Time  Nick        Message
05:57 nore        any thoughts on #966?
13:08 nore        PilzAdam, what do you think of #966 now?
13:08 nore        https://gist.github.com/Novatux/7179966 <-- to test it
13:37 troller     hmmmm, weather fix after your moving - https://github.com/proller/minetest/commit/dda672d03988b8c8581ed80b0d1b8eca1fec2014
16:17 VanessaE    will someone PLEASE FIX THE FUCKING vector math lib so that it asserts properly?
16:17 PilzAdam    what doesnt work?
16:18 VanessaE    [10-27 21:55] <cy1> oy... there's a bug in builtin/vector.lua
16:18 VanessaE    [10-27 21:56] <cy1> yeah, vector.new(0,0,0) will cause that assertion error instead of returning {x=0,y=0,z=0}
16:18 VanessaE    [10-27 21:56] <cy1> You'd only notice it with relative vectors, or really close to 0,0,0
16:18 VanessaE    this is what cayuses the assert() fail on my servers.
16:18 VanessaE    causes*
16:20 PilzAdam    what Lua version do you use?
16:20 VanessaE    I expect other functions in the lib have the same issue.
16:20 VanessaE    LuaJIT, so I guess 5.1
16:20 VanessaE    (2.0.2 or some such)
16:20 PilzAdam    assert(0); local a = 0; assert(a) <- this does not crash
16:20 VanessaE    it does for me and for cy1
16:20 VanessaE    in the vector lib I mean
16:20 PilzAdam    thats weird
16:21 VanessaE    and I've been having these untraceable crashes for weeks now
16:21 PilzAdam    print(_VERSION)
16:21 PilzAdam    ^ can you check your Lua version?
16:21 VanessaE    I just told you, it's LuaJIT.
16:22 VanessaE    (frok git, last commit was two days ago, 8944e27762712679e50962598ed1be8dc1fbe7be)
16:22 VanessaE    from*
16:22 VanessaE    idk what cy1 uses, he didn't say.
16:23 PilzAdam    btw, why do we do assert(v) instead of assert(v.x and v.y and v.z), ShadowNinja?
16:24 nore        PilzAdam, what do you think of the new version of #966?
16:26 thexyz      aaaand since we're talking about lua what's the reason behind https://github.com/minetest/minetest/blob/master/builtin/misc_helpers.lua#L107
16:27 thexyz      instead of something like math.sqrt(x * x + y * y)
16:28 thexyz      ShadowNinja: plz
16:29 thexyz      and maybe it should actually do something like math.sqrt(x * x + y * y + (z or 0) * (z or 0))
16:29 PilzAdam    thexyz, http://en.wikipedia.org/wiki/Hypot#Motivation_and_usage
16:30 thexyz      oh, I see
16:31 PilzAdam    it was kahrl's idea IIRC
16:31 thexyz      that's smart
16:31 nore        so, what about https://github.com/minetest/minetest/pull/966 ?
16:31 PilzAdam    nore, have you found solutions to all the problems I mentioned?
16:32 nore        yes, try it now
16:32 ShadowNinja PilzAdam: That wouldn't work, we need assert(type(v) == "table" and v.x and v.y and v.z) which will be a little slower, although it would ensure the function suceeded.
16:32 nore        even craft preview is changed
16:32 PilzAdam    how did you solve it?
16:32 nore        I added another registration function, to get craft prediction
16:32 PilzAdam    ShadowNinja, right
16:32 nore        and it changes craft preview as well
16:32 thexyz      shouldn't we do something similar for three coordinates instead of calling it two times?
16:33 ShadowNinja PilzAdam:Do you think I should add that?
16:33 PilzAdam    yes
16:34 PilzAdam    nore, so the client has to wait for the server now instead of just stacking up the result?
16:34 nore        it was already like that (the client already had to wait)
16:35 nore        proof: the craft preview code was in server.cpp
16:35 PilzAdam    the client currently only waits for the result once, but the stacking happens without waiting for the server
16:35 PilzAdam    or did I miss something?
16:35 nore        well, now if the craft preview does not change, it is like before
16:35 nore        if for the same recipe, the output changes, there it has to wait
16:36 nore        no, the stacking waits for the server too
16:36 PilzAdam    oh, youre right
16:37 nore        PilzAdam, to test: https://gist.github.com/Novatux/7179966
16:37 nore        first one: put a stack of steel ingots in the center of the craft grid
16:37 nore        second one: put stone in the center of the craft grid, and a stone axe above it
16:38 PilzAdam    of course I wont use your code to test it
16:38 PilzAdam    its more likely to find bugs when I write something else
16:39 nore        well, you can look at the code to see how it works
16:40 PilzAdam    I think this should be redesigned a bit: the engine should call a Lua function to get the craft result, and on_craft shouldnt be allowed to change the output
16:40 PilzAdam    this way clients always show the correct output item
16:41 nore        well, but if you want a random result each time
16:42 nore        and don't want the players to know what they will get
16:42 nore        i.e. I want something to give a random number of wood, between 2 end 5
16:42 PilzAdam    then you create a "random item" as the result and catch the placement in on_move or so
16:42 nore        if the user knows, he will try again and again until having 5
16:42 PilzAdam    hmm
16:43 nore        no, you can't do that
16:43 nore        with the current way, I show wood as craft preview
16:43 nore        and then I give a random number of it when it is crafted
16:43 nore        or, it allows secret recipes
16:44 nore        -> you make an invisible item as output, and if the player clicks it, give the output
16:44 VanessaE    player crafts a steel door, and they might get a bonus gift of a locked chest when they commit to the craft
16:47 PilzAdam    nore, does the client show the result of craft_predict?
16:47 nore        yes
16:47 nore        (that was the hardest part to do)
17:00 nore        PilzAdam, so, does it work?
17:05 VanessaE    so what's going to be with the vector lib?
17:28 nore        PilzAdam, so, what do you think of it now?
17:46 ShadowNinja PilzAdam: Comments? http://ix.io/8MK
18:08 nore        PilzAdam, so, what do you think of #966 now?
18:08 VanessaE    ShadowNinja: http://ix.io/8MK  seems reasonable to me.
20:22 sapier      https://github.com/minetest/minetest/pull/971 please test this one I almost got crazy when trying to fix all variants that may happen
20:23 sapier      and guys next one someone else is up to fixing microsoft compatibility
20:24 PilzAdam    sapier, you are the only person I have ever seen that doesnt put a space after commas in function parameter lists
20:24 sapier      yea bad habit sorry
20:25 sapier      I'm gonna fix it ... but believe me if that was the only problem I'd have that shit finished days ago
20:26 sapier      any other coding style comments?
20:27 celeron55   wtf is this 8D
20:27 sapier      yeah wtf is right comment for this crap ;)
20:28 celeron55   init_gettext() that restarts the whole process
20:28 PilzAdam    I like that you fixed the warnings in irrUString.h
20:28 sapier      exactly that's the only way to make msvc compiled minetest use the updated LANGUAGE environment variable
20:29 sapier      for localization multiple different effects happen
20:30 sapier      first thing is gettext is using something to decide what language to return
20:30 thexyz      did you figure out what exactly it uses?
20:30 sapier      after lots of test it seams this decision is only made once on load of dll ... no chance to make a change later (at least for msvc builds)
20:30 sapier      -seams +seems
20:30 sapier      second thing is character encoding
20:31 sapier      you can change the character encoding in msvc build by using setlocale BUT this only works for windows named locales
20:31 sapier      thats what the lookup table is or
20:31 sapier      for
20:32 thexyz      is there a full list somewhere?
20:32 thexyz      so that we can dump it into a huge file and forget about it forever
20:32 sapier      actually yes ... about 180 entrys not counting variants and charset variations ;-)
20:33 thexyz      well, I don't really feel like adding a new if every time user asks us to add a new locale
20:33 sapier      I'm not even sure this is gonna work on winxp
20:34 thexyz      well it's deprecated anyway
20:34 sapier      I tried linux mingw msvc using system default, environment, minetest.conf
20:34 sapier      where minetest.conf overrides everything else while environment only overrides system default
20:35 thexyz      I think a good idea would be to declare some macro, i.e. SHITTYLOCALE("fr", "French") and have a huge file shittylocales.something which we will include
20:35 thexyz      (if the only option is to have a full list of locales, of course)
20:36 sapier      if someone has an idea to do this better I'd be glad
20:37 thexyz      well I have none except for the suggestion of having all locales listed, not just ones we have translations for
20:37 sapier      btw due to gpl reasons we can't even fix the gettext update issue. If I understood correctly you're not allowed to distibute gettext dlls' built with current msvc
20:38 thexyz      why?
20:39 sapier      they mentioned something about msvc binary parts linked into dll and a gpl incompatibility of microsofts permissions for those parts
20:40 thexyz      oh wow
20:40 sapier      I didn't completely understand but considering strange behaviour of that dll I'd not be surprised if they'd had to include some system specific things to dll to get it work
20:40 celeron55   i would guess winapi has a way to list all of those needed locale names
20:40 sapier      yea .. list microsoft language id and name ;-)
20:41 celeron55   http://msdn.microsoft.com/en-us/library/windows/desktop/dd319091%28v=vs.85%29.aspx
20:42 celeron55   the code could just linearly find what it needs using EnumSystemLocalesEx
20:42 sapier      interesting they mention the short locale name there
20:42 celeron55   this is terrible but i guess better than a static list inside minetest
20:42 sapier      so at least in vista they did still recognize those names
20:43 sapier      ok they use - instead of _ but converting this shouldn't be that hard
20:44 sapier      but guys took me 1 week to figure that crazy stuff out :-) I'm for merging and fix later if someone really cares
20:45 celeron55   you should write an entry for TDWTF 8)
20:46 sapier      whats TDWTF? :-)
20:46 celeron55   (well probably not; they don't do stuff like this really)
20:46 celeron55   http://thedailywtf.com/
20:48 sapier      :-)
20:48 sapier      if clang has other localization issues someone else has to fix it ;-P
20:48 celeron55   i completely hate it how microsoft basically abandoned their native API and hacked .NET on top of it
20:49 celeron55   they're going to fail hard some day when their foundations break into pieces that nobody can fix 8)
20:49 sapier      I never did program for windows so I can't really say anything about it .. I only used qt on windows
20:50 sapier      but still someone should test my pull request on other windows compilers e.g. msvc 2010
20:50 sapier      I'd not be surprised if other things would happen
20:55 celeron55   what i think is that you probably shouldn't jump so straight into doing but think first
20:55 celeron55   with the same effort maybe gettext could've had thrown away and something less crappy put into use instead
20:56 celeron55   hacking a program to work with a hacky library on a broken OS is just terrible
20:58 sapier1     sorry connection lost ... I tried to find a replacement for gettext but haven't found a single one
20:59 kahrl       maybe we could code our own? it can't be that hard to be honest
20:59 sapier1     and gettext works fine for mingw
20:59 sapier1     it's just msvc that causes issues
21:01 sapier1     I don't think it's hard it's just some work to do it correct and performant
21:01 celeron55   it's not like we even need correct or performant
21:01 celeron55   8D
21:03 sapier1     maybe but I don't feel good in replacing a lib working just fine just because some creepy microsoft compiler that not even has to be used
21:03 sapier1     directx is working on mingw too
21:04 celeron55   directx on mingw is definitely good; we've never had that before
21:05 sapier1     why not? it's no problem at all?
21:05 sapier1     you don't even have to do something special?
21:06 celeron55   nobody has bothered to build irrlicht by themselves
21:06 sapier1     I see :-) irrlicht was one of those packages that did almost instantly build
21:07 celeron55   (for windows, that is)
21:07 sapier1     I think I'm gonna publish my build instructions in wiki
21:07 sapier1     on linux I don't have to build libs to match my compiler ;-)
21:09 sapier1     could someone allowed to plz add a "build instructions" page to "Dev Docs" ?
21:13 celeron55   doing that
21:14 celeron55   aactually
21:14 sapier1     hmm it's at core right?
21:14 celeron55   there is already a "compiling minetest" page
21:15 celeron55   it's under "Core"
21:15 sapier1     ok I obviously wouldn't expect it there but I'm gonna use it
21:16 celeron55   it's a bit odd... but it doesn't fit under "dev docs" either
21:16 sapier1     true
21:16 celeron55   well i'll fix this in a certain way
21:16 sapier1     I guess I need to update my docs they don't really match the style used there
21:18 celeron55   is it good now?
21:18 sapier1     I dodn't use a fixed directory structure but set the paths in cmake-gui ... that's a lot of work once but you're free to place your libs where they build
21:18 sapier1     Instructions? what about "howtos"? ;-)
21:18 celeron55   well dunno; it's good enough
21:19 sapier1     ok :-) I'm fine with it
21:19 celeron55   i think maybe you shouldn't edit the existing page
21:19 celeron55   let's add a second page with a windows build done in your way?
21:20 celeron55   we need a title for that
21:20 sapier1     I intended to add a link on top at first stage, my doc isn't finished right now but could be helpfull for others
21:21 celeron55   if you just append stuff to the existing page, that's obviously fine so in that case go ahead
21:22 celeron55   i think it's important to have the simplest way to build on windows documented and to not horribly mangle it in pieces within a document about building the whole universe
21:23 sapier1     my way isn't simplest but by now it's most clean way to do it
21:29 sapier1     http://dev.minetest.net/Build_Win32_minetest_including_all_required_libraries my instructions those are my personal notices they may not be complete so comments are welcome