Wormnest has joined #openttd
17:26:09 *** tokai|noir has joined #openttd
17:26:10 *** ChanServ sets mode: +v tokai|noir
17:38:00 <glx> FLHerne: all these reverse_lookup to get name, while the name is available when the function is called
17:38:25 <FLHerne> glx: Yes, it's silly, I'm trying to fix it
17:38:28 <glx> func(id_d[self.value], self.pos) could simply also pass self.value
17:39:02 <FLHerne> The problem is tracing down every case where some callback gets included in id_dicts...
17:39:12 <glx> then all (info, pos) functions need to be fixed yes
17:39:15 <FLHerne> Miss one, and it crashes at some point
17:39:32 <FLHerne> This is where a statically-compiled language would be nice :p
17:40:04 <FLHerne> (well, C function pointers aren't checked anyway IIRC, but...)
17:43:53 <FLHerne> Hm, if *all* of them take `info, pos` it's easy to find them
17:44:00 <FLHerne> But how do I know that for sure?
17:46:15 <glx> for sure there's parse_var, and func60x() lambda, all the functions in global_constants.const_list
17:48:16 <glx> I'm checking all ".reduce(" :)
17:48:48 <FLHerne> I wonder if the right solution is for these big dict things to be instances of a custom class
17:49:30 <FLHerne> So instead of passing around a dict and a special function, we just pass around a dict implementing a custom __getitem__
17:49:43 <FLHerne> Except then `pos` gets lost
17:49:54 <FLHerne> Unless it's passed in
17:53:21 <FLHerne> glx: If you're looking as well, I worry about duplicated effort :p
17:54:08 <glx> well I'm just looking in the code, not coding :)
17:54:31 <glx> it's like a virtual review :)
17:55:31 <glx> based on reduce_constant() and reduce() calls I think the list is complete
17:56:47 <FLHerne> Just the global_constants ones and the two from action2var, you mean?
17:58:36 <glx> there are some calls with [id_dict] but as it's a simple list without function in this case
17:58:52 <glx> else it would be [(id_dict, function)]
18:06:41 *** cit|zen has joined #openttd
18:08:27 <Speeder> I mean, I think it is sad when companies can attend to something like less than 1% of the produced cargo
18:10:27 <DorpsGek_III> [OpenTTD/OpenTTD] JGRennison commented on issue #7970: Crash in crash handler - Assertion failed at line 2981 of window.cpp https://git.io/JvsUh
19:09:38 <FLHerne> glx: I found another one in action2layout :p
19:15:49 <glx> I see the one in action2layout, but not the other
19:39:04 <DorpsGek_III> [OpenTTD/nml] FLHerne opened pull request #153: Codechange: Make id_list callbacks take a 'name' argument https://git.io/JfVHV
19:41:37 <FLHerne> I still don't like the way this works much
20:00:34 *** Progman has joined #openttd
20:04:58 *** Gustavo6046 has joined #openttd
21:12:47 *** supermop_Home has joined #openttd
22:06:10 *** Wormnest has joined #openttd
22:27:36 <DorpsGek_III> [OpenTTD/nml] glx22 approved pull request #153: Codechange: Make id_list callbacks take a 'name' argument https://git.io/JfVbO
22:28:05 <glx> and the patch is quite simple
22:46:22 <FLHerne> glx: It seems silly to pass in both the key and the value, when half the functions are specific to a particular dict anyway
22:46:37 <FLHerne> But none of the alternatives I tried seemed much better
22:47:05 <DorpsGek_III> [OpenTTD/OpenTTD] Chr12t0pher commented on issue #8153: Autoreplace does not report failures due to refittability https://git.io/Jf0oB
22:47:07 <glx> yeah, but at least it removes silly reverse lookups
22:47:38 <FLHerne> glx: Silly bikeshed question, should I s/name/key/ ?
22:51:03 <DorpsGek_III> [OpenTTD/nml] FLHerne merged pull request #153: Codechange: Make id_list callbacks take a 'name' argument https://git.io/JfVHV
22:52:53 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on issue #8153: Autoreplace does not report failures due to refittability https://git.io/Jf0oB
23:08:14 <DorpsGek_III> [OpenTTD/OpenTTD] Chr12t0pher opened pull request #8169: Fix #8153: Report autoreplace failure when new vehicle cannot carry the cargo https://git.io/JfVAv
23:20:02 <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh commented on pull request #8169: Fix #8153: Report autoreplace failure when new vehicle cannot carry the cargo https://git.io/JfVAK
23:34:43 <DorpsGek_III> [OpenTTD/nml] glx22 updated pull request #152: Codechange: Reintroduce 0.4 syntax for compatibility https://git.io/JfVeZ
23:40:57 <DorpsGek_III> [OpenTTD/nml] FLHerne commented on pull request #152: Codechange: Reintroduce 0.4 syntax for compatibility https://git.io/JfVxB
23:44:52 <glx> FLHerne: yeah it prints the number
23:46:59 <FLHerne> glx: Also, that patch has a different wording for the warning, because I thought your version is a bit vague
23:47:29 <FLHerne> (and because I have a personal dislike of computers pretending to be polite)
23:56:08 *** Gadg8eer has joined #openttd
23:56:27 *** Gadg8eer has joined #openttd
23:59:54 *** Gadg8eer has joined #openttd
continue to next day ⏵