Time  Nick       Message
04:49 kahrl      ... look at that
04:49 * VanessaE looks
04:50 VanessaE   what, where? :D
04:50 kahrl      sapier: the mainmenu lua could read any file readable to the current user, even if all file operations were restricted to the minetest directory
04:50 kahrl      libcurl supports file:// URIs :D
04:55 VanessaE   I...'m not so sure that's a safe thing to allow :)
05:00 kahrl      should we set CURLOPT_PROTOCOLS to CURLPROTO_HTTP | CURLPROTO_HTTPS?
05:00 kahrl      what about FTP, FTPS?
05:01 VanessaE   well
05:01 kahrl      (and set CURLOPT_REDIR_PROTOCOLS to the same, of course)
05:01 VanessaE   I see no problem allowing FTP
05:01 VanessaE   secure connections makes sense if you're just talking about fetching files from some random ite that enforces such
05:02 VanessaE   site*
05:02 VanessaE   (imagine using github as a media server :D )
05:02 VanessaE   redirects...mmmh  I dunno about that
05:04 kahrl      we already talked about redirects and the consensus was to allow them
05:04 VanessaE   I can see a situation where a Minetest server is using some random website for its media files, without necessarily having direct control over that site
05:05 kahrl      yeah - lets say somebody controls a media server but not the minetest server, but wants to shut it down temporarily and just forward to cdn.minetest.net
05:05 VanessaE   site owner decides "ok, this is costing me too much bandwidth, let's outsource".  Next thing you know, 2 or 3 redirects later, everything being served up is some variant of goatse.
05:06 VanessaE   that's the only argument against it that I can see, other than redirects of course being a bit slower than direct access.
05:07 VanessaE   (this is the part where sapier is supposed to sign on and start in on this conversation :D )
05:34 VanessaE   for some reason, it feels like we're forgetting some important protocol..
05:35 VanessaE   can it do rsync?
05:37 kahrl      oh there's a list of the ones it can do in curl.h
05:37 VanessaE   what, you want I should actually READ THE CODE!? ;)
05:37 kahrl      http://code.google.com/p/msysgit/source/browse/mingw/include/curl/curl.h?r=19288e3798dc2003b9a67bbcfd437e7de3e34ef1#749
05:38 VanessaE   gopher! :D
05:38 kahrl      oh wow
05:38 VanessaE   AH SCP
05:38 VanessaE   that one is useful
05:39 VanessaE   too bad it can't do NFS ;)
05:40 VanessaE   really though, http/ftp is enough
05:40 kahrl      just mount it and it can ;)
05:40 VanessaE   true :D
05:41 VanessaE   (seriously, gopher?  does anyone use that anymore?)
06:00 thexyz     PilzAdam: I'll do that later, there are too many conflicts because 4bbb78a99b5d2a8446a1946eb44df01b618de4cc
10:18 PilzAdam   thexyz, oh damn, havent thought about merging weblate before pushing that one, sry
11:45 PilzAdam   https://github.com/minetest/minetest/pull/871
13:30 celeron55  >Next thing you know, 2 or 3 redirects later, everything being served up is some variant of goatse.
13:30 celeron55  that actually doesn't matter because minetest can check the hashes of the contents
13:30 celeron55  (i don't think it does though?)
13:30 celeron55  maybe it should
14:04 thexyz     wait wut wut
14:04 thexyz     I asked for redirects because that'd allow me to abuse cloudflare free plan
14:04 thexyz     and yeah, it definitely should check hashes
14:35 jojoa1997  Hello
15:14 rubenwardy Is anyone working on the lua main menu?
15:15 rubenwardy I started rewriting it to have a better style, and I am wondering whether that collides with any one
15:15 PilzAdam   there are some open pull requests
15:16 rubenwardy Yeah, mine will conflict
15:16 rubenwardy But it will be easy to adjust the commits to mine.
15:16 PilzAdam   what do you mean with "better style"?
15:18 rubenwardy like this: https://gist.github.com/rubenwardy/d05011fb13ddb398bae6
15:37 sapier     kahrl mainmenu doesn't enforce any security policys
15:40 sapier     rubenwardy actually I just added i18n support to mainmenu
15:48 sapier     rubenwardy do you intend to use a single file per tab or add dialogs to those files (if the dialog is relevant for that tab only)
15:48 rubenwardy not sure
15:49 rubenwardy sounds logically
15:49 rubenwardy *logical
15:49 sapier     your design is similar to what I used for modmanager and modstore (I've been to lazy to update the other tabs too) ;-)
15:50 sapier     modstore and manager need tab private data too so there's an additional difference
15:50 sapier     menu.world_to_del would be private too if this dialog wasn't called by different tabs ;-)
15:51 sapier     but you should merge the i18n branch first unless you want to add it manually later I doubt automatic merge is possible
15:52 rubenwardy I will add it manually
15:53 sapier     np don't forget about the fixes in guiFormspecMenu.cpp
15:54 sapier     and plz don't revert the naming fixes for those TP functions it's just ugly to have anything lower case and 3 functions TPddfhdsg
16:01 rubenwardy is menu.render_TP_list(TPlist)
16:01 rubenwardy correct?
16:02 sapier     no
16:02 sapier     I replaced TP by texture_pack
16:02 rubenwardy ok
16:03 sapier     yes it's longer but you don't have to guess "what does TP mean in this special case"
16:03 sapier     most time it's obvious but e.g. filterTP isn't
16:04 sapier     sfan5 minetest is still missing color predefines this was something YOU said you'd add it after you broke the previous readable color selection
16:25 nore_      can someone look at #858 and #859 please?
16:26 nore_      those are fixes for mainmenu and formspec-related things
16:26 rubenwardy I can not access github
16:27 Calinou    it is odwn
16:27 sfan5      sapier: ..
16:28 rubenwardy I am going to have to go now
16:28 sapier     sfan5 I need to add those damn color macros in all of my mods it's just annoying to have hundreds of copys
16:29 thexyz     wait, wait, what color macroses do you need?
16:29 sfan5      ^ was just about to ask that
16:29 nore_      looks like they are under a ddos :(
16:29 thexyz     have we agreed on some set yet?
16:30 sapier     maybe you can imagine right exactly what color 19A222 is  but I CANT
16:30 nore_      sapier: light green?
16:30 sapier     there already was a set ... I don't say you can't add more but at least that minimalistic ser
16:30 thexyz     I vote for this http://en.wikipedia.org/wiki/X11_color_names
16:30 thexyz     this set was shit (sorry)
16:31 sapier     I don't know what color this is I need to start a program to update forst
16:31 thexyz     I mean, YLW? GRN? Are you serious?
16:31 sapier     a little bit too much  :)
16:31 thexyz     any objections to "Color name charts"?
16:31 sapier     no need to keep the exact names
16:32 thexyz     from the wiki article I linked
16:32 sapier     the 3 char lenght was a commitment to ease parsing
16:33 sapier     isn't that already a little bit too much?
16:34 sapier     as I'm already complaining .... -O1 is still crap for DEBUG BUILD
16:36 thexyz     sapier: what do you mean?
16:37 sapier     if you set cmake build type to debug you get compiler optimization -O1 which is useless for debugging ... and guess what I always need to compile twice
16:37 thexyz     http://www.w3.org/TR/css3-color/#svg-color
16:37 thexyz     sapier: no, I'm talking about "a little bit too much?"
16:38 Calinou    <sapier> maybe you can imagine right exactly what color 19A222 is  but I CANT
16:38 Calinou    it's easy
16:38 Calinou    palettes suck
16:38 Calinou    it's much easier for people to learn how to use hex colors than a palette
16:38 Calinou    I never remember IRC colors, or worse, X11 colors
16:38 sapier     we don't force that palette but just allow ppl to use it for convenience
16:38 Calinou    (and don't even know how to use X11 colors anyway)
16:39 sapier     I don't want to think about what "dark green" I used for this part neither do I want to look if #FFAACC was whit grey or light lemon if I read code
16:39 Calinou    keep gcolor2 or such at hand
16:41 celeron55  any kind of pretty colors generally aren't anything that people actually know names for
16:41 thexyz     I'm with sapier on that; it's easier to type darkred for me than to look what darkred is
16:41 sapier     there's a app for everything of course but thats not an excuse to not add a simple lookup table increasing usability drasticaly
16:41 celeron55  (i'm against named colors; even just for KISS)
16:41 sapier     I'm not talking about hundreds of colors just the basic colors and maybe light and dark versions
16:42 sapier     I doubt #112233 is KISS
16:42 sapier     maybe last S
16:42 sapier     ;-)
16:42 celeron55  why not?
16:43 celeron55  ;-) ;-) ;-)
16:43 sapier     because you don't program mods and don't need to lookup those names for code you didn't look at for lets say 2 months
16:43 celeron55  put colors in variables in the code then
16:43 thexyz     that's what we were going to do
16:44 celeron55  myniceblue = "#3344cc"
16:44 sapier     yes so we have 200 mods all using slightly different colors ... that's going to give a very consistent feeling when playing minetest
16:45 sapier     I already use those variables for my mods but I guess other mods use slightly different dark green values
16:45 celeron55  so you intend to add some kind of minetest.colors = {somecolor="#something", ...}
16:46 celeron55  or what?
16:46 thexyz     yes
16:46 sfan5      what if modders use their own colors instead of built-in "standard" colors, that not consistent too
16:46 sapier     I more thought about something like COLOR_GREEN = "#00FF00" but thats a minor detail
16:46 celeron55  then i am strongly against using something like that "green" is #00ff00
16:46 celeron55  because that's a horrible color
16:46 celeron55  make a good palette
16:46 sapier     that's been an example ;-P
16:47 sfan5      00ff00 is lime green
16:47 celeron55  and don't stick to any programmer colors
16:47 thexyz     there already are color palettes
16:47 thexyz     that are supported everywhere
16:47 thexyz     (hint: http://www.w3.org/TR/css3-color/#svg-color)
16:47 sfan5      lets use HTML colors
16:48 sapier     I don't care about what exact colors so I'm fine with svg as well as html colors
16:49 celeron55  i think someone who has actually ever made any kind of acceptable graphics should be consulted on this palette issue
16:49 sapier     I only wanna write something like "textlistelements = COLOR_GREEN .. "the text for element," .. COLOR_RED .. "some other element," .. COLOR_BLUE .. "another element"
16:49 thexyz     ah
16:49 thexyz     that feeling when c55 ignores you
16:50 thexyz     no let's make our very own palette
16:50 thexyz     because NIH
16:50 celeron55  thexyz: i looked at that; what then?
16:50 * sfan5    is not the person who made "acceptable" graphics (hint: initial apple texture)
16:50 celeron55  thexyz: i'm not qualified to choose anything
16:50 sapier     rba maybe? or vanessae?
16:51 celeron55  ...
16:52 celeron55  they are simply programmers too
16:52 PilzAdam   Im against predefined colors, modders should be forced to think about their own colors, and not just use what someone thinks is a good color
16:52 sapier     ok who is our graphics guru then? *g*
16:53 sfan5      PilzAdam: <sapier> [...] that's going to give a very consistent feeling when playing minetest
16:54 sapier     we don't disallow ppl to use different colors predefined colors are just a hint
16:54 Calinou    green would be 008000, bright_green would be 00ff00, and dark_green would be 004000
16:54 Calinou    (about the "green" discussion)
16:54 thexyz     okay, so can anyone explain the problem with w3c set?
16:54 thexyz     those are well known
16:54 Calinou    or use a pre-made set
16:54 Calinou    PilzAdam: you use kubuntu, and you say that? :p
16:54 celeron55  if consistency is wanted and you want to look good, then you need some kind of an actual palette, like http://colorschemedesigner.com/#3142lmmVQSSg2
16:55 celeron55  with something like thexyz linked, you don't get consistency
16:55 celeron55  they are completely different things
16:55 thexyz     what is "consistency"?
16:56 thexyz     to me, it's when red in all mods is red
16:56 sapier     e.g. if you have 5 dialogs all using "green" for something good and every green is different that's not consistent
16:56 celeron55  sapier clearly wants some kind of stylistic palette
16:56 celeron55  thexyz just wants to write programmer colors as words
16:56 sapier     that'd be best
16:57 celeron55  it's a different thing
16:57 thexyz     20:47  <      sapier> | I don't care about what exact colors so I'm fine with svg as well as html colors
16:57 celeron55  we could even implement both though
16:57 thexyz     20:49  <      sapier> | I only wanna write something like "textlistelements = COLOR_GREEN .. "the text for element," .. COLOR_RED .. "some other element," .. COLOR_BLUE .. "another element"
16:57 sapier     I'd be fine with thexyz's version too but yes stylistic set would be even better suite my intention
17:03 nore_      now that github is back online, could someone look at 868 and 869 please?
17:03 celeron55  minetest.colors.svg = {} and minetest.colors.theme = {}? 8) there's the problem of figuring out how to base the themed color naming
17:04 thexyz     and then suddenly it's easier to type #008000 than minetest.colors.svg.seriously.wtf.are.you.doing.green
17:05 celeron55  you can just do local c = minetest.colors.svg and then you have c.red and whatever
17:05 celeron55  namespacing is a much used concept for a reason
17:05 celeron55  it's insane to pollute the global namespace with colors
17:05 thexyz     why not?
17:05 nore_      thexyz, that is the main problem
17:06 celeron55  because you can do it cleaner with only benefits
17:06 thexyz     like? the first benefit is that every mod will include >local c = minetest.colors.svg
17:07 celeron55  the first benefit is that the constants aren't ugly and prefixed with bloat that you can't get rid of
17:07 celeron55  local green = COLOR_GREEN; local red = COLOR_RED; yeah fun
17:07 thexyz     did I say that it's better to use COLOR_GREEN?
17:08 thexyz     even global `colors = {}` is better
17:09 PilzAdam   why not a function minetest.get_colors("svg") that returns a table?
17:09 thexyz     shit
17:09 PilzAdam   so every mod can call it whatever they want
17:09 celeron55  thexyz: well you can just make that a default alias for minetest.colors.svg; whatever; but having only a global "colors" would be masked if someone has their own variable "colors"
17:11 thexyz     PilzAdam: why not a metaclass which returns a class which then we use to create an object which has a very convenient :get_colors method which returns table with colors? so every mod can just call it whatever they want
17:11 celeron55  you just have to have a sane hierarchy of namespaces and then tug it down to some easier default if it seems to be needed
17:12 thexyz     I'm fine with minetest.colors as long as we have some shortcut like mt.c
17:12 thexyz     I'm not good with naming
17:14 celeron55  does the API currently eat any other kind of colors than those html six digit hexes?
17:16 celeron55  i can't really tell based on lua_api.txt
17:16 nore_      sapier, I tested what happened with the \n not removed by 859, and it looks good.
17:16 celeron55  it's written too badly
17:17 celeron55  everything is vague
17:17 thexyz     it's just trolling
17:17 thexyz     > see colorkeys
17:17 thexyz     and there's nothing about colorkeys, moreover, they're dropped now
17:17 celeron55  yeah
17:18 celeron55  also like "color in hexadecimal format RRGGBB"
17:18 celeron55  while there is no place for color
17:18 thexyz     in some places you should use `#color`, in others `color`
17:18 celeron55  also nobody tells whether there is # or not
17:18 thexyz     (or so it seems)
17:18 celeron55  also 0xFFFFFF is mentioned
17:19 celeron55  lol 8D
17:19 thexyz     in HUD
17:19 thexyz     and dev.minetest.net is seriously behind
17:20 thexyz     moreover, some methods use camelCase while others are underscore_separated
17:20 celeron55  i'm sure this is how PHP developers feel
17:22 thexyz     they have docs and examples
17:22 thexyz     and stackoverflow
17:22 celeron55  we have stackoverflow too! http://stackoverflow.com/questions/8142968/irrlicht-uses-wrong-opengl-version
17:23 nore_      sapier, I tested what happened with the \n not removed by 859, and it looks good.
17:23 nore_      is there anything more to do?
17:23 thexyz     no answers, how typical
17:24 celeron55  that's how stackoverflow works
17:24 celeron55  you ask a question and then everyone tries to not answer it but rather post useless comments!
17:25 sapier     I don't think so nore_ but maybe others do have comments too
17:26 sapier     I'd be fine with COLOR=minetest.color.svg and use COLOR["green"] in mod too
17:26 sfan5      thexyz: #<color> is when color is optional (list elements) and <color> if its required
17:27 sapier     as long as I don't have to copy a x-elements color table from mod to mod and keep that damn thing in sync ;-)
17:27 sapier     is there something except listelements that use color right now?
17:27 sfan5      IIRC yes
17:28 sapier     any chance to make this consistent for both?
17:28 sfan5      umm
17:28 sfan5      https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L1001
17:28 sfan5      apparently someone forgot to remove something when copy-pasting
17:29 sapier     no the upper description 993 is wrong
17:30 sfan5      also: the #<color> vs. <color> things makes sense since otherwise the engine wouldn't know whether the listelement is "F1A99BSome text" without color or "Some text" with color #F1A99B
17:30 sapier     descriptions usually don't use optional fields
17:30 sapier     yes in listelement we need the # (or some other escape character to mark a color)
17:30 sfan5      Line 999 is wrong
17:31 sfan5      it refers to colorkeys
17:31 nore_      sfan5, but how does it know that if text begins with #?
17:31 sfan5      nore_: ##F1A99B -> "#F1A99B" without color
17:31 sapier     colorkeys is at 993 ... why does everyone always have different line numbers than I have??
17:31 sfan5      https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L999 refers to colorkeys
17:33 sapier     ok now I have two konqueror windows with same file but different line numbers ;-/ ... I hate kde
17:33 sfan5      haha
17:34 sfan5      I'll work on a fix for the colorkey mess in lua_api.txt
17:34 sapier     yes :-) you made it you should fix it ;-P
17:34 nore_      sfan5, about formspec-related things, what do you think of #868 and #869?
17:35 sapier     something completely different what about a minetest.getusectimestamp() function for tracing purposes?
17:37 nore_      the important question is where time is spent, not how much...
17:37 nore_      but it coukd still be useful
17:37 sfan5      what about a profiler interface for lua?
17:37 sapier     could be implemented with this fct
17:38 sapier     it'd be a wrapper layer between core and mods but with second granularity like os.time provides there's no use for profiling
17:38 thexyz     you can have a profiler in pure lua
17:38 sapier     and socket isn't installed by default
17:39 sapier     no you can't thexyz
17:39 thexyz     no you can
17:39 sapier     all profilers I know need accurate time
17:39 sapier     and by default lua doesn't provide a timesource with reasonable resolution
17:39 sfan5      lua_api.txt fix pushed
17:41 sfan5      nore_: #868 should be merged IMO, I have that bug too; #869 too maybe
17:41 sapier     I already use lua profiler for mobf but as long as it requires the socket lib to be installed I can't tell users "enable profiler and send it to me" in case of performnce issues
17:42 sapier     sfan5 this one isn't a bug but an irrlicht feature ;-)
17:43 thexyz     http://lua-users.org/wiki/PepperfishProfiler
17:43 sapier     but I'm for 868 too although I don't like singleclick world starting
17:43 sfan5      I don't think the Irrlicht devs will do it like this "feature" annoys one user lets remove it ;)
17:43 sfan5      s/;\)/;-)/
17:44 sapier     os.clock()??
17:44 PilzAdam   nore_, what about completly removing single- and doubleclick to start a world? since singleclick is not intuitive and doubleclick doesnt work correctly
17:44 thexyz     >event.typ >DCL
17:44 thexyz     aww
17:45 sapier     you approved it two months ago thexyz don't complain about it now without better suggestion
17:45 celeron55  i think the lua gui stuff should be modified to accept # in any color, so that the color lists could be in format "#rrggbb"
17:45 nore_      PilzAdam, I can do thqt
17:46 thexyz     sapier: yeah, what about "type"
17:46 celeron55  (or, well, color tables)
17:46 nore_      celeron55, I agree
17:47 thexyz     sapier: I think os.clock is good enough to "tell users "enable profiler and send it to me" in case of performnce issues"
17:47 sapier     hmm noone realized that typo before
17:48 nore_      should I rem8ve clicking to start world or not?
17:48 thexyz     sapier: also, I'd prefer having constants for events instead of strings
17:48 sapier     thexyz I'm not sure if this is in ms for all architectures
17:49 nore_      sapier, it was a typo? I thought it was to shorten the variable names..
17:49 sapier     thexyz there are no constants in lua
17:49 sapier     unless you want to abuse a global variable as constant
17:49 nore_      use DCL instead of "DCL"
17:50 nore_      with DCL=1
17:50 thexyz     sapier: can you tell me the difference?
17:50 sapier     a global variable pollutes namespace
17:50 nore_      a constant does that too
17:50 sapier     if someone does DCL = 10 it'll fail horribly
17:50 sapier     there are no constants in lua nore_
17:51 thexyz     okay, then yeah, I'd prefer to "abuse a global variable as constant" instead of using "DCL"
17:51 celeron55  we use strings in lua
17:51 celeron55  but what is that DCL thing
17:51 thexyz     I'd also call it something more reasonable
17:52 thexyz     like EVENT_DOUBLE_CLICK
17:52 celeron55  i don't even know what you're talking about, but it doesn't sound at all like what we tend to call things
17:52 sapier     just change it ;-)
17:52 * sfan5    prefers s/DOUBLE_CLICK/DOUBLECLICK/
17:52 sapier     formspec is only designed for buttons
17:52 celeron55  knowing nothing about it other than that it's doubleclick, i'd call it "doubleclick"
17:53 thexyz     celeron55: and then you make a typo somewhere
17:53 sapier     so textlist needed to add information to determin if it's a doubleclick  or singleclick thats what "DCL" is used for
17:53 sfan5      has settings a metatable on _G any effect on setting variables normally?
17:53 celeron55  thexyz: the api is full of those kinds of things; you just need to be careful and that's it
17:53 thexyz     though there's no difference
17:53 sfan5      if it does you could make pseudo-constants
17:54 celeron55  thexyz: also, using a random variable name in lua is nil, isn't it?
17:54 sapier     if you want to change it change it now it's gonna break any mod using textlists
17:54 thexyz     celeron55: there are some tools which can check your code
17:54 sapier     yes it is
17:55 sapier     so if you use DCL you have to add DCL=1 SCL=2 .. or whatever singleclick is right now
17:55 thexyz     https://code.google.com/p/lua-checker/
17:55 celeron55  sfan5: for sure you can make some kind of constants in lua with metatables, like you can make almost anything with metatables in lua
17:55 nore_      singleclick is CHG IIRC
17:55 sapier     ok CHG=2
17:55 sfan5      metatables only work for arrays IIRC
17:56 celeron55  sfan5: yeah you need to put them in some table/object kind of thing
17:56 nore_      s/arrays/tables
17:56 thexyz     EVENT_CLICK
17:56 celeron55  altough in lua the global namespace is a table too
17:56 thexyz     holywar in process …
17:56 sfan5      let me try if you can set the metatable of _G
17:56 celeron55  dunno if you can add a metatable to that 8)
17:57 sapier     no in lua it's one single stack knowing the index you can access anything
17:57 sapier     like addresses in c
17:57 celeron55  net.minetest.api.gui.formspec.v1.event.click.variant.double.immediate.typefield.variant
17:58 sapier     isn't metatable overkill?
17:58 thexyz     seriously, I'd prefer this to "CHG" and "DCL"
17:58 sapier     guys you need to distinguish singleclick from doubleclick only
17:58 celeron55  to be clear: i am not proposing to use anything for enums in minetest's lua api; i want to stick with strings
17:58 sapier     it's passed to lua by string ... the events already are convenience features
17:58 celeron55  strings in the style that we already use
17:58 PilzAdam   <celeron55> net.minetest.api.gui.formspec.v1.event.click.variant.double.immediate.typefield.variant <- since when do you code java?
17:59 sapier     if you don't whant the event mechanism you can parse the event yourself too
18:00 nore_      www.lua.org/pil/14.2.html
18:00 nore_      sfan5, above
18:01 nore_      for metatable of _G
18:01 sapier     e.g. if field["yourtextlist"] == "CHG:1" then select first element
18:01 sfan5      nore_: mhm
18:03 sapier     thexyz you can change it in guiFormspecMenu.cpp to whatever you want but to it prior 0.4.8 if this is once out in a stable version it's no option to break all mods using it
18:04 nore_      about 868, should I remove click-to-start-world?
18:05 nore_      and for 869, is there anything more to do?
18:05 sfan5      869 seems fine
18:06 sapier     nore_ you should write somewhere that sever as well as mod description may look different after this change
18:06 nore_      in news?
18:07 sapier     in puhh request description will be enough
18:07 nore_      ok, I will do it
18:07 sapier     as old behaviour isn't in a stable version too right now i guess it's no problem too
18:09 nore_      sapier, done
18:10 sapier     thexyz I guess os.clock will be fine for some things but I'm not sure if ms is really enough ... at least on windows and linux it's ms ... does anyone know if this is true for bsd too?
18:13 kahrl      I'm trying out something else
18:13 kahrl      on listbox click, check if the same item was previously selected (by a user event, not setSelected)
18:14 kahrl      by the way, does the API currently set all listboxes to DCL if one of them is doubleclicked?
18:15 sapier     I don't know I didn't have that usecase by now
18:17 sapier     if you check for same element you need to do some timekeeping too
18:17 kahrl      oh, I only check if irrlicht reports SELECTED_AGAIN
18:18 kahrl      so that should do the timekeeping
18:18 sapier     no it doesn't
18:18 sapier     wasn't there a bug with a event beeing triggered right at begining?
18:20 kahrl      testing...
18:20 kahrl      (though someone with 1.7* needs to test the end result too)
18:23 kahrl      should stuff like explode_textlist_event be in the global namespace? and where are INV, CHG, DCL documented? do any mods use them?
18:24 nore_      it's too recent, so I don't think so
18:24 sapier     mobf uses them
18:25 sapier     but as I said by now its in no stable minetest version
18:25 sapier     and I'm not gonna declare new mobf version stable on top of a git minetest version
18:27 sapier     still why do you realize this 2 months after merge and more than 3 months after initial pull request
18:27 kahrl      because I didn't read it all?
18:28 sapier     if this really is that critical why didn't anyone else realize it?
18:28 PilzAdam   sapier, it would be more visible if you had documented it somewhere
18:29 sapier     of course I could have documentet exactly those points you discover later ... if I'd knew what youd discover later
18:29 sapier     if I documented any single thing I did for mainmenu you'd not have found that single point in that huge documentation too
18:30 PilzAdam   is there anything else that is not documented in menu_lua_api.txt?
18:30 sapier     of course ... I guess all those things added after initial pull request may be candidates for it
18:30 sapier     doubleclick support was one of those too
18:31 sapier     initial mainmenu didn't have doubleclick support but someone said "no no no that can't be done without doubleclick"
18:31 kahrl      oh, I forgot to add some of the functions I moved to ModApiUtil to menu_lua_api.txt
18:31 sapier     ;-P
18:32 sapier     but I remember adding DCL documentation I don't know why it was lost in final version of doc
18:33 kahrl      get_dig_params, get_hit_params, get_password_hash
18:39 PilzAdam   thexyz, is it possible to rebase the weblate branch before merging into master? this way all the commits would be together in the history
18:41 kahrl      sapier: if we stick to the INV/CHG/DCL convention, what would you call the type where a listbox is unchanged
18:41 kahrl      UNC?
18:43 sapier     why send a event if nothing changes?
18:43 sapier     you could consider "not changed" as special case of changed too aka change to same
18:43 celeron55  NOP
18:43 celeron55  8D
18:44 celeron55  send it each millisecond for each item
18:44 thexyz     PilzAdam: yeah maybe
18:44 thexyz     what's the problem with merge btw?
18:45 sapier     millisecond? you're that yesterday now its us or even ns
18:46 kahrl      bah! planck time
18:46 sapier     use your imagination half planck time would be interesting
18:47 sapier     no really what's "not changed" good for?
18:49 kahrl      isn't the formspec supposed to send all fields when one changes?
18:49 kahrl      it's weird that all the textlists are reported as changed
18:49 sapier     as far as I know for buttons only the one clicked is sent
18:49 kahrl      hmm
18:50 sapier     actually only the one changed should be reported if there's more this is a bug
18:50 kahrl      ah okay, yes it does that, misread the code
19:07 PilzAdam   thexyz, its easier to skip the translate part in the commit log if all are in one place rather than spammed all over the place
19:14 thexyz     oh, I see
19:14 thexyz     we're using rebase for some odd reason
19:31 kahrl      I managed to fix one problem, which is the initial setSelected() counting as a click
19:32 sapier     good
19:32 kahrl      there's another one I haven't fixed yet, namely dragging counting as a doubleclick
19:34 kahrl      well, I guess for this I do actually need to store the time
19:36 proller    support for 63 levels liquid - https://github.com/proller/minetest/compare/liquid63 - very smooth flowing with lliquid_finite (not ready for merge now)
19:47 nore_      so, about 869, is there anything to be done, or can it be merged?
19:47 kahrl      nore_: sapier just left :P
19:47 nore_      and I guess 868 isn't needed anymore
19:48 nore_      kahrl, sapier said that he thought nothing more was needed, but that I had to ask the others about it
19:49 nore_      in the logs of today, a few hours ago
19:49 kahrl      I see
19:53 nore_      kahrl, with what you are doing, 868 isn't needed, is it?
19:54 kahrl      yep
19:54 nore_      or should it be added nevertheless?
19:55 kahrl      well I don't like it jumping into the game on a single click
19:56 kahrl      though if the other devs do, they can override my opinion
19:56 nore_      I can completely disable starting on click, if needed
19:56 nore_      just a few lines to delete
19:57 kahrl      nah, with my patch the doubleclick behaviour is kept without being annoying
19:58 nore_      you shiukd close it then
19:58 kahrl      when my patch is ready and agreed upon
20:00 nore_      I reckon there were other core devs agreeing 869, too
20:00 kahrl      https://github.com/kahrl/minetest/commit/589aacd6c94f6021f7bec1a0531acc1fa38cefb5
20:00 nore_      sapier, sfan5 said maybe
20:01 kahrl      sapier isn't a core dev
20:02 nore_      isn't he?
20:02 kahrl      he develops the core sometimes :P
20:02 nore_      but sfan5 is one anyway
20:02 kahrl      but not officially a core dev
20:02 kahrl      sfan5 is, yes
20:03 PilzAdam   kahrl, that patch works with 1.7.3
20:04 kahrl      nore_: how did the discussion about the \r \n problem end?
20:04 kahrl      PilzAdam, good
20:04 nore_      that it was ok, see the logs from today
20:05 PilzAdam   kahrl, now we need some more black magic to stop moving the slider when selecting an item from a long list
20:05 nore_      I tested it and did not see bad things, plus it has not been in a stable release yet
20:05 kahrl      PilzAdam: yep that keeps annoying me too
20:06 nore_      I have a question about modstore: does it take forever to load for you too?
20:07 nore_      and I have nothing on pages 2, 3,4, but page 1 is full
20:07 PilzAdam   nore_, it takes a while
20:08 PilzAdam   I guess the empty pages are due to mods that dont have a .zip release uploaded yet
20:08 nore_      but there are no screenshots nor description
20:09 nore_      and pages are still there
20:09 PilzAdam   I consider this a bug
20:09 kahrl      I haven't played with it yet
20:09 kahrl      it only shows installed mods for me
20:09 nore_      the button download?
20:10 kahrl      oh!
20:10 kahrl      I thought that would download the selected mod...
20:10 nore_      btw, PilzAdam, any opinions on 869?
20:11 PilzAdam   I am not familiar with the escaping stuff
20:11 PilzAdam   I just know that it tends to break stuff ;-)
20:12 nore_      I tested it, and did not see anything break
20:16 kahrl      didn't see anything breaking on a quick glance
20:17 kahrl      some server descriptions were empty, but they are empty in current master too (I guess they are really empty)
20:17 kahrl      code seems reasonable, so in my opinion it can be merged
20:18 nore_      lines 32 and 33 of modstore.lua should be changed
20:19 nore_      use get_texturepath instead of get_gamepath
20:20 nore_      and line 173 should use math.ceil and not math.floor
20:21 kahrl      what about defaulttexturedir in mm_textures.lua?
20:22 nore_      perhaps too, I did not check those files whan I added get_texturepath
20:22 nore_      some grep -R is definitely needed
20:23 nore_      kahrl, so it can be merged?
20:24 nore_      or do we need one more core dev?
20:24 kahrl      right now it's 1.5 devs agreeing to #869 so not yet
20:25 kahrl      brb
20:36 PilzAdam   869 is good if sapier agrees to it
20:37 nore_      he does, see earlier
21:40 Tesseract  Finally got a segfault in a debuger: http://pastebin.ubuntu.com/5990642/
21:41 PilzAdam   do you have the full backtrace?
21:41 Tesseract  I think it is reproducable by  loading a world, havind a mod crash, leaving the error screan for a few minutes, and then strting the world again.
21:41 Tesseract  One sec...
21:45 Tesseract  Is there a dump to file option for bt full?
21:45 PilzAdam   dunno
21:45 Tesseract  I have every variableavailabil when it crashed, which is quite a lot...
21:45 Tesseract  variable available*
21:46 Tesseract  How about a non-full bt...
21:46 Tesseract  http://pastebin.ubuntu.com/5990651/
22:05 kahrl      I'll merge 869 then. What about the doubleclick fix?
22:05 PilzAdam   it works, so its good
22:14 PilzAdam   since doubleclick works now: https://github.com/PilzAdam/minetest/commit/8d76f3eb35ce1ad7b8b4f24dd3b2ad4f2d6de1e6
22:15 PilzAdam   this adds the ability to use doubleclick in the modmanager too
22:18 kahrl      I think that has to wait until the scrolling problem is fixed
22:18 kahrl      maybe?
22:21 PilzAdam   I dont think so; its less useable but nothing breaks
22:22 PilzAdam   it works fine for the first 13 mods in the list, and for the others there is still the checkbox
22:24 kahrl      ah, I thought it was like !old_style_mod_selection i.e. without the checkbox
22:24 PilzAdam   its basically both ways combined
22:26 kahrl      ah, I think I can fix the scroll problem
22:26 kahrl      in an extremely ugly but working way
22:27 kahrl      basically: get a list of all the children of the listbox
22:27 kahrl      one of them is the scrollbar, gets its position when each listbox is destroyed, set it back to the old value after a new listbox is created
22:27 kahrl      get*
22:28 kahrl      I checked the CGUIListBox code, thankfully it doesn't try to cache the getPos() of the scrollbar
22:29 kahrl      if this stops working in a future release of irrlicht, the best solution would probably to "fork" CGUIListBox
22:29 kahrl      +be
22:52 kahrl      unbelievable
22:52 kahrl      it works
22:55 kahrl      https://github.com/kahrl/minetest/commits/textlistscroll
22:57 PilzAdam   nice
22:57 PilzAdam   works here too
22:57 kahrl      push?
22:57 PilzAdam   yea
22:57 PilzAdam   now the menu is actually useable again
22:59 Tesseract  Hmmm, automatic_face_movement_dir doesn't work with y velocity...
23:00 PilzAdam   Tesseract, because there is no way to set the pitch of an entity yet
23:00 kahrl      Tesseract: hehe, sapier said that nobody would need that in the near future
23:01 PilzAdam   kahrl, no, he said that about animation speed
23:01 kahrl      oh, right
23:01 Tesseract  Oh, well I need to set the pitch for a certain mod to look good at all...
23:02 PilzAdam   Tesseract, the only way to set pitch is by attaching the object to something
23:03 Tesseract  PilzAdam: I can set the pitch it is attached with?
23:03 PilzAdam   kahrl, can I push that doubleclick support for modmanager list now?
23:03 kahrl      PilzAdam, sure
23:03 PilzAdam   Tesseract, you can set the pitch of the child, not the parent
23:36 PilzAdam   https://github.com/PilzAdam/minetest/commit/b7458e6324b5c2a1328300c769da7041ac4f9f86
23:37 PilzAdam   oh, that table.sort() call in modmgr.lua:323 is redundant
23:40 PilzAdam   should game mods be always added at the bottom?
23:46 kahrl      yeah, I think that's fine
23:48 PilzAdam   https://github.com/PilzAdam/minetest/commit/360d44860cef9d170cfe5edeac208b6f4c9f2735
23:48 VanessaE   regarding colors:  check colors.txt in unified dyes.  I've long since established a standard RGB <-> name mapping standard there (except black and white might be off by 1 or two points)
23:49 VanessaE   (rest of backlog:  tl;dr)
23:49 VanessaE   oops, double standard there :D
23:50 PilzAdam   VanessaE, you dont want to read how kahrl fixed the doubleclick problem?
23:50 VanessaE   tt;tl;dr.
23:50 VanessaE   (too tired; too long; didn't read)
23:53 kahrl      PilzAdam: one problem: when "Hide mp content" is unchecked the modpack is not necessarily at the top of the mods it contains
23:53 VanessaE   glad you fixed it though, kahrl  :)
23:53 kahrl      e.g. minetest-mod-mesecons is below all the mesecons_... mods
23:54 PilzAdam   let me see how to fix this...