Time  Nick        Message
01:28 kahrl       These 'No path found for "/base.txt"' lines are rather stupid
01:28 kahrl       rebuildShaders calls generate_shader even for the null shader with name "" and id 0
01:29 kahrl       if no one objects, I'll fix that and add a missing infostream output in afterContentReceived: https://gist.github.com/kahrl/5931110
03:30 hmmmmm      dafuq
03:31 kahrl       ?
03:31 hmmmmm      i think there's a nodebox bug or something:  http://i39.tinypic.com/2lo17he.png
03:31 hmmmmm      i'm just placing them, nothing special at all
03:34 kahrl       what are you placing? default:snow?
03:35 hmmmmm      yup
03:35 VanessaE    what am I seeing here?
03:35 VanessaE    oh the lighting, duh
03:42 hmmmmm      actually forget that
03:42 hmmmmm      the hell is going on.....
03:42 hmmmmm      it's just way more noticable with snow nodes i guess
03:48 kahrl       I changed mgv6 to generate default:dirt_with_snow with default:snow on top and it has no lighting glitches
03:49 hmmmmm      yeah, sorry about that, nevermind
03:52 kahrl       actually there are occasional patches of snow that are brighter than the rest
03:52 kahrl       the one I see right now is at the top edge of a cliff
03:52 hmmmmm      patches... like, on chunk boundaries?
03:54 kahrl       http://i.imgur.com/ObQUUW1.png
03:54 hmmmmm      hrmm
03:55 hmmmmm      that's probably an unrelated bug
04:07 hmmmmm      woooooooooooooooooooww
04:07 hmmmmm      turbodumb
04:08 hmmmmm      i placed this call after calcLighting()
04:19 hmmmmm      some things i don't really like about the addition of snow things:
04:19 hmmmmm      1. why can't you throw snowballs like in the snow mod?  doesn't make much sense to remove something that adds another dimension to the game
04:20 hmmmmm      2. why was the alpha channel removed from the ice texture?  if it's above 128 it'll register as no transparency, and it'd give the added benefit of being easily enabled later on
04:20 VanessaE    2. I believe that was because it's slightly glitchy.
04:20 VanessaE    1. because pilz expected you to only need the nodes for the mapgen
04:21 hmmmmm      3. why isn't alpha transparency on by default for ice?  the other transparent node it'd fight with would be liquid, but it's going *right on top* of water
04:21 hmmmmm      solid ice looks absolutely horrible.  in fact, i'd be ashamed to use this block in my mapgen.
04:21 VanessaE    3. because of the aforementioned glitchiness
04:22 hmmmmm      4. when ice is broken, why doesn't it create a water source block like in minecraft?  if you don't do it, you'll get water flowing *from the ocean* inside of the gap left by digging a part of the ice sheet
04:22 VanessaE    now THAT I can't answer :P
04:23 hmmmmm      so if plizadam doesn't do all of this stuff, i'm going to do it myself.  end of story, no discussion.
04:23 hmmmmm      it's not his own game.
04:23 hmmmmm      this would also be my first ever minetest_game commit
04:24 VanessaE    the general complaint is that he's been treating it as if it were his game.
04:24 VanessaE    and that a game with lots of features and full configurability is...what was his word
04:24 hmmmmm      we trust him to make generally sane decisions
04:24 VanessaE    [07-04 18:16] <PilzAdam> a 100% configureable game would piss me off
04:24 VanessaE    [07-04 18:16] <PilzAdam> since its against the point of a "game"
04:25 hmmmmm      and because most of us couldn't be bothered with the content part of minetest
04:25 hmmmmm      it's just too tedious and it's not as technically intensive, the whole opportunity cost thing
04:25 hmmmmm      somebody else who doesn't mind spending their time doing it ought to do it
04:26 VanessaE    but see, there's a different side to consider:  without game content, the engine becomes little more than an academic exercise.
04:26 hmmmmm      but these are things i need to complete my vision of it
04:26 hmmmmm      not really mine, basically everybody who wants what i want
04:26 hmmmmm      which is nearly everybody to be honest
04:27 hmmmmm      (everybody wants it the way minecraft has it, not because we're copying minecraft, but because they made the correct and sane choices for implementing snow things like this)
04:27 VanessaE    G*d no, I personally don't want us going necessarily the way of MC
04:27 hmmmmm      i know that's your knee-jerk reaction just as everybody else
04:28 VanessaE    well wat
04:28 VanessaE    wait
04:28 VanessaE    don't get me wrong
04:28 VanessaE    MC has stuff that's useful
04:28 hmmmmm      like their grass is green
04:28 hmmmmm      that's a sane decision i'd say
04:28 VanessaE    well surew
04:28 VanessaE    sure*
04:28 hmmmmm      ice is translucent and not ugly... sane as well (they have the same problem we do by the way, it's just not completely buggy)
04:29 hmmmmm      break an ice block, it makes a gap in the water
04:29 hmmmmm      what's the simple way to fix this?
04:29 hmmmmm      replace it with a water source
04:29 hmmmmm      not copying MC, just doing things the sane way
04:31 VanessaE    that begs the question - when do youbreak it, versus merely collecting it?
04:34 VanessaE    but that's trivial really
04:34 hmmmmm      i'm not good with the graphics stuff, so i'll just ask:
04:34 hmmmmm      is it possible for *us* to fix the z buffering issue?
04:35 VanessaE    when last I talked with c55 about it some months ago, it was declared to be ...well... non-trivial
04:35 hmmmmm      if not, is it possible for *us* to mask out the transparent object like minecraft?
04:35 hmmmmm      and if not, how plausible is it for us to start shipping a custom version of irrlicht with our own patches?
04:35 VanessaE    well frankly,
04:36 VanessaE    I'd opt for that last option, especially after the recent weirdness re: shaders and irrlicht 1.8.x
04:36 hmmmmm      hohoho :/
04:36 hmmmmm      it'd be an absolute last resort to do that
04:37 VanessaE    I hesitate to comment too much further on this, lest I blow some random detail and bring c55 down on me :-/
04:40 VanessaE    aside from the layering glitches, would it have any other benefit?
13:55 hmmmm       so i've been thinking about minetest_game
13:56 hmmmm       right now i'd have to say it's in an abysmal state.  there's only an occasional tweak or whatever, i don't see any development going on.
13:56 hmmmm       let's just take a couple examples of things
13:56 hmmmm       in default/textures i see tnt-related images - but we don't even have any TNT
13:56 proller     i want to commit sponge
13:57 hmmmm       so how about that.  we don't have any tnt, which is, to me at least, a pretty basic element of gameplay
13:57 hmmmm       can't even throw snowballs
13:58 proller     now tnt mod works and nuke
13:58 hmmmm       no initiative has been made to add simple mobs to minetest_game
13:58 proller     nuke hardcore mese very powerful
13:58 proller     current mobs too stupid
13:59 Zeg9        stupid but at least they are here
13:59 Zeg9        they would be*
13:59 hmmmm       it's funny, because there's so much engine development going on and we don't have a very good game to showcase it all
13:59 proller     rain from mobs in my sky server
14:01 Zeg9        I don't know if minetest_game uses half of the available API ;)
14:02 hmmmm       that's because the new api only exists for the specific modmakers that wanted it for their own mods
14:02 hmmmm       and everybody else thinks, "my mod works, i shouldn't have to rewrite part of it so it's any better"
14:03 Zeg9        by the way, will mgv7 have L-system trees by default?
14:03 proller     maybe create fork of minetest_game witout PilzAdam ? 8)
14:03 proller     minetest_pilzadam_free and make all new features to it
14:04 hmmmm       it'll have whatever decorations you'd like
14:05 proller     with tnt now possible to destroy locked chests
14:05 hmmmm       proller, i have other things to do that are bigger issues, it's just that i'm surprised something like minetest_game that's  1). easy to contribute to  and  2). has a lot of people that would like to contribute to it   has nothing going on at all
14:06 proller     we have too small number of players -> small comunity
14:07 Zeg9        nothing gets merged so nobody does anything.
14:08 hmmmm       i think it'd be best if we had a designated minetest_game team to work on it
14:08 hmmmm       a lot of the things that people want merged aren't the greatest ideas, and not merging them is justified
14:09 hmmmm       we need   1). good ideas  2). action
14:09 hmmmm       right now there's neither
14:09 hmmmm       well, I am full of good ideas, if I were working on minetest_game i'd have that shit in tip-top shape
14:10 proller     neither is stealed from minecraft, its good, but we need more underground zones
14:10 Zeg9        +1
14:11 proller     and more mobs in every zone
14:37 proller     maybe select best (useful, ...) mods from forum (maybe via voting) ant push to _game ? like carts
14:41 hmmmm       object->playaer collision needs to be disabled
14:41 hmmmm       sick of this
14:41 Calinou     why?
14:41 Calinou     there needs to be a way to disable it
14:41 Calinou     but it should be kept on by default
14:41 hmmmm       stand under sand, dig it
14:41 Calinou     nodes are walkable by default too
14:41 hmmmm       and it keeps hopping up and down glitchily
14:42 hmmmm       same with snow after you cut down a tree and the leaves start decaying
14:45 Zeg9        Disabling is so much easier than fixing...
14:46 hmmmm       are there any circumstances when you'd like a sand node to not just cover the object on the ground?
14:47 Zeg9        No, but with mobs or vehicles...
15:07 PilzAdam    hmmmm, I have a pull request that disables object <-> object collision for falling and dropped nodes
15:08 PilzAdam    and re minetest_game: the problem is that we dont have a goal there, and if we set a goal 70% of the people will start a shitstorm because its not what they want
15:08 hmmmm       yeah fuck them
15:08 PilzAdam    the problem with minetest_game is that its too central
15:08 hmmmm       this is a horrible excuse for no progress
15:09 hmmmm       people don't complain when minecraft gets a new feature they didn't ask for, and they also don't complain when minecraft doesn't add one of their suggestions
15:10 hmmmm       why should this be different?  because of the "community" misconception? (thanks.... celeron)
15:11 proller     dont disable collision between falling sand!
15:11 proller     ang gravel
15:11 PilzAdam    people already say that I am a dictator; if I merge something they say "you only merge what you want!" or "you only copy Minecraft!", and when I dont merge things they say I am an naysayer
15:11 hmmmm       that's because you seem to be doing it alone
15:12 hmmmm       if there was a dedicated *team* for minetest_game it'd seem better
15:12 hmmmm       people don't complain that i'm the emergethread or mapgen dictator
15:13 PilzAdam    because nobody notices your changes :-p
15:13 hmmmm       except when it's 5000% faster
15:13 hmmmm       (hah, i literally mean 5000%)
15:13 PilzAdam    oh, I hear people complaining about "slow mapgen" once a week
15:14 hmmmm       not my fault they use shitty mods
15:22 celeron55_  PilzAdam: did you read the discussion a couple of days ago where i talked about gameplay stuff?
15:23 PilzAdam    yes
15:23 PilzAdam    I am currently experimenting with a new game
15:24 celeron55_  one thing that will solve a lot of "dicatator" accusations is to publish a rough plan for the future so that people know what is coming
15:24 celeron55_  the "dictator" feel comes when nobody knows what to expect
16:52 sapier      can someone tell me who decided to use [ for furnace textures? I've got a rocket ready to shoot her/him to moon
16:56 hmmmm       yeah i heard about that, apparently it broke
16:56 hmmmm       formspec strings are poorly designed and i wish it could be done over.
16:57 sfan5       couldn't formspec use lua tables
16:57 sapier      of course  .... [] are key characters for formspec to make them available for text I had to escape them ... not knowing someone beeing crazy enough to make different parsing style for each single element
16:58 sapier      formspec could use amost everything but it can't be really cleaned up without breaking compatibility
16:58 hmmmm       dude you can escape [ and ] with \
16:58 hmmmm       i added this a while back.
16:59 sfan5       like {{label=true, text="This is a [ text ]", x=0, y=-1}, {listbox=true, items={"foo", "bar"}, x=3, y=0, width=25, height=125}}
16:59 sfan5       ^ much easier to understand IMO
16:59 hmmmm       of course
16:59 ShadowNinja sfan5: Or better, a type attribute.
16:59 sfan5       but that requires every mod to change and after hours of discussion that resolves to "no!"
17:00 sapier      if you want to blow up syntax I suggest using a language designed for things like that
17:00 hmmmm       who originally conceived of the idea of formspec
17:00 sapier      I hate to say but xml is the way to go in that case
17:00 sapier      but that won't happen
17:00 hmmmm       indeed, i was going to say XML here
17:00 hmmmm       don't see what's wrong with Lua tables though
17:01 sfan5       sapier: if it would be added there would a lua function that converts the text format to lua tables format
17:01 hmmmm       that's what i would've done
17:01 sfan5       why would we need to add XML parsing if it can be done with lua tables
17:01 sapier      it just doesn't fix the problem
17:02 sapier      problem is the text format beeing not exact enough and sometimes crazy to parse
17:02 hmmmm       well then what is the problem if that doesn't fix it?
17:02 sapier      I don't see any use in adding a now format to have even more complexity
17:02 hmmmm       yes, but why do we need a text format in the first place?
17:03 sfan5       lua tables are better since the attributes are named, not like "1;foo;5;12;-1;bar" and you don't know what is what
17:03 sapier      it's sent as text string from server to client
17:03 hmmmm       you don't see me running around making text formats for ores, do you?  "default:stone_with_coal[1;6;20;default:stone,default:desert_stone]"
17:03 hmmmm       lel
17:03 sapier      lua tables just don't help anything as the gui is built in c
17:03 sfan5       :D
17:03 sapier      and I don't want to write a parser for lua tables ... so if you volonteer to write it ;-P
17:03 hmmmm       sapier, lua tables can be serializes no problem
17:04 hmmmm       hey hold on a minute
17:04 hmmmm       we're just saying "what if", nobody is making anything new
17:04 sapier      I know ;-)
17:04 hmmmm       but the thing is
17:04 sapier      I just say problem with lua tables is getting them to c++
17:04 hmmmm       in the network protocol we can use binary data
17:05 hmmmm       we can have all fields present and so on
17:05 sapier      for xml there are plenty of libs out there that can parse them
17:05 sapier      wrong issue hmmmm
17:05 hmmmm       so serialization of fields we expect in a lua table would be trivial
17:05 hmmmm       well then what is the issue?
17:05 sapier      not sending is the problem but interpretation
17:05 hmmmm       because i am seeing literally no issue here
17:06 sapier      exactly no issue on sending it's no difference if it's a serialized table or a xml file
17:06 hmmmm       it's no difference
17:06 sapier      the difference is when you want to really build the gui from it
17:06 hmmmm       then why add XML to the mix?
17:06 hmmmm       you're doing it with lua
17:06 hmmmm       there's no way lua isn't going to be part of the process
17:06 sapier      did you ever parse a non fixed format table with arbitrary child elements from c++?
17:07 hmmmm       that's an XML job
17:07 hmmmm       except we're not parsing it, the lua interpreter is
17:07 hmmmm       my point is, in no case would we need to do parsing
17:07 sapier      hmmmm the gui isn't built in lua ;-)
17:07 hmmmm       minetest relies on lua in a fundamental way
17:07 sapier      you need to translate img[] to a irrlicht element
17:08 sapier      we're talking about completely different things
17:08 hmmmm       still don't see how it's different from using lua tables in any of the existing api
17:09 sapier      it's not different it just adds a lot of additional parsing effort in c++
17:09 proller     pleeeeeeease, never use xml
17:09 proller     use json
17:09 sfan5       not json
17:09 proller     yaml
17:10 proller     why not json ?
17:10 sfan5       why use XML or JSON or YAML(which isn't suited for this case) if we can use Lua tables with fields
17:10 sapier      yes of course I'm gonna use a language that I need to write 2 dinA4 pages to get 20 variables .... json is crap unles you write a browser app
17:10 sfan5       nah
17:10 sfan5       but still, there is no need to use some markup language
17:11 sapier      sfan5 because there already are librarys to access elements of a xml document while we would have to implement this on our own for lua tables
17:11 proller     xml is crap everywere
17:12 sapier      xml is pretty good if you want to descripe gui elements
17:12 hmmmm       sapier, still not seeing the problem... we do this in literally all of our lua apis
17:12 hmmmm       it's not difficult at all
17:12 hmmmm       you're making it out to be the hardest thing ever
17:12 sfan5       sapier: thats not a problem, modders won't have to use a markup language but lua tables they can easily work with in many ways
17:12 sapier      hmm do we have a parser somewhere?
17:12 hmmmm       by using lua, we:
17:12 hmmmm       1). reduce bloat
17:12 hmmmm       2). remove an additional dependency
17:12 hmmmm       3). increase speed
17:12 hmmmm       4). make things fit together and look nicer
17:12 sfan5       ^
17:13 sapier      no we don't ... guiFormSpecMenu.cpp will at least tripple size
17:13 hmmmm       you must be not seeing what i'm thinking of then....
17:13 sapier      maybe
17:13 sfan5       otherwise we'll end up having XML for formspecs, YAML for world.mt, <Simple conf format> for minetest.conf and JSON for mapgen settings
17:14 sapier      json in mapgen ....
17:14 hmmmm       yeah this is ridic.
17:14 hmmmm       sapier, proller's stuff
17:14 sapier      as there's no chance to drop current formspec format this discussion is of no use
17:14 sapier      I neither will implement xml nor will I add lua table format
17:15 sapier      I'll fix formspec to do exactly same as before no more no less ;-)
17:17 celeron55_  XML is quite a bloated format
17:18 sapier      And I'm completely with you that using lua tables for gui in mod might be an interesting aproach it just doesn't make life more easy on evaluation
17:18 celeron55_  also, to some earlier questions: formspecs existed looooong before MT had even heard of Lua
17:18 proller     json is good for storing stuctured data
17:19 sapier      is here anyone except proller who likes json?
17:19 hmmmm       json makes sense because of the whole website thing
17:19 hmmmm       and since it's already there, why not use it for other things
17:19 hmmmm       i don't particularily like it though
17:19 sapier      jes on websites but not within a c++ application
17:19 proller     players/* can be 146% better with json instead line-position format
17:20 celeron55_  sapier: can you suggest a very lightweight XML subset library that you would use?
17:20 sapier      proller we need to convert your json parameters at least twice once to c++ second to lua and everytime we need to evaluate the generic style json type
17:21 celeron55_  sapier: also, make an example formspec
17:21 hmmmm       lol roflmao
17:21 proller     xml better than json ?
17:21 sapier      no celeron55_  as I don't really want to add xml support right now if ever this is a mid/long term change
17:21 hmmmm       celeron, it was a simple thought experiment, there were never any plans on replacing formspec
17:21 hmmmm       we just said that formspec was super crappy and were wondering how we would've done it differently
17:22 proller     once to c++ - its one call
17:23 sfan5       I don't think it makes much sense to use json if you have lua and could just use tables
17:23 celeron55_  hmmmm: okay, not changing it to anything then i guess
17:23 proller     lua tables = lua code, its must be runned as program
17:23 celeron55_  sfan5: the thing there is that you can't "just use Lua" when you have to put stuff through network and actually handle the data in C++
17:23 sfan5       celeron55_: yeah
17:23 proller     json - fast parsing format for storing data
17:24 sapier      proller if this is that easy you may be able to cleanup convert_json.cpp
17:24 celeron55_  of course from a mod's standpoint a lua interface could be best, not depending on whatever happens in the background
17:24 proller     json ok for  "through network"
17:24 sapier      it's 300 lines of code for 13 variables
17:25 sfan5       I think we could just have a format that serializes a Lua Table like Minecraft uses NBT
17:25 celeron55_  oh also: do you people know that irrlicht's gui system actually supports an XML format of it's own? 8)
17:26 sapier      a lua table needs to be either parsed on server beeing sent to client in a proprietary binary format
17:26 * sfan5     read smth. about that
17:26 celeron55_  spoiler: it's super crappy too
17:26 hmmmm       sapier, tbh convert_json.cpp is written bloatily
17:26 sapier      or resent to lua engine on client to be parsed into c++ from the lua engine there ... btw we don't even have a lua stack on client while game is running
17:27 hmmmm       a lot of if (big thing here.size()) { \n some_long_thing = big thing here; \n } \n else { blah = false \n }
17:27 sapier      hmmmm ok remove 50 lines still 250 lines for 13 variables is way too much
17:27 hmmmm       which is basically 7 LoC for something that would be done in 2
17:28 sapier      can you give an example hmmmm? I didn't find a shorter way to read what I needed
17:28 proller     sapier, you can make ModStoreModDetails as json::value and delete 300 lines of code.
17:28 sapier      proller no I can't
17:28 sapier      as I need to translate it to a lua table
17:28 proller     like it used in serverlist
17:29 proller     its simpler
17:29 hmmmm       well, not putting the else on a separate line as the closing bracket for an if would be a start (and more fitting to minetest's official code style)
17:29 hmmmm       really a macro would be ideal for this
17:29 sapier      ok doing all those cosmetic fixes still 200 lines will remain
17:29 proller     sapier, guiLuaApi.cpp:   - lua_pushstring(L,current_mod.description.c_str());  + lua_pushstring(L,current_mod["description"].AsString().c_str());
17:30 hmmmm       he needs to fail if it's empty though
17:30 hmmmm       (for some of them)
17:30 proller     typedef Json::Value ModStoreModDetails
17:30 sapier      and I don't want to mix up lua and c++ code again
17:31 proller     Json::Value - its good dynamic data container
17:31 proller     you can use in instead of struct
17:31 sapier      it isn't it doesn't even convert a text to int on its own
17:32 proller     you can make func text2int and use where needed
17:32 sapier      AsInt() won't return a int from a text even if it is possible
17:32 proller     and not use copypaste as now
17:32 sapier      it's your turn proller you tell this can be cleaned up plz do it
17:33 sapier      but don't mixup lua and c again
17:33 proller     ok, but with typedef Json::Value ModStoreModDetails ?
17:33 sapier      no
17:33 sapier      purpos was to have correct type information available directly in c++
17:34 sapier      typdefing is only moving parser from a to b
17:34 proller     for what ?
17:34 sapier      I don't want to bloat another file that already is way to big by "if type is that then"
17:35 kahrl       why does this stuff have to go through the C++ side at all? why not do this all in lua?
17:35 sapier      do we have a json lua parser?
17:35 kahrl       I don't know but it would certainly be useful for lots of things
17:36 sapier      of course we could add a json xml parser too ;-)
17:36 kahrl       take json, return lua table
17:36 sapier      and xml lua
17:36 proller     universal json=>lua maybe good
17:36 proller     and not very hard to make
17:36 sapier      xml => lua will be exactly same ... json and xml are both bloated text formats
17:37 kahrl       what's up with xml? we already have jsoncpp
17:37 proller     but xml is fat, stupid and not usable
17:37 sapier      I prefere files I can read ..
17:37 proller     you can read xml and cant json ??
17:37 kahrl       umm
17:37 proller     its.. strange..
17:37 kahrl       what exactly would change, these things are already in json
17:38 sapier      at least xml tells what you're closing atm
17:38 kahrl       vim % command does exactly that
17:39 sapier      of course I use vim to write big lecture everyday ;-P
17:39 sapier      I don't like xml guys but I json is worse
17:39 proller     sapier, you can use json formater in any editor
17:39 proller     or console
17:40 sapier      http://json.org/example.html is a very small but speaking example ... the 4th } from below is closing what json element?
17:40 hmmmm       thaht is such a contrieved example
17:41 sapier      it's the first one google told me ;-)
17:41 kahrl       why are they mixing spaces and tabs
17:41 sapier      btw it's from json.org ... I guess they need some advertisement lessons
17:43 proller     sapier, http://paste.org.ru/?94ybth - its speaking
17:44 proller     jsoncpp can write formated like in or oneline
17:44 sapier      I'm still waiting for the name of the tag closed with 4th } from below
17:44 proller     "GlossList": {
17:44 sapier      good ;-)
17:45 sapier      I just don't see any benefit for json ... it's almost same as xml just using different separators and a imho less readable syntax
17:45 proller     its 30%+ smaller
17:46 proller     parser much faster
17:46 kahrl       json is almost 100% like lua tables
17:46 hmmmm       XML is horrible
17:46 kahrl       so anyone who knows one can work with the other
17:46 sapier      30% smaller is bouglt by less readability
17:46 proller     default libxml have vulnerability
17:46 hmmmm       if you call MXL more readable...
17:46 hmmmm       XML*
17:46 sapier      and parser beeing faster is just something you tell
17:47 hmmmm       well good thing we don't have plans for this :)
17:47 hmmmm       but if we were to ever work on something in the future where we have to choose between json and xml, we are sure as hell not using xml
17:47 sapier      I don't see anything that would explain a difference for equally well written parsers
17:47 proller     xml have 2+ ways to write one thing (attr, tags, mix)
17:48 sapier      on the other hand in json you can only decide a document beeing invalid on last char
17:48 proller     in any program json much simpler than xml
17:49 proller     in xml you must handle every fucked tag
17:49 sapier      ok make modders write gui in json I can imagine their reaction ;-)
17:50 kahrl       what?
17:50 kahrl       I thought we were talking about the serialization format
17:50 kahrl       the lua api would be tables
17:51 kahrl       no one would ever see the json unless they were debugging or working on the formspec engine
17:51 sapier      converting back and forward 20 times?
17:51 proller     i will try remake convert_json.cpp but on next week
17:52 sapier      nonono imho whatever user writes needs to be directly formated to gui elements
17:52 proller     sapier, no big problem to convert
17:52 sapier      no hurry proller it's working as is it's just ugly
17:53 proller     i can make 300% less ugly
17:53 proller     or not ugly
17:54 sapier      if you want to transform lua tables to json on server just to transform it to gui elements on client that's just adding an additional error location
18:11 sapier      am I really the only one using a debugger to find bugs?
18:11 sfan5       no
18:11 sapier      so how can anyone debug with -O1????
18:12 sapier      this debug build option used by minetest is just useless
18:12 sapier      in best case it's a verbose build but NOT debug
18:13 sfan5       shouldn't ''' ([A-Za-z_0-9]+)\[(<[A-Za-z0-9_ ]+>[,;\]])+ ''' match image[<X>,<Y>;<W>,<H>;<texture name>] and put them all into groups?
18:14 kahrl       sapier: it still allows me to get a backtrace in case of a crash, which is all I need usually
18:15 sapier      I usually debug step by step and you can't step if all your variables are optimized and operations are reordered
18:15 kahrl       step by step debugging is too tedious for me
18:15 sapier      it's best way to find out whats happening
18:17 sapier      don't know what you're aiming at sfan5?
18:18 sfan5       sapier: I wanted regexes to do something faster for me, but that regex only catches the "texture name" entry instead of all entires
18:18 sfan5       entries*
18:18 sapier      I see
18:30 sapier      did someone ever try do escape [ by using \?
18:34 sapier      I'm asking because I don't get any \ passed from lua to c++
18:34 sfan5       maybe lua realizes that \] is not a valid string escape sequence and ignores it?
18:35 sapier      hmmm ... so we need at least doubleescaping
18:36 sapier      lol ... "\[" in lua is "[" in c++ but "\\[" in lua is "\\[" in c++ ... any explanations?
18:37 sfan5       wat
18:37 sfan5       that makes sense
18:37 sapier      wait ... guess this is special display format by debugger
18:38 sapier      yes ... stupid eclipse debug notation
18:48 sapier      PilzAdam are you sure escaping has been done by \ before?
18:50 hmmmm       \\[ should be \[ in C++
18:50 hmmmm       hrmm
18:50 sapier      it is ... while if using \[ \ is ignored completely
18:50 hmmmm       yes..... \[ in lua should be [ in C++
18:51 hmmmm       that only makes sense
18:51 sapier      but [ in lua is  [ in c too ;-)
18:51 hmmmm       [ in lua should be part of the formspec identifier
18:51 sapier      should
18:51 hmmmm       while \[ is a way to use [ in a string
18:52 sapier      but someone was as crazy to use [ in texture name too
18:52 sapier      no it isn't hmm you need to write \\[ to use [ in a string
18:52 hmmmm       are you sure???
18:52 hmmmm       that's not the way it was designed
18:53 hmmmm       oh
18:53 sapier      unless we reparse the format string in lua and replace \[ by \\[
18:53 hmmmm       i mean the data representation of \[
18:53 hmmmm       the actual two characters
18:53 sapier      I don't see any difference between \[ and [ from c++
18:53 ShadowNinja hmmmm: It might be LuaJIT escapes.
18:53 hmmmm       yes, those are luajit escapes
18:53 hmmmm       so you need two levels of escaping here
18:54 hmmmm       one to escape the backslash from within lua
18:54 sapier      so a formspec that works on luajit doesn't work on lua?
18:54 hmmmm       and another to do the actual formspec parsing escape
18:54 ShadowNinja Will that work in regular Lua?
18:54 ShadowNinja +
18:54 hmmmm       yes it should
18:54 sapier      it'd add an andditional \
18:54 hmmmm       why did you say "it might be luajit escapes" then?
18:55 ShadowNinja And +1 for table formspecs.
18:55 sapier      not an option atm shadowninja ;)
18:55 ShadowNinja I know some escapes had to be changed in builtin. I think they were \[ and \].
18:56 ShadowNinja s/changed/removed/
18:56 sapier      with plain lua I need at least two \ to get a single one to lua
18:56 hmmmm       that's pretty reasonable
18:56 sapier      yes but is it same for luajit?
18:57 hmmmm       it should be.  how come you don't know?
18:57 hmmmm       if it's not, then luajit has a *serious* problem, not following the lua language spec
18:58 sapier      I don't have lujit installed
18:58 sapier      I just gave up to check every possible build variant of minetest
19:01 sapier      ok I'm gonna add a workaround for image[ to support unescaped [ within image name ... that's ugly but possible
19:02 sfan5       after spending much time on replacing that regex with a python program that does the same; here is what the regex should have helped me with: http://titanpad.com/oRKZzLBSg7
19:03 sapier      ok now start coding the table to json/xml parser and then the json/xml to irrlicht gui interpreter
19:04 sapier      alternatively just serialize, implement a client side lua stack and then write a lua2irrlicht interpreter ;-)
19:05 sapier      btw you forgot the more challanging elements
19:07 ShadowNinja sfan5: Definitely should be type="foo". How do you handle image=true, list=true?
19:08 hmmmm       what on earth did i do :(
19:08 hmmmm       people are taking this very seriously
19:08 sapier      if we are redesigning formspec I demand container formats
19:09 sapier      hmmmm as long as no coder does write it nothing is gonna happen ;-)
19:09 hmmmm       this is something we don't want
19:10 sfan5       ShadowNinja: the [type]=true thing is a dumb idea
19:11 sfan5       hmmmm: we don't want? IMO forspecs would be much easier to make with tables
19:12 thexyz      make a lua function which converts your table thing into a string
19:12 ShadowNinja We can keep the old code for a release or two to give mods a chance to update.
19:12 hmmmm       great, but, don't change the entire universe because it might be better
19:12 sfan5       thexyz: was about to do that
19:12 hmmmm       formspecs right now are sufficient and it doesn't need messing around with
19:12 sapier      ShadowNinja I already told you just creating table is only about 5% that needs to be done
19:13 hmmmm       it's not what minetest 'needs'
19:14 PilzAdam    hmmmm, re throwing snoballs: for the same reason we dont have mobs (yet), its just too laggy to wait for the server and it wouldnt be usable at all on slightly laggy servers
19:14 hmmmm       how is it possibly laggy
19:15 PilzAdam    all object handling is done server side
19:16 PilzAdam    no client prediction at all (when creating, setting velocity, or hitting nodes)
19:16 hmmmm       yes, that's fine....
19:16 sapier      that's wrong pilzadam acceleration is predicted by client
19:16 hmmmm       ah see, even better
19:16 sapier      that's nut much of course
19:17 PilzAdam    sapier, Im talking about the creation and "hitting a node" part
19:17 sapier      https://github.com/minetest/minetest/pull/806 plz test and merge before mor ppl get upset
19:17 hmmmm       so?
19:17 hmmmm       enough with the "it's too laggy" thing
19:17 sapier      that never can be done client side
19:17 hmmmm       "it's not even there" is worse
19:17 hmmmm       you're expecting the impossible out of minetest
19:17 sapier      mob hitting a node?
19:18 hmmmm       sapier, he wants object node collision to be predicted as well
19:18 sfan5       what?
19:18 sapier      you can't expect a mob to react within ms if you tell server only every 20s ... neither will any other player react within ms in this case ;-)
19:18 PilzAdam    hmmmm, also re ice: its not transparent because its just too glitchy, it looks horrible above water; it doesnt create a water source when destroying it because thats illogical, you get the ice block in your inventory and dont melt it
19:18 hmmmm       all in all, for the people who don't have lag problems or those who play singleplayer, you're preventing them from having decent gameplay
19:19 PilzAdam    hmmmm, no, I want snowball -> snow node prediction
19:19 sapier      hmmmm mobf already does predict object collision server side
19:19 sfan5       PilzAdam: that'd require it hardcoded, or prediction data sent to the client
19:19 PilzAdam    or client side Lua
19:19 hmmmm       creating a water source when destroying an ice node is necessary, so holes of water don't form in the freaking ocean
19:19 hmmmm       that my friend is what's illogical
19:19 sapier      we don't even have a on_collision callback on server ;-) but I already was thinking about adding it
19:19 hmmmm       i said this about 3 times
19:20 PilzAdam    hmmmm, you can create wholes in frozen seas
19:20 sapier      sfan5 it's not that impossible it seams
19:20 sapier      once client side lua is added simple things like that could be done
19:20 sapier      -seams + seems
19:21 hmmmm       ice above water, i didn't realize it was that glitchy until i tried it out myself.  it seems to flicker between transparent and not transparent within whole mapblocks for some reason which you don't see at all on an individual basis
19:21 hmmmm       snowballs and the ice thing, there's no excuse for that
19:21 hmmmm       and if you don't want to do it, i'm going to do it
19:21 sapier      what's up with snowballs?
19:22 hmmmm       when you dig a snow node you get a snowball
19:22 hmmmm       which you can't do anything with
19:22 hmmmm       normally you'd be able to throw it but this was removed
19:22 sapier      why not throw? as in throwing mod?
19:22 sfan5       ; can't be escaped in formspecs, can it?
19:22 sapier      now it can
19:22 PilzAdam    hmmmm, go to a frozen sea IRL and break some ice, there is no water magically appearing
19:22 PilzAdam    hmmmm, there is water _under_ the ice
19:22 sapier      but I haven't tested
19:23 hmmmm       yes, when i break part of the ice sheet it doesn't pour in at all
19:23 hmmmm       it doesn't matter if it's realistic or not, it matters so it doesn't make the ocean look odd
19:24 PilzAdam    it would be interesting to break the first layer of ice, then wait until the water under it freezes and continue that, until you get a nice whole
19:24 PilzAdam    *hole
19:26 hmmmm       between the cheddar cheese blocks for copper, american cheese for gold bars, can't mine a cactus, can't throw snowballs because "it's too laggy", breaking ice deforms the freaking ocean, i don't know which i should fix first
19:26 hmmmm       minetest_game is broken and it seems like you don't care to fix it
19:27 PilzAdam    ehm, ever punched a cactus with your hand IRL? of course you need an axe
19:27 PilzAdam    feel free to create better textures, though
19:28 hmmmm       i really wish i could just do it all myself and get it over with
19:28 hmmmm       but that takes time and skills that would take even more time to develop
19:28 hmmmm       and i'm going to start working and have absolutely no time at all for minetest after that
19:29 hmmmm       so i'd rather spend what i have left on finishing what i immediately have to deal with
19:30 PilzAdam    you seem to be one of those guys who think that everything that differs from MC is bad
19:30 hmmmm       no, it doesn't take one of those people to see that what is currently there is shit.
19:30 sapier      does punching a cactus by hand damage player?
19:30 PilzAdam    sapier, nope
19:31 hmmmm       i really wish i could entrust that these things would get fixed by somebody
19:31 PilzAdam    but punching a rock doesnt damage you either
19:31 hmmmm       somebody that isn't me
19:31 PilzAdam    s/fixed/changed
19:31 sapier      maybe if it did need for an axe would be more straight forward?
19:31 PilzAdam    since nothing is broken
19:31 hmmmm       fixed.
19:31 hmmmm       pilzadam, you can use minimal and nothing is broken
19:32 hmmmm       pilzadam, this is why people see you as a "dictator", because you're not receptive at all
19:32 hmmmm       do you not see it in this very conversation?
19:33 hmmmm       everything i've said so far was shot down for really stupid and irrelevant reasons
19:33 PilzAdam    what I see is that daily people come and want to change minetest_game in what they think is the best
19:33 hmmmm       and it's not even your game
19:33 hmmmm       if this was your own personal game for minetest i'd say that's alright
19:34 PilzAdam    so broken z-buffering is "no reason"?
19:34 hmmmm       i recanted that request
19:34 hmmmm       how about all other 30
19:35 PilzAdam    I cant change the textures since I cant draw them
19:36 PilzAdam    and for the ice -> water source thing I cant see how removing parts of the ice will "deform" the ocea
19:36 PilzAdam    *ocean
19:38 hmmmm       i am hereby officially calling for another to work on the minetest_game component
19:38 hmmmm       anybody interested...?
19:40 PilzAdam    hmmmm, celeron55_ asked for more people to work on minetest_game months ago
19:40 hmmmm       not loud enough apparently
19:40 hmmmm       i should try the forums
19:40 PilzAdam    I guess it would be best if you ask them to post their "plans" for minetest_game
19:41 hmmmm       yes, he wants long term plans, but i'm just talking about a couple of simple fixes (simple for someone who has good enough knowledge of the modding api)
19:42 Taoki       I kinda do. But not sure how reliably I can work on stuff now
19:42 PilzAdam    hmmmm, so... you just ignore that I dont see the point in your "fixes" and ask someone else to do it?
19:43 hmmmm       because i explained it to you as much as i can and you keep saying the same thing
19:43 PilzAdam    I explained my point too
20:00 celeron55_  wat
20:02 VanessaE    who?
20:03 celeron55_  i'm confused about the tininess of the things you are arguing about like they were everything that exists
20:04 sapier      I guess those tiny things are only placeholders for the real problems :-/
20:06 VanessaE    celeron55_: where I come from, it's a generally accepted fact that it's "the small stuff" that has the heaviest impact on something or someone, versus "the big issues", because "the small stuff" tends to pile up far worse.
20:11 celeron55_  but many of the mentioned small stuff is also highly based on preferences of each person
20:12 celeron55_  anyway, the situation of minetest_game is quite stuck in every aspect
20:12 sapier      that's what a moderator is needed to find a common base that everyone is willing to accept
20:12 celeron55_  i would solve this if i knew how
20:13 celeron55_  no, there isn't anything that everyone would accept
20:13 sapier      only accept not like
20:13 celeron55_  well maybe, except that saying it like that gives too much power to those who don't accept things easily
20:14 sapier      another option is to find a way to manage differences
20:14 proller     btw i want to add sponge for liquid_finite- https://github.com/minetest/minetest_game/pull/185/files
20:14 VanessaE    there's an easy way to fix this.  Add every option everyone's been wanting (WITHIN REASON) but leave most of them disabled by default.
20:14 PilzAdam    VanessaE, ehm, why not keep them as mods then?
20:15 VanessaE    because people are idiots and can't even figure out how to rename a folder to make a given mod work?
20:15 sapier      adding all options is a big challange to architecture maintenance gets more and more difficult
20:15 PilzAdam    VanessaE, thats not a good reason
20:15 VanessaE    I can't begin to count the number of times I've fielded that issue just on the half a dozen or so mods I maintain.
20:15 sfan5       thexyz: here you go: http://pastie.org/8113749
20:15 VanessaE    it's a perfect reason.  It's 100000x easier to click a checkbox than to download, rename, and install a mod.
20:16 sapier      vanessae that problem is already fixed
20:16 hmmmm       not only that, but it is kind of a big pain to have to install 50000 mods
20:16 hmmmm       because the default game is too crap
20:16 hmmmm       mods should just be "hey that's neat" instead of "i need this in order to play the actual game"
20:16 sapier      modmanager autorenames mods installed from zip to a sane way
20:17 proller     its hard to select good and not buggy mods from all, base mods must be more stable
20:17 PilzAdam    hmmmm, then you are too limited in the way you use mods
20:17 VanessaE    sapier: we don't have a built-in mod manager though.
20:17 sapier      we do
20:17 PilzAdam    also placing water sources when destroying ice isnt something that is really needed to play the game
20:17 sapier      was added along with formspec menu
20:18 VanessaE    it was?
20:18 * VanessaE  looks
20:18 hmmmm       you also don't need brick nodes to play the game
20:18 hmmmm       it's simply lacking completeness
20:18 proller     but melting ice to water when hot - need!
20:18 VanessaE    again, why does every stinking thing added to the game HAVE to have some active use?
20:18 sapier      yes it's disabled as it's most advanced feature is moddb integration
20:18 VanessaE    (i.e. as hmmmm said, bricks)
20:18 sapier      and moddb isn't ready right now I want
20:18 celeron55_  public announcement: celeron55 is looking for someone who has the courage and the time to pull minetest_game forward without stopping at every naysayer including me, /msg me
20:19 celeron55_  this design by committee simply does not work
20:19 sapier      it to be at least helpfull before enabling modmanager by default
20:19 hmmmm       it's not even design by committee, because if you don't commit it yourself, pilzadam won't commit it
20:19 celeron55_  design by naysayers
20:19 VanessaE    sapier: oh that's seriously busted.  "Configure" button doesn't even work.  Settings tab erases all other tabs when clicked.
20:19 sapier      no that's not modmanager vanessae
20:19 hmmmm       i don't have some kind of "grand plan" for minetest_game
20:19 sapier      and those bugs are already fixed in 806
20:19 hmmmm       i just wanted a couple nagging issues fixed
20:20 proller     modmanager useles on server
20:20 PilzAdam    hmmmm, see? you are like everyone else in the forum
20:20 sfan5       thexyz: fixed http://pastie.org/8113761
20:20 VanessaE    sapier: I know, but I can't even *get* to the mod manager, and I already have #806 installed, unless you've since amended it.
20:20 hmmmm       like you can't mine a cactus with a mese pick even
20:20 hmmmm       i'm sorry, that's just retarded
20:20 PilzAdam    you think that I work on my own, but fact is that I never commit things without at least asking someone else (like VanessaE)
20:20 sapier      have a look at minetest.conf.example
20:20 sapier      modmanager != configure
20:20 VanessaE    PilzAdam: he knows that.
20:20 sapier      modmanager is a separate tab!
20:21 sfan5       sapier: BTW I made a lua thing which converts the lua table thing to the string format http://pastie.org/8113761
20:21 hmmmm       pilzadam, if i'm saying it and i'm not even on the forums, could it be that they were right?
20:21 sapier      wait did you just tell configure doesn't work i 806?
20:21 VanessaE    sapier: I don't see such a tab on mine, and I'm definitely using #806 (your game selector bar is as we last agreed on)
20:21 sapier      main_menu_mod_mgr = 1
20:21 sapier      in config
20:21 PilzAdam    hmmmm, no, people on the forums usually dont see the discussion in IRC, and they see only me in the commit log
20:22 sapier      lol error("Someone started Internet Explorer")
20:22 VanessaE    Ah, that enabled the manager all right.  Good.  Configure button for worlds is still busted though.
20:23 sapier      good idea sfan but now we have three error locations ;-/
20:23 VanessaE    sapier: why isn't that mod manager enabled by default?
20:23 sapier      becaus it's full power is only unleashed after moddb is cleaned up
20:23 VanessaE    ah
20:23 sapier      you need modstore_* parameters from example too
20:24 sfan5       sapier: I just wanted to make as clear as possible what may be passed as arguments with what type
20:24 sapier      without those only install works but not download
20:24 VanessaE    well that solves one issue then of getting mods easily, but it still doesn't fix the main contention about toe overall content of minetest_game
20:25 sapier      sfan5 thatd be another additional way to do it lua -> old formspec format
20:25 PilzAdam    sapier, when creating a new world the world should be selected in the list (so you can hit directly hit "Play") </random>
20:25 sapier      true vanessa .... btw main_menu_game_mgr = 1 could be interesting for you too vanessa
20:25 sfan5       sapier: builtin could hook in front of minetest.display_formspec(?) and check type(argument) and convert if necessary
20:25 PilzAdam    sapier, there is also no error message when a world with that name already exists
20:26 VanessaE    s/toe/the/
20:27 sapier      sfan5 thatd all be possible but still the question is: "Is it worth adding a whole layer adding new errors?" imho it's not but thats a personal preference
20:27 sapier      ok configure/error/selection
20:28 sfan5       sapier: the stuff that is handeled by c++ is usually not so well error-checked, its better when you've got a clear error instead of a segfault/similar
20:29 sapier      vanessae what exactly happens when you hit configure?
20:29 sapier      sfan I just cleaned up error handling for formspecs ;-P
20:30 sapier      configure button works for me vanessae can you show me how it's looking for you?
20:31 VanessaE    sapier: it switches the background from my currently-selected game to the gemeric "minetest" background.  that's it.
20:31 VanessaE    sapier: if I then click on some tab like "settings", it goes to that menu and all the other tabs disappear until I hit escape.
20:32 thexyz      sfan5: this code looks really dirty
20:32 VanessaE    generic*
20:32 sfan5       thexyz: where?
20:32 sapier      hmm sounds like the settings dialog results in some error for you ... windows or linux?
20:32 VanessaE    sapier: also, you need to fix pressing enter to connect to a server (it doesn't)
20:33 thexyz      sfan5: everywhere, actually
20:33 sapier      pilzadam said it's only world connecting ... not a big deal eiter
20:33 sapier      +h
20:33 VanessaE    it's a major deal.  Enter should commit the main action the user would normally expect, without exception.
20:33 sfan5       thexyz: how would you suggest to do it "cleanly"?
20:34 VanessaE    anything else is inconsistent.
20:35 sapier      no it's a 3 line fix so no big deal ;-)
20:35 VanessaE    sapier: THREE LINES!?  are you TRYING to bloat out the engine!?!
20:35 VanessaE    ;)
20:36 sapier      of course didn't you realize I just added 14000 lines ;-P
20:36 thexyz      sfan5: have templates for each element type, like "size[{x},{y}]" for invsize, "image[{x},{y};{w},{h};{texturename}]" for image, etc
20:36 thexyz      then make it process those
20:36 thexyz      well, format obviously can be changed
20:37 * sfan5     doesn't like making templates, because it would be less code
20:37 VanessaE    sapier: :P
20:38 sfan5       s/because/just because/
20:38 sapier      format can't be changed ;-) we need to support old fomspec at least until minetest 0.7 ;-)
20:38 thexyz      sapier: you only read my last message, didn't you?
20:39 sapier      no the last 3 ones
20:39 thexyz      well, I wasn't talking about formspec specification
20:39 sapier      but maybe I misinterpreted what you were trying to say
20:39 thexyz      I mean, sfan5's code is copypasta
20:39 hmmmm       why are you seriously entertaining the whole formspec idea
20:39 hmmmm       it was just a "what if"
20:40 hmmmm       actually carrying that out it misguided effort
20:40 sapier      and noone is willing to do the real work
20:40 thexyz      sapier: what "real work" are you talking about?
20:40 sfan5       thexyz: he means implementing it into the engine
20:40 sapier      what sfan5 currently is doing is just adding another layer of indirection
20:40 thexyz      as I see it, we should make it easier for modders to make formspecs, the idea of converting table-like formspec into string one without changing anything in the engine seems pretty reasonable
20:41 sfan5       I didn't say that code is coming to an use
20:42 sapier      yes if you are a friend of adding additional error sources thats reasonable
20:42 sfan5       <sfan5> sapier: the stuff that is handeled by c++ is usually not so well error-checked, its better when you've got a clear error instead of a segfault/similar
20:42 sapier      as I already told formspecs are 100% error checked by now
20:42 thexyz      sapier: that's better than rewriting everything just because
20:44 sfan5       can someone tell me how tabheader is supposed to work?
20:44 sapier      yes but is it really worth adding so many additional error sources for a benefit that even depends on modders personal preference? We didn't even look for the details that make it even worse
20:44 sapier      l977  in lua_api.txt?
20:44 sapier      caption is the tab name
20:45 thexyz      sapier: crap, I don't get what you're talking about so I'll just give up
20:45 sapier      there are arbitrary number of captions
20:45 sapier      atm the only thing that can be wrong is the formspec the modder wrote
20:45 thexyz      right
20:46 sfan5       sapier: so "tabheader[<X>,<Y>;<name>;<caption 1>,<caption 2>;...;<caption n>;<current_tab>;<transparent>;<draw_border>]" like that?
20:47 sapier      after adding sfans changes after core detected an error user won't have any idea what is wrong as error message is about something he didn't write
20:47 thexyz      sapier: then sfan5's code should do all the validation
20:47 sapier      yes I add caption n in doc
20:48 sapier      thexyz core just cant rely on it
20:48 thexyz      core won't crash
20:48 sapier      if it relied on sfans check it might crash for sure
20:49 sapier      accessing element 5 of a 3 element vector isn't quite a good idea ;-)
20:49 thexyz      I'm not saying we should remove all checks
20:49 thexyz      I'm saying that another layer of abstracion won't hurt
20:49 sapier      I'm for kiss
20:50 thexyz      the first S in KISS is for "Simple, IIRC
20:51 sfan5       wat, drawborder is uses the first time and draw_border the second time
20:51 sapier      yes ... more layers are not reducing simplicity ... but I admit that depends on point of view
20:51 thexyz      it's not "Simple" to write formspecs by hand
20:52 sapier      it is
20:52 thexyz      okay
20:52 sapier      it's just not c++ ;-)
20:53 thexyz      I guess sleep is more productive than fighting over such argument
20:53 sapier      formspecs are (except of some inconsistencys) very very simple
20:53 PilzAdam    sapier, formspec isnt simple
20:53 PilzAdam    and its a PITA
20:54 sapier      no it is simple it's just a flat list of gui elements
20:54 sapier      what can be more simple?
20:54 thexyz      it's simple in terms of code
20:54 thexyz      it's hard to use
20:54 PilzAdam    an layout manager would be simple (see java)
20:54 thexyz      there's just that long string
20:54 thexyz      where all the stuff is located
20:54 sapier      did you ever try to write any gui application without a designer?
20:55 PilzAdam    sure
20:55 sapier      what toolkit did you use ?
20:55 PilzAdam    swing
20:55 sapier      ok I don't know that one ... java crap ;-P
20:56 sapier      all c++ toolkits are way more complex than formspec
20:56 Calinou     suddenly: cube 2 GUIs
20:56 * Calinou   runs
20:56 thexyz      sapier: nice way to win an argument
20:56 sapier      no I just can't say anything about swing and wont learn it just do decide if it's better or even worse
20:57 PilzAdam    sapier, swing has plenty of layout managers
20:57 sapier      ok that's a completely different topic
20:58 sapier      our guis don't have layouts lua table doesn't change that ... of course lua tables would be a way to add it ... but reading sfans changes that doesn't seem to be his intention
20:59 thexyz      sfan5: I suggest you to release this thing as separate mod
20:59 sfan5       sapier: intention to ?
20:59 sapier      add layout based formspecs
20:59 sapier      aka container gui elements
20:59 sfan5       heres one that works with tabheader: http://pastie.org/8113854
20:59 sfan5       ^ sapier: could you confirm that I did it right?
21:00 thexyz      sfan5: I think you should also rewrite it so there's not so much copypasta
21:00 sfan5       ..You mean my intention is to add layout based formspecs? no.
21:00 sfan5       thexyz: if i release this as a mod, probably nobody will use it
21:01 sapier      not all elements are required sfan
21:01 sapier      tabheader looks correct
21:02 sapier      you should have a look at 806 I updated docs added missing elements and parameters
21:02 thexyz      sfan5: well, then I can only hope you won't include it into minetest/minetest without cleaning up this code
21:02 sfan5       thexyz: its ok, although I don't like templates I'll clean it up how you want it
21:03 VanessaE    sapier: please squash your commits on #806 ?
21:03 sapier      I don't want to squash every 2h vanessae
21:04 VanessaE    heh
21:04 sapier      :-)
21:04 sfan5       oh, and my current implementation of formspec would require one change in the engine: being able to use #RRGGBB instead of those colorkeys
21:06 sapier      I used colorkeys to make it easy ;-)
21:07 Calinou     did sapier implement color codes? :o
21:07 sapier      as I can't remember what color 27733 is
21:07 Calinou     well, yey, useless 24 bit :P
21:07 hmmmm       you should know
21:07 sapier      tell me ;-)
21:07 hmmmm       0xFF00FF for example
21:07 hmmmm       what's a mixture of red and blue
21:07 hmmmm       hrm.. purple!
21:08 sapier      and what is 27733
21:08 VanessaE    magenta :P
21:08 hmmmm       is that hex or not?
21:08 VanessaE    0x027733... a bit greenish-blue :)
21:08 Calinou     tip: gcolor2
21:09 sapier      I don't want to start gimp just to find out what colornumber to use ;-)
21:09 Calinou     gcolor2 isn't gimp
21:09 hmmmm       the flexibility ought to be there
21:09 Calinou     flexibility?
21:09 thexyz      wow
21:09 sapier      no it isn't
21:09 thexyz      crap, how should I phrase it
21:09 sapier      it's the small issues that make programming a pain
21:09 hmmmm       well say, if # is prefixing the color, expect an RGB color code
21:10 thexyz      you mean, there's no way to use a specific color?
21:10 hmmmm       if the other thing, expect the constant
21:10 sapier      not an option as # is already used to identify a color key
21:10 hmmmm       whose bright idea was that..
21:10 hmmmm       alright, then, ##?
21:10 sapier      is used to escape a #
21:10 sapier      :-) sorry
21:10 hmmmm       woop woop
21:10 hmmmm       all the sane options are taken
21:11 hmmmm       how about $
21:11 Calinou     +1 for $
21:11 sapier      possible
21:11 Calinou     but why not 12 bit colors instead of 24 bit
21:11 hmmmm       that's not too insane, it's a hex identifier for 6502
21:11 Calinou     like, $f8f for pink
21:11 Calinou     that is enough by far
21:11 hmmmm       because nobody thinks about those...
21:11 sapier      but I won't code it
21:11 Calinou     (ideally, we should also have alpha bits :P)
21:11 thexyz      crap
21:11 Calinou     eg. $f8f8 for half-transparent pink
21:11 thexyz      instead of implementing table in lua with those mapping you did color codes in engine
21:12 sapier      but we need to add $$ to escape $ then
21:12 Calinou     yes, $$ to escape it
21:12 thexyz      which don't correspond to any standart
21:12 sfan5       IMO it should be like "^ color in Hexadecimal format #RRGGBB" --http://titanpad.com/oRKZzLBSg7
21:14 sapier      once all bugs are fixed with formspec menu I may be willing to implement this but not before
21:14 sapier      so if you have pressing need for it you'll have to add it yourself
21:14 thexyz      I may sound a bit celeron55-y but that's just fucking stupid
21:15 sapier      I like the idea even if it requires to add  an additional lookup table to make colors usefull in for old style formspecs
21:16 sapier      vanessae do you play minetest on windows?
21:16 thexyz      no one uses it now, just make it #rrggbb before anyone notices
21:16 sfan5       ^
21:17 VanessaE    no
21:17 sapier      I already said I'm not going to add any new feature anyone wants just to get a single bugfix merged
21:17 sapier      I really wonder why configure is broken for you vanessae
21:18 thexyz      sfan5: can you do that since sapier is busy with bugfixes?
21:18 sfan5       thexyz: after cleaning my code yeah probably
21:18 sfan5       s/code /code, /
21:18 thexyz      sfan5: alright
21:18 sapier      replacing current color scheme by rgb colors isn't quite a challenging task but still some time to do it right
21:20 sapier      vanessae do you have that configure bug for all worlds or only for special ones?
21:21 VanessaE    sapier: I just picked a random world from my (somewhat short) list
21:21 ShadowNinja sapier: minetest.dig_node is broken. I think you were the last to touch it.
21:22 sapier      are you sure?
21:22 ShadowNinja Yes, it fails with an obscure error message and commenting it out fixes it.
21:23 sapier      can you post the error message?
21:23 ShadowNinja I did, in #minetest.
21:23 ShadowNinja Basically error:error:... (null)
21:25 sapier      minetest.dig_node or minetest.env:dig_node()?
21:27 VanessaE    PilzAdam: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/images/default_stone_broken.png
21:27 VanessaE    how's that look?
21:27 VanessaE    er, wrong channel
21:27 VanessaE    meh.
21:29 ShadowNinja sapier: Both.
21:31 sapier      I don't see any error in there
21:33 sapier      how is it called from lua?
21:35 sapier      VanessaE i pushed some minor fixes for modmanager that might be related to your problem ... could you post debug log of config button press somewhere?
21:35 VanessaE    wait, lemme get the latest commits in first
21:53 VanessaE    hrm, perhaps I should have moved that over here.
21:54 VanessaE    sapier: "failed to initialize world" happens for most of my worlds, here's debug.txt (log level 4) for one such instance:  http://pastebin.com/CDELAVEa
21:55 VanessaE    sapier: also, I like the window-in-window thing for the file browser for installing a mod.  nicely done.  Note that you should make it possible to install a mod from an already-extracted folder (I presume it expects zip/tgz/etc archives?)
21:55 sapier      yes it does
21:55 sapier      zip
21:56 VanessaE    sapier: and here's a log from one run where it misinterpreted a single click as a double and started the world without my say-so:  http://pastebin.com/K8hGPD0e
22:00 sapier      VanessaE I don't have any hope to fix it I'm not deciding if it's a single or double click this is done by irrlicht
22:01 VanessaE    ok
22:01 sapier      unless there is some irrlicht setting I don't know about to make it less sensitive I don't see any place to change right now
22:01 VanessaE    there probably is
22:02 VanessaE    it's a pretty basic function to set the double-click threshold, or so I've always seen elsewhere anyway
22:02 VanessaE    (usually for accessibility reasons)
22:04 sfan5       thexyz: does this looks alright? http://pastie.org/8113971
22:11 VanessaE    sapier: also, it keeps forgetting the IP of the last server I connected to.
22:15 sapier      already fixed
22:15 sapier      I'm currently working on pwd for enter btn
22:16 VanessaE    it ain't fixed.
22:16 sapier      not pushed yes ;-)
22:16 VanessaE    so push :)
22:18 sapier      once I fixed the password issue
22:18 VanessaE    k
22:18 VanessaE    ok
22:18 VanessaE    great, now ALL worlds are failing to initialize
22:18 VanessaE    even on my server
22:19 sapier      :-) don't know what you're doing :-)
22:20 VanessaE    18:19:01: ERROR[main]: ERROR: An unhandled exception occurred: ServerError: Failed to initialize world
22:20 sapier      great
22:20 sapier      unhandled :-) I guess some information is missing
22:21 VanessaE    sapier: the server was running fine a few mins ago.  I just did a routine restart to install a texture file.
22:22 sapier      strange
22:23 VanessaE    lemme re-clone without your extra patches.
22:26 VanessaE    sapier: yep, it works fine without #806 in place.
22:26 VanessaE    you bwoke it :(
22:27 sapier      great /
22:30 VanessaE    I wonder if it's 3bda6aa that did it.
22:31 sapier      the escape changes?
22:32 VanessaE    possibly
22:32 VanessaE    grasping at straws here
22:36 VanessaE    I just know that the first two commits in there seemed okay (no actual functional issues), and I don't see how the mod manager crash fix or the enter-for-multiplayer (which still didn't help anyway) would be the cause.
22:37 sapier      do you connect by enter or button?
22:37 VanessaE    by the button, because as I said, enter does not work
22:38 sapier      ok ok :-)
22:39 VanessaE    bbl
22:43 sapier      ok at least I can reproduce
22:48 sapier      argh
22:56 sapier      VanessaE should be fixed
23:11 hmmmm       hmm
23:11 hmmmm       https://github.com/minetest/minetest/blob/922a30ee47edbd6255cd6cbea0f89f21c607094c/src/mesh.cpp#L295
23:11 hmmmm       sapier, did you touch this last?? or who did
23:12 hmmmm       if it fails to create the second image, the function won't fail but mesh will be null
23:12 PilzAdam    hmmmm, click on the "blame" button at the top
23:12 PilzAdam    it was kahrl
23:12 PilzAdam    https://github.com/minetest/minetest/commit/6a76c226e10e92c3e3339096f07f8ab065e2098b
23:12 hmmmm       kahrl, you!
23:15 hmmmm       really the appropriate thing to do here, if img2 == NULL, is to drop img1 and return null
23:19 VanessaE    sapier: please squash those commits
23:20 sapier      once they work I'll do that vanessa
23:25 VanessaE    sapier: worlds seem to work now (no more failed-to-initialize errors)
23:26 sapier      I know it's been a regression due to adding duplicated worldname ... didn't realize a world is always tried to create if it's started
23:28 VanessaE    now, fix the shaders checkbox
23:28 VanessaE    when "true", it should write a '2', not 'true' to the config.
23:29 PilzAdam    wasnt it consensus that enable_shaders will be changed to a bool?
23:29 sapier      I won't add 2 instead of true for sure either it's gonna be a dropdown or shaders will be enabled disabled by true/false
23:29 VanessaE    PilzAdam: sure - when the rest of the system can handle it being a bool.  right now, that is not the case.
23:29 hmmmm       this is supposed to be a stopgap solution to fix the broken behavior until the new shader system is in place...
23:29 PilzAdam    well, there are only one or two places that needs to be changed
23:30 VanessaE    sapier: fix it.  0 or 2 or gtfo
23:30 sapier      definitivly not
23:31 PilzAdam    lol, german_word + ly != english word, sapier
23:31 VanessaE    either way, the checkbox does not work, and if used even once, shaders are disabled until I manually edit the config.
23:31 hmmmm       be sure to fix all the shader levels if you're going to make it a bool
23:31 hmmmm       not writing a 2 will break everything as it is right now
23:31 sapier      ok ok .. not in my lifetime I'll make a checkbox switch something to 2
23:31 PilzAdam    hmmmm, the only shaders we have currently check if enable_shaders >= 2
23:32 hmmmm       yes, but he wants it to write a bool
23:32 PilzAdam    so, changing the checks would be perfectly fine
23:32 hmmmm       which checks
23:32 hmmmm       he'd need to change *all* of the checks for enable_shaders
23:33 VanessaE    sapier: enter on multiplayer tab still doesn't work right - "No world selected and no address provided.  Nothing to do."   (yes, all four fields were filled out as usual)
23:33 hmmmm       it doesn't really matter, RBA is going to change all of that in a little while, but just make sure that it actually works
23:33 VanessaE    sapier: connect button works though.
23:33 PilzAdam    that are only about 9 lines to change
23:38 sapier      VanessaE i was wrong about the doubleclick ... it's even worse irrlicht doesn't support doubleclick at all
23:39 sapier      changed shaders to be a bool setting
23:41 VanessaE    sapier: one of those conflicts with something else
23:43 VanessaE    aa57afe does.
23:44 sapier      who changed shaders recently?
23:44 PilzAdam    RBA
23:44 PilzAdam    he added bumpmapping
23:45 sapier      I guess it's a minor collision to his changes
23:50 sapier      ok rebased and squashed vanessa
23:50 VanessaE    ok
23:51 VanessaE    rebuilding...
23:57 VanessaE    sapier: enter works correctly now, shader checkbox also works as expected now
23:57 Exio        what?
23:57 VanessaE    sapier:  minor glitch:  the "Configure" button should not change the background.
23:57 Exio        bumpmapping? some tweaks to lights too or just with fake-lights?
23:57 sapier      yes I'm gonna fix this tomorrow
23:57 VanessaE    (it changes from the background for the currently-selected game to the default one
23:57 VanessaE    )
23:57 PilzAdam    Exio, fake lights
23:57 Exio        ok
23:58 VanessaE    sapier: it still keeps forgetting the last server address I typed in.
23:59 sapier      ok same thin :-) tomorrow