Time Nick Message 01:14 hmmmm ugh i totally understand how sapier felt when doing the formspec menu 01:15 hmmmm someone should make a VB6-like drag and drop formspec editor 01:20 VanessaE hmmmm: compounded by having to replicate every last feature of the old C++ menu AND having to rebase about 50 times :) 01:20 hmmmm i can understand how he made so many screwups with it too. 01:21 VanessaE I don't know about a drag-and-drop formspec editor, but I definitely remember seeing something that might make it a *little* easier 01:21 hmmmm thing is, the editor would probably be just as much work.... :/ 01:21 VanessaE damned if I can remember what it was, some tables-to-formspec format that was 10x easier to read 01:22 kaeza VanessaE, stress, by thexyz 01:22 kaeza or a similar concept by sfan 01:23 ShadowNinja It would be mice if formspecs were tables. You could add a Lua function that converts strings to the table format. Unfourtunately that would probably break the way they are sent over the network, making the reverse(what sfan did) easier. 01:23 VanessaE http://sfan5.duckdns.org/minetest-formspec-luatable.txt 01:23 VanessaE that ^^^ 01:23 VanessaE good call. 01:23 hmmmm i forget exactly what, but there was a heated argument about all this 01:24 hmmmm i agree that lua tables are probably the easiest way to represent this data 01:24 hmmmm the current formspec string format could be used only for serialization 01:24 hmmmm but an end user should not have to work with this crap 01:24 VanessaE agreed 01:25 kaeza could be useful too to simulate e.g. expanders or such controls 01:25 VanessaE that obscure syntax is why I suck at making formspecs. 01:25 proller why not serialize to json? 01:25 VanessaE NO! 01:25 hmmmm there's the json again 01:25 hmmmm i predicted it 01:25 VanessaE jesus, let's just get lamefun in on this (re: everything is HTML5+JS) 01:25 proller JS != JSON 01:26 hmmmm well 01:26 hmmmm if i were doing this all over again from scratch 01:26 hmmmm i'd do JS for the scripting and XML for the formspec serialization 01:27 proller why xml? 01:27 VanessaE proller: I was referencing his undying advocacy for an unwanted language, not the language specifically. 01:27 ShadowNinja XML is a bit bulky. 01:27 hmmmm XML is easily compressible, and can represent hierarchies somewhat better 01:27 hmmmm the tools to work with it might be a little easier as well.... json in syntax might be okay, but libjson is pretty gross 01:28 hmmmm UIs are inherently heriarchical 01:28 proller json can represent hierarchies too 01:29 proller and more unambiguously than xml 01:29 ShadowNinja JSON isn't all that bad, it is somewhat similar to Lua syntax. 01:29 proller and json better compressible 01:30 ShadowNinja s/is/has/ 01:30 proller and much faster to [de]serialize 01:31 ShadowNinja Which reminds me that we should have minetest.serialize_json() or similar. 01:31 proller all languages have similar syntax for data represention 01:31 proller i prefer perl way ;) 01:31 ShadowNinja Well, most. 01:33 proller lua->json - its 1-2 days to work or less 01:33 proller already have json->lua 01:33 ShadowNinja Then do it. :-) 01:34 ShadowNinja parse_json should probably be renamed to deserialize_json for consistency. 01:35 PilzAdam if you already want to break formspec compatibilty then use Lua tables and BKVL for network 01:35 proller hmmmm, i'm +1 for js v8 instead lua 01:36 hmmmm hahahaha 01:36 PilzAdam we should ban everyone from the core devs who suggest to replace Lua with JS 01:36 hmmmm i hope you realize that i was speaking hypothetically 01:36 proller js shit too, but less than lua 01:36 hmmmm if you want to change to JS, you're better off starting a whole new project 01:37 proller yes, too hard 01:37 VanessaE for once I have to agree with PilzAdam 01:38 VanessaE don't use a new language for formspecs, whatever it may be. either stick with what we have (yuck) or use Lua. 01:38 VanessaE er Lua tables. 01:40 ShadowNinja But how should we serialize it to send to clients? The current system is terrible, JSON would probably be best IMO as we don't need annother lib and it may be easier to convert. 01:41 PilzAdam ShadowNinja, BKVL 01:41 proller messagepack? 01:42 proller but for rare transfers json can be good 01:42 ShadowNinja PilzAdam: Do we have BKVL? Is someone going to actually merge it if not? 01:43 proller http://msgpack.org/ 01:43 PilzAdam its on the TODO list for 0.5 01:43 ShadowNinja messagepack would be good, but we need another lib. 01:43 hmmmm so this is what things are looking like so far 01:43 hmmmm http://picpaste.com/pics/2013-09-17-214118_802x626_scrot-8XHQ0f0M.1379468577.png 01:43 hmmmm http://picpaste.com/pics/2013-09-17-214131_802x626_scrot-W75CmsNp.1379468596.png 01:43 PilzAdam and celeron55 already has some basic BKVL implemented 01:43 proller but whole protocol must be switched to this 01:44 proller hmmmm, cool! need this for updated math 8) 01:45 VanessaE hmmmm: nice 01:45 PilzAdam hmmmm, I would like to see more support for games setting the mapgen 01:45 VanessaE that's just what we've been needing 01:45 hmmmm there's plenty of support. 01:46 ShadowNinja hmmmm: Is there a way to get the water level from a mod? 01:46 hmmmm yes 01:46 PilzAdam e.g. I want to set settings that are not overrideable by the user 01:46 ShadowNinja How? setting_get won't work. 01:47 ShadowNinja The menu looks good BTW. 01:47 PilzAdam ShadowNinja, on_mapgen_init() 01:47 hmmmm ShadowNinja, minetest.register_on_mapgen_init 01:49 ShadowNinja Alright, because some mods (Like nether) may break if you change that. 01:50 hmmmm i can break the nether mod if i alias all the nodes to air 01:50 hmmmm ..if you do crazy things with the settings, crazy things will happen... that's the way the world is 01:51 ShadowNinja If you add a setting to the GUI like that then mods should be prepared for users that change it. 01:52 ShadowNinja Even if it is just if water_level ~= 1 then minetest.log("error", "This only wouks with a water level of 1") end 01:52 ShadowNinja works* 02:23 thexyz VanessaE: I prefer my format, it has more abstraction https://github.com/xyzz/minetest-stress/blob/master/tests/test_formspice.lua 02:24 thexyz i.e. you don't really need "invsize" element 02:24 thexyz you don't really need to differ label and vertlabel, it's hard to remember all those names, just add "vertical=true" 02:25 thexyz sfan5's formspecs is just another way to write string-based formspecs, it's easier, yeah, since you don't need to do escaping and can write arguments in any order but it could be even better 05:42 sfan5 "PANIC: unprotected error in call to Lua API (bad argument #3 to '?' (table expected, got nil))" what is this? 05:57 kahrl sfan5: when did it happen? 05:57 sfan5 some hours ago on my hungergames server while it(hungry_games) was filling the chests 06:53 hmmmmm i need a hyperlink and an option button element for formspec 07:06 hmmmmm and there's no way to make a label wider..... great. 07:13 hmmmmm what do you guys feel about adding a hyperlink formspec element, for windows it'll use ShellExecute and for everything else it'll use xdg-open 07:13 hmmmmm er how do you do it in Mac? 07:14 hmmmmm ah, "open" 07:14 Jordach hmmmmm, people could abuse the hyperlink 07:15 hmmmmm yeah, of course there should be an option to disable them, and they'll display as plain labels 07:15 hmmmmm also there shouldn't be any text, just the href 07:16 Jordach hmmmmm, a server-side disabling of them would be more effective than a client side (people would possibly still click it) 07:17 hmmmmm http://picpaste.com/pics/2013-09-18-031633_800x600_scrot-fOIgQ3RA.1379488614.png 07:17 Jordach HOLY FUCK THAT'S AWESOME 07:18 Jordach no seed entry? 07:18 hmmmmm that's in the simple settings 07:18 Jordach ah 07:18 Jordach didn't see 10:53 celeron55 >js shit too, but less than lua 10:53 celeron55 lol, lua is like fixed javascript 10:54 celeron55 anyway nice that someone cares to work on the UI more 8) 10:54 Kray js is bretty good 10:54 Kray 5/5 would recommend 10:55 celeron55 there's no valid reason to think that javascript wouldn't be worse than lua other than if one prefers a C-like syntax 10:56 proller {} better than begin..end 10:56 celeron55 javascript has so many design mishaps that cause very obscure errors to happen 10:58 celeron55 of course it would be a better choice because of availability but anyway, this is the standpoint of the language itself 10:58 celeron55 +from 10:59 celeron55 ha someone tried using moonscript for minetest mods? 8) 10:59 celeron55 +s 10:59 celeron55 http://moonscript.org/ 11:00 proller function (function() lalala end) lallaa end 11:02 celeron55 it could make sense to go even so far as to include moonscript in minetest if people like it 11:06 celeron55 (it's not C-like though, but does remove the keyword noise of lua) 11:06 VanessaE heh. writing in js and compiling it to Lua which is then compiled again into whatever you wanna call LuaJIT's output :D 11:07 Jordach languageception 12:53 thexyz celeron55: you can't, it uses metatables 12:53 thexyz and metatables are BAD BAD BAD INSECURE 12:54 thexyz anyway, mediawiki has a Lua extension 12:55 thexyz http://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Differences_from_standard_Lua this has some stuff on how they modified setmetatable, getmetatable 13:18 celeron55 thexyz: sounds like something minetest could possibly do too; dunno 17:15 harrison one prefers to have the scripting language == the language that the renderer itself is coded in 17:15 harrison for minetest this may be impossible 17:16 harrison but it is still worth thinking about 17:16 harrison why is it impossible (if it is) ? 17:17 harrison why the separation of language and scripting language? 17:20 Exio4 rewrite minetest in java and you'll have the thing that you are saying 17:20 hmmmm "one prefers to have"? what? who is the "one" you're speaking of? by "one" you mean "you" right? 17:22 harrison such might be surmised 17:22 harrison ugh java 17:23 harrison is the passive voice now deprecated? 17:26 hmmmm I don't know *anybody* who wants the scripting language to be C++ 17:27 harrison well, that is not an attractive subjuctive to contemplate, certainly 17:28 harrison subjunctive 17:31 harrison but why is this true? is it necessarily true that a system-programming-level language cannot be its own scripting language? 17:31 hmmmm i dunno, but it's certainly not a #minetest-dev topic 17:53 harrison i defer to your expertise 18:22 proller цфффте ьууукпу 18:22 proller waaaant meeerge 19:51 celeron55 is it so that nothing has progressed in terms of getting a wider variety of (larger) trees in the game? 19:51 celeron55 i don't even remember where that was left off last time; was there some technical issue? 19:59 celeron55 i would much rather see that progress rather than for example this proller's weather stuff 19:59 ShadowNinja L-System treespawning uses a older version of the lighting code, which makes bigger trees slow down the mapgen a LOT. I am not sure why, better ask hmmmm. 20:01 celeron55 a way that works speed-wise is to do trees like pilztest does, but that's a bit hacky (but works extremely well once you just get the trees in the format) 20:08 ShadowNinja MTS is pretty fast I beleive, I don't know if you can generate good-looking big trees with it though. 20:32 PilzAdam celeron55, hows that hacky? 20:33 ShadowNinja I thought MTS was designed with trees in mind. But I don't know why we couldn't just use L-Systems. 23:45 kahrl has anybody started working on libmtmap yet?