Time  Nick        Message
02:45 Tesseract   Something something #1606 something something
02:45 ShadowBot   https://github.com/minetest/minetest/issues/1606 -- Add mod security by ShadowNinja
02:52 est31       o_O
02:59 Tesseract   Zeno`, kahrl: +/- 1 on mod security ^ ?
03:01 Tesseract   0.5 (or a nonsensical version jump) is coming up, so compat is officially not an issue, and it's 99% compatible anyway.
03:01 Tesseract   Also, 99% of internet stats are made up.
03:01 Tesseract   :-P
03:02 est31       so the standard way to use an insecure function is to check whether a function is nil, and do request_insecure_environment() otherwise?
03:03 est31       Isn't it better to make request_insecure_environment() return sth when mod security is disabled, and the mod isn't listed?
03:03 est31       ah ok
03:03 est31       its only a doc isse
03:03 est31       issue*
03:04 Tesseract   est31: No.  `ins_env=_G if req_ins_env then ins_env = req_ins_env() end if not ins_env then error() end ins_env.ins_func()`
03:04 Tesseract   At least that's how I did it for the IRC mod.
03:04 Tesseract   It doesn't really matter exactly how you do it.
03:05 Tesseract   And if you don't care about compat that gets about 3 lines shorter.
03:05 est31       yes
03:06 Tesseract   (mod compat with old Minetest)
03:06 est31       doc should feature that though
03:06 est31       https://github.com/minetest/minetest/pull/1606/files#diff-e62796adc1d9474bfa497540cf203d47R98
03:06 Zeno`       I'm not a modder so I don't feel qualified enough to yay or nay on this commit. BUT from memory my main concern when this was first opened was that there be an ability to bypass the mod security and this is now possible. So I have no objections any more either
03:07 Zeno`       Network, apparently, will be broken soon as well so I agree that the two should be merged at the same time
03:08 Zeno`       If there's going to be disruption anyway :)
03:08 Tesseract   Zeno`: I don't claim that it's perfectly secure, I can't really know for sure.  But I know of no holes, and something is better than nothing IMO (as long as you're not claiming perfect security).
03:09 Zeno`       I meant a way for the modder to explicitly bypass it -- and you've added that feature, so all is well
03:10 Tesseract   Zeno`: The only mod that's really affected is the IRC mod.  The two others just need a few lines of compat code.
03:10 Zeno`       yep
03:10 Zeno`       I'll +1 it but I'd prefer if another dev +1's it as well (for something this big)
03:11 Tesseract   (and that only needs users to change a setting, in adition to the compat code)
03:11 Zeno`       what doth kahrl think?
03:12 Zeno`       Now (around about) is certainly the time to merge it
03:12 Zeno`       I can't wait for the flurry of "bug" reports :D
03:14 Zeno`       https://github.com/minetest/minetest/issues/2274#issuecomment-73924800
03:14 Zeno`       I'll merge that in a second
03:15 est31       http://irc.minetest.ru/minetest-dev/2015-02-11#i_4151665
03:16 est31       1,2, or 3?
03:16 est31       Tesseract, Zeno` ^
03:17 Zeno`       I think live with the warning
03:17 Zeno`       Option 2 isn't that bad either... since it's bundled and we don't modify it (AFAIK)
03:18 est31       btw its this change https://github.com/minetest/minetest/pull/2290/files#diff-597e4a4ad36b3232059614f27748f932R576
03:19 Tesseract   I vote 2 or 3.
03:19 Tesseract   We don't need the warnings since we don't touch that code.
03:20 Zeno`       Yep, so (2) is probably the best solution
03:20 est31       The first warning that got fixed did help
03:21 Zeno`       help?
03:21 est31       it pointed out that the cmake file is wrong
03:21 est31       and now, on posix systems, we don't use tmpnam anymore
03:23 Zeno`       yep, so that should stay
03:25 est31       ok, apparently you can selectively remove warnings, gonna do that instead
03:26 * VanessaE  peeks in
03:47 est31       what's weird, I've removed all -W* stuff from src/lua/CMakeLists.txt
03:47 est31       and still the warning is shown
03:47 est31       even added -Wno-empty-body
04:46 est31       ok, the cmake scripts are broken
04:46 Zeno`       excellent work :D
04:47 est31       as it seems, this only works when I actually remove the -Wall from the src/CMakeLists.txt
04:48 est31       at least it adds the warning again when I only add the wall to the one in src/lua
04:59 est31       Meh, dont want anymore.
04:59 est31       dunno how to fix that, sorry
05:00 est31       if you dont like the change of *comment positions*, then we will have to live with the warning.
05:00 est31       so whats your position?
05:02 Zeno`       it's not ideal having the warnings there but it's a third party package so *shrug*
05:03 Zeno`       how did your patch fix the empty body warning anyway?
05:03 Zeno`       that's not the Lua warning I get either heh
05:04 Zeno`       I only get loadlib.c:76:21: warning: ISO C forbids conversion of object pointer to function pointer type [-Wpedantic]
05:05 Zeno`       which is, of course, a pedantic warning
05:08 est31       My package fixed the warning by proper identation
05:08 est31       see this one:
05:08 est31       https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62184#c4
05:09 Zeno`       yeah I always put ; on it's own line. Didn't know indentation mattered though :/
05:11 Zeno`       I still don't get why indenting the whole line should suppress the warning. That's bizarre
05:12 est31       because this way the compiler thinks the line after that way is idented.
05:13 est31       and then it thinks that the ; might have got there by accident
05:13 Zeno`       https://github.com/est31/minetest/blob/master/src/lua/src/lauxlib.c     why doesn't line 570 generate the warning?
05:14 est31       because the line following it has the same identation level
05:15 Zeno`       but line 576 is your fix (add indentation) and line 577 doesn't have the same indentation level after the fix (it did before)
05:16 Zeno`       I think gcc needs to fix their warning personally
05:16 est31       ok, I'll move the while two to the left
05:20 Zeno`       ahh ok... it was originally 3 to the left
05:21 Zeno`       that makes more sense
05:21 Zeno`       (as to why the warning was emitted)
05:21 est31       I can move it if it makes sense (meaning if you accept it then)
05:22 Zeno`       I didn't say I was going to merge it. I brought it up because I was curious as to how it fixed the warning :)
05:25 Zeno`       est31, what version of gcc do you use (out of curiosity as well)
05:28 est31       4.9.1-16ubuntu6
05:32 est31       perhaps its also because of something else in the patch
05:33 est31       will you merge it when I remove this second warning fix completely?
05:43 Zeno`       If another dev agrees
05:53 est31       ok removed it you can make your +1
05:56 est31       better warning message...
05:57 est31       commit*
06:00 Zeno`       yep I'm fine with that. Lua does not come with cmakelists.txt anyway
06:01 est31       ok a github +1 too?
06:01 Zeno`       yep
06:02 est31       fine
06:02 est31       :)
06:04 Zeno`       Tesseract, are you ok with #2290 now?
06:04 ShadowBot   https://github.com/minetest/minetest/issues/2290 -- Fix gcc warning for lua by est31
06:05 Zeno`       actually I think he's already said that bit was fine. So I'll merge in 10 minutes unless someone objects
06:11 * Tesseract +1
06:15 Zeno`       Pushed (I changed the commit message, est31)
06:23 est31       lol is this still valid https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L17-19
06:57 nrzkt       oh, no warning on compilation for lua mkstemp, good est31
08:25 kahrl       BlockMen is back :) https://github.com/minetest/minetest/pull/2295
08:28 VanessaE    *gasp*
08:29 nrzkt       it's good or a problem ? don't know the history :)
08:29 est31       so, just to make, sure I'm not doing work in vain since 1.5 hrs, could you comment #2275?
08:29 ShadowBot   https://github.com/minetest/minetest/issues/2275 -- lua_api.txt should have a since field for every function
08:29 kahrl       as far as I'm concerned, good :)
08:30 est31       I have started with version 0.4.7 is that early enough?
08:30 VanessaE    how much you wanna bet he'll be confused as all fuck about the changes in the development model? :)
08:31 kahrl       est31: additions to the wiki are always welcome
08:32 est31       Is lua_api.txt too verbose then?
08:32 kahrl       probably so
08:32 kahrl       haven't thought about it too much
08:34 est31       there will be e.g. a (>=0.4.8) added
08:36 est31       the wiki is broken
08:37 est31       I wont fix it
08:38 nrzkt       i think wiki must be update, especially for this versioning in API
08:38 est31       so you dont object to it being in lua_api.txt?
08:39 nrzkt       i don't know for the lua_api.txt, i prefer wiki instead of flat files for those things
08:42 est31       Tesseract, you labeled it, and didnt close it, what do you think?
08:55 est31       can any core dev give their +1 to #2225 ?
08:55 ShadowBot   https://github.com/minetest/minetest/issues/2225 -- Fix for getCraftRecipes implementation is very slow by gregorycu
08:56 nrzkt       hmmmm is on the thread
08:58 est31       ?
08:58 nrzkt       kwoelr is hmmmm
08:59 nrzkt       erf my 0.4.11 stable has crashed by assert:
08:59 nrzkt       In thread 805406800:
08:59 nrzkt       Debug stacks:
08:59 nrzkt       DEBUG STACK FOR THREAD 805406400:
08:59 nrzkt       #0  int main(int, char **)
08:59 nrzkt       #1  Dedicated server branch
08:59 nrzkt       #2  void dedicated_server_loop(Server &, bool &)
08:59 nrzkt       (Leftover data: #3  void Server::step(float))
08:59 nrzkt       DEBUG STACK FOR THREAD 805408400:
08:59 nrzkt       #0  virtual void *ServerThread::Thread()
08:59 nrzkt       #1  void Server::Receive()
08:59 nrzkt       (Leftover data: #2  void Server::SendBlocks(float))
08:59 nrzkt       (Leftover data: #3  void Server::SendTimeOfDay(u16, u16, f32))
08:59 nrzkt       (Leftover data: #4  void ItemStack::serialize(std::ostream &) const)
08:59 nrzkt       (Leftover data: #5  void ServerMap::loadBlock(std::string *, v3s16, MapSector *, bool))
09:00 nrzkt       DEBUG STACK FOR THREAD 807432400:
09:00 nrzkt       #0  virtual void *CurlFetchThread::Thread()
09:00 nrzkt       DEBUG STACK FOR THREAD 807432800:
09:00 nrzkt       #0  virtual void *EmergeThread::Thread()
09:00 nrzkt       (Leftover data: #1  MapBlock *ServerMap::loadBlock(v3s16))
09:00 nrzkt       (Leftover data: #2  void ServerMap::loadBlock(std::string *, v3s16, MapSector *, bool))
09:00 nrzkt       (Leftover data: #3  void ItemStack::deSerialize(std::istream &, IItemDefManager *))
09:00 nrzkt       /wrkdirs/usr/ports/games/minetest/work/minetest-minetest-be358c0/src/util/pointer.h:279: void SharedBuffer<unsigned char>::drop() [T = unsigned char]: Assertion '(*refcount) > 0' failed.
09:02 * VanessaE  looks at nrzkt
09:02 est31       is it by chance main.cpp:882: int main(int, char**): Assertion '0' failed
09:02 VanessaE    [02-12 04:06] * nrz2@VEVn (nrz2@VE-Vanilla) has left #minetest   <--- that's your crash?
09:03 nrzkt       no no :p this is a crash on my own server
09:03 VanessaE    oh ok
09:03 nrzkt       it's not a problem on the client :)
09:03 * VanessaE  closes the ssh session that she was just about to open to check the server's log :P
09:06 VanessaE    and with that, I'm off to bed. :)
09:06 VanessaE    night
09:06 nrzkt       what does everybody think about #2285 ? could you reply to it
09:06 ShadowBot   https://github.com/minetest/minetest/issues/2285 -- No damage in creative mode
09:06 nrzkt       good night
09:07 kilbith     i'm thinking about replace those checkboxes by a dropdown
09:07 kilbith     exclusive modes
09:41 nrzkt       clang reports two unused variables: #2296 if nobody cares i merge tomorrow
09:41 ShadowBot   https://github.com/minetest/minetest/issues/2296 -- Remove unused variables to make clang happy by nerzhul
09:46 est31       gcc reports those too
09:46 est31       at least the post one
10:18 Megaf       nrzkt: Hi there, welcome to the team
10:19 Megaf       or is not him?
10:19 Megaf       nevermind
10:19 Megaf       I know we have a new core developer, who is he?
10:19 Zeno`       nrzkt
10:19 est31       where did you use him?
10:19 Zeno`       aka nazgul aka black rider
10:20 nrzkt       hi Megaf, thanks
10:20 Zeno`       it's not really nazgul but nazgul is easy to remember
10:20 Megaf       Ok, so I'm not mistaken
10:20 Zeno`       lol :)
10:20 Zeno`       I'll be back later
10:20 nrzkt       nrztk is ner'zhul but nrz is already taken then i add Kael'Thuzad suffix=> nrzkt
10:21 nrzkt       yes i'm a wow gamer
10:22 nrzkt       est31 what's the problem ?
10:22 nrzkt       #2275
10:22 ShadowBot   https://github.com/minetest/minetest/issues/2275 -- lua_api.txt should have a since field for every function
10:23 est31       if devs think the version doc shouldn't be added, this should be respected.
10:25 est31       I'm just trying to find a way to get feedback for changes I'm doing, not that they get rejected when they are finished.
10:28 Megaf       we could use github for documentation control too
10:28 Megaf       put the docs on git.io
10:28 nrzkt       http://dev.minetest.net/Translation isn't deprecated ?
10:28 Megaf       or github.io
10:29 nrzkt       #2052 is fine for me, but util/update-po.sh does many changes, not only on cz
10:29 est31       like http://rubenwardy.com/minetest_modding_book/lua_api.html ?
10:29 ShadowBot   https://github.com/minetest/minetest/issues/2052 -- Update czech translation by JakubVanek
10:33 est31       I think its wrong to maintain two different repos for documentation .. the wiki and the lua_api.txt. Best would be to have one, which autogenerates the wiki and a shorter api doc
10:34 nrzkt       i think everything must be on the wiki, this also permit to have a page history
10:35 est31       git has histories too
10:35 est31       wiki has the problem that you dont have the overview
10:35 kahrl       using exclusively the wiki would be fine if there was a way to attach wiki changes to pull request
10:35 nrzkt       yes but it's less userfriedly for mod developper and community, i think
10:35 kahrl       s
10:36 nrzkt       PR < 100, yeah !
10:57 nrzkt       kahrl #2295 is ok for you ?
10:57 ShadowBot   https://github.com/minetest/minetest/issues/2295 -- Fix gettext on MSVC by BlockMen
10:57 kahrl       Haven't really looked at it yet
10:57 gregorycu   celeron55: template <typename T> struct InferenceDisabler { typedef T type }; template <typename T> void Foo(typename InferenceDisabler<T>::type arg);  Foo(5); // error Foo<int>(5); ok
10:58 est31       ^ ?
10:59 gregorycu   There was a discussion on how to disable type inference for templated functions, I believe this is the clearest way
10:59 est31       oh ok
11:00 gregorycu   What was wrong with pipeworks?
11:01 est31       when
11:01 gregorycu   You got thanked, less than 24 hours ago for fixing pipeworks
11:02 est31       ah that one, there was a regression, you placed a locked chest, and it wasnt registered to you as owner
11:02 est31       was done by HEAD^
11:04 gregorycu   est31, do you have shaders turned on?
11:16 gregorycu   Oops
11:16 gregorycu   est31, do you have shaders turned on?
11:17 est31       why
11:17 est31       no
11:18 gregorycu   That was my next question
11:19 gregorycu   I'm trying to see there are many people suffering from a performance issue with shaders
11:21 est31       ok
11:26 gregorycu   Simple optimisations and it goes from 80fps to 120fps
11:30 nrzkt       i haven't seen performance issues on shader because my machine is too powered, but improving FPS could be great
11:31 gregorycu   Actually
11:31 gregorycu   This type of perf issue has nothing to do with your graphics card
11:32 gregorycu   Probably
11:32 gregorycu   Out of interest, what is the difference between your FPS with shaders turned on and off?
11:33 nrzkt       i can try at work now, because i have a intel GPU
11:34 gregorycu   What do you mean?
11:34 nrzkt       the FPS/shader issue will be shown
11:35 gregorycu   I don't understand
11:35 gregorycu   It's possible that more powerful graphics cards are more affected by this
11:38 celeron55   most systems do get slightly lower performance when shaders are enabled
11:39 celeron55   some don't, and some very old ones go unplayably slow
11:39 gregorycu   Yep, it's the degree of the performance degradation that I'm interested in
11:39 gregorycu   I wouldn't call 120 -> 80 slight
11:40 gregorycu   In fact, that's shaders, vs shaders with less constant updates
11:40 celeron55   can you define the test setup
11:40 celeron55   what settings?
11:40 celeron55   and how much to let the world load and so on
11:41 celeron55   getting consistent results takes a bit of effort
11:41 gregorycu   No options on the left, or middle,  shaders, bumpmapping, normalmaps and occlusion turned on
11:41 nrzkt       with my intel GPU: no shader: 60fps constant, with: 60 but sometimes lower to 40.
11:42 gregorycu   You have frame limiting turned on, don't you?
11:42 gregorycu   I look at an island, wait until the map loads (the frame rate to settle)
11:42 gregorycu   It's the same island, I don't touch the mouse while it loads
11:42 celeron55   my intel on linux also caps to 60 while minetest is configured at 120/240
11:42 celeron55   it'd probably require fullscreen to go further but that'll probably just mess things up so i won't even try
11:42 gregorycu   What is 120/20 ?
11:42 gregorycu   240
11:43 gregorycu   Max frame rates?
11:43 celeron55   fps_max = 240
11:43 celeron55   wanted_fps = 120
11:43 gregorycu   Ok
11:43 celeron55   those values make it run uncapped at most times
11:43 kilbith     or /set -n fps_max 0
11:44 celeron55   don't use -n
11:44 celeron55   has calinou spread that to every person in this community?
11:44 gregorycu   I have a 4 line change to limit the number of settings updates
11:44 gregorycu   err... shader updates
11:44 kilbith     yes, the tip comes from Calinou
11:44 gregorycu   If people try this, we can see if other people benefit from this improvement
11:44 celeron55   tell everyone to not use -n unless absolutely required; it's designed to make people not typo their settings
11:44 celeron55   with -n it will add new settings
11:45 celeron55   without -n it will check that it's an existing setting
11:45 gregorycu   Anyone game enough to try?
11:46 sfan5       "vblank_mode=0 ./minetest"
11:46 sfan5       run minetest like that to disable the forced vsync on intel
11:46 JakubVanek  yep
11:46 JakubVanek  to vblank_mode
11:47 celeron55   seems to do the trick
11:47 gregorycu   If anyone is keen: https://gist.github.com/gregorycu/5704c5bf1d2266193976
11:47 gregorycu   (this is not a solution, just a test to see if other people are affected by the same performance hit)
11:47 celeron55   so to everyone doing this test: set your viewing range to be locked to a value where the game loads terrain in a reasonable amount of time
11:47 celeron55   so that you are measuring the fpses at a constant draw distance
11:48 celeron55   (for me 200 is a good value for testing)
11:48 gregorycu   I'm excited about potential 50% perf improvement for pretty much no work
11:51 sfan5       w/o shaders: 100 fps w/ shaders: 80 fps
11:51 celeron55   50 with shaders, 70 without shaders, with shaders with patch varies between 60-70
11:51 kahrl       nrzkt: did you want to close #1148?
11:51 ShadowBot   https://github.com/minetest/minetest/issues/1148 -- Please include Lithuanian translation (lt.po) from http://translate.minetest.ru/projects/minetest/core/
11:52 celeron55   (the variance might come from the GPU getting heated up and throttling)
11:52 celeron55   so yes, there's a measurable difference
11:53 gregorycu   Ok, that's good enough for me
11:56 Megaf       nrzkt: est31: I think all doc should be in a git wiki,
11:57 gregorycu   I think you're quite likely to get higher variance at greater fps
11:57 gregorycu   (More data getting sent)
12:06 est31       Megaf, what does c55 think? other devs?
12:06 Megaf       I can't see why not use the git's built in wiki
12:07 Megaf       ^ celeron55
12:07 Megaf       kahrl: ^
12:07 Megaf       Opinions on using git's wiki as only doc?
12:07 celeron55   there is no wiki in git
12:07 celeron55   do you mean github
12:07 Megaf       yep
12:08 celeron55   it's too cramped
12:08 celeron55   it's meant for smaller amounts of documentation than what minetest uses
12:08 Megaf       https://help.github.com/articles/about-github-wikis/
12:09 Megaf       I think it's just perfect
12:09 gregorycu   The other option is to store it in source control, and use a tool to populate the wikis
12:10 Megaf       I believe we should give a try to github wiki thing and see how it looks
12:10 Megaf       if it doesn't look ok, than we just move on
12:11 celeron55   you haven't given even one good reason to use it over what there currently is
12:11 celeron55   also, in my opinion minetest's download must come with a compact reference documentation
12:12 celeron55   it saves the day when your internet is broken for whatever stupid reason
12:12 gregorycu   Just a text file?
12:13 celeron55   lua_api.txt works fine for that purpose; it also works fine for the purpose of having a canonical documentation which can be updated with reasonable effort when coding the actual implementation
12:13 gregorycu   PilzAdam: You wanna help out a brother with something?
12:13 celeron55   if a wiki is the canonical documentation, then people are likely to just not update it with the code
12:13 Megaf       and what about having the documentation in HTML4? Then it could be included in minetest and the very same file could be used on a site
12:13 gregorycu   Why 4? What about 4.1?
12:13 est31       the documentation is already in markdown
12:14 Megaf       4.x*
12:14 Megaf       I mean, not HTML5 to be lighter
12:14 celeron55   no this sounds like change for the sake of change
12:14 est31       thats the problem I think. The doc only gets updated when sb codes the implementation.
12:14 gregorycu   If this is an issue, maybe an github issue is best, I don't know what problems we are trying to solve
12:14 celeron55   what real-world problem are you solving?
12:15 est31       PRs should only be merged when both wiki and lua_api.txt have sufficiant docs
12:16 est31       thats my issue -- a missing good resource for the lua api
12:16 est31       neither wiki nor lua_api.txt fulfill this requirement
12:16 est31       wiki is outdated and incomplete
12:16 celeron55   you don't solve that by forcing more friction into development; you solve that by having people who maintain the extended documentation
12:16 est31       lua_api is just too concise
12:17 gregorycu   Do you mean dense?
12:17 est31       no, dense means everything is there
12:17 est31       but honestly I've tried to code only with the lua_api doc, and I've failed
12:17 celeron55   there's no way on earth that lua_api.txt could be extended so much that everyone would be happy with the extent of information in there
12:17 gregorycu   You mean concise in a negative sense, I don't understand what you mean?
12:17 celeron55   it requires a wiki
12:17 est31       Ive ended up either inspecting source or copying from other mods
12:17 celeron55   and there's no way developers would write that much information either
12:18 celeron55   these are practical constraints
12:18 gregorycu   concise: giving a lot of information clearly and in a few words; brief but comprehensive.
12:18 gregorycu   too concise?
12:18 gregorycu   Did you mean terse?
12:18 Megaf       there could be a wiki, wherever you want, and it could be converted to html and that covnersion included in minetest. So there will be only one thing to maintain and update
12:19 gregorycu   I actually have an idea
12:19 Megaf       or at least som articles of it
12:19 gregorycu   (Strange isn't it)
12:19 est31       they shouldnt write full examples and so on, just a sane api documentation with what is returned, when its returned, what the params are, and which minetest version supports the functionality
12:19 celeron55   wiki-only docs will create huge issues when different versions of minetest
12:19 gregorycu   Some sort of markup which is checked in (like lua_api) which populates the wiki, however, additions can be made to the wiki
12:19 celeron55   with the single canonical file which gets updated along with the code, the docs always match the code when you pull a version
12:19 celeron55   it has infinite upsides
12:20 celeron55   gregorycu: something like that could possibly work
12:20 est31       I also think that its great to have a single place to get an overview, and I also think that better doc should be bundled with minetest
12:20 gregorycu   The best sites are the ones with documentation, followed by user comments
12:21 est31       you can then auto-generate the wiki from the bundled doc
12:21 gregorycu   Where user comments are discussions about the API
12:21 gregorycu   Examples etc.
12:21 celeron55   gregorycu: yes, when it's done like that, the official doc can be quite short
12:21 est31       I also think I used "I also think" too often :)
12:22 celeron55   but it does require someone to implement the automation
12:23 celeron55   but every change requires someone to create something
12:23 gregorycu   lol
12:25 celeron55   anyway, so, i mean, feel free to try something, but there are reasons why things have been done the way they are
12:25 celeron55   don't go too idealistic because that rarely works
12:28 est31       current api doc is already helping alot
12:31 gregorycu   thanks, I wrote it myself
12:31 est31       ?
12:33 est31       and I guess there are more important issues to be resolved
12:34 est31       but some devs already object to me adding *version* documentation to lua_api.txt
12:34 celeron55   hmm what would that looks like?
12:35 est31       a simple (>=0.4.7) after the function name, or in its own line when the function name is a headline.
12:35 celeron55   with that tight format adding those to every function and field gets hard to read fast
12:35 est31       and when a param got added (param >=0.4.7)
12:35 celeron55   when most of the time they aren't important
12:36 est31       they are important when you want to know which version your mod supports minimum
12:36 gregorycu_  hmm...
12:36 gregorycu_  can't you just annotate?
12:36 gregorycu_  Or whatever the git version of that is?
12:36 est31       ?
12:37 gregorycu_  "annotate" allows you to see when a line was added in source control
12:37 gregorycu_  What commit did it
12:37 est31       you mean git blame?
12:37 gregorycu_  Well, added or changed
12:37 est31       that doesnt work unfortunately
12:37 celeron55   the information can be gotten that way, yes; it's not very accessible though
12:37 gregorycu_  So negative
12:37 JakubVanek  i'm not fan of microsoft but their .NET documentation on MSDN is really good
12:37 est31       you want to change the doc later on
12:38 est31       alot of ppl did typo fixes and so on
12:38 est31       then some people move whole sections
12:38 celeron55   JakubVanek: they probably have an entire documentation team working on it, we can't aim for the same
12:38 est31       and so on
12:38 celeron55   the problem is
12:38 gregorycu   We don't need a whole team
12:38 gregorycu   Our API isn't that massive
12:39 JakubVanek  i meant just documentation style
12:39 celeron55   if lua_api.txt is split to more files (so that it can be more verbose), we probably have to ditch markdown because nothing else than github actually supports links to other pages in markdown
12:39 gregorycu   Give us an example page
12:39 celeron55   and html is kind of crappy
12:40 est31       so we can render the markdown with a github compatible renderer?
12:40 celeron55   you have to be careful to not make the build process horrible so that it actually ends up in buildsd
12:40 JakubVanek  https://msdn.microsoft.com/en-us/library/system.console.writeline(v=vs.110).aspx
12:40 celeron55   builds*
12:40 neoascetic  doesn't bitbucket support links to relative pages?
12:40 JakubVanek  oh there aren't any examples on that page
12:41 JakubVanek  https://msdn.microsoft.com/en-us/library/system.net.sockets.tcpclient(v=vs.100).aspx
12:41 celeron55   how about something like https://readthedocs.org/?
12:43 est31       yes. mt already has sth like that http://rubenwardy.com/minetest_modding_book/lua_api.html
12:43 neoascetic  jekyll may be used for the docs
12:43 celeron55   it only does a part of the process and doesn't even allow user comments, though
12:43 est31       its basically lua_api.txt rendered into html
12:43 est31       we could bundle that
12:44 est31       every now and then (at least every release) sb does a "update bundled doc" commit
12:45 celeron55   that's messy
12:46 neoascetic  why write lua api description by our hands? if we have lua api, we may use some kind of docstrings for lua and generate it
12:46 kahrl       since when does rubenwardy's lua_api.html have a scrollbar under every piece of monospace text
12:47 celeron55   kahrl: it's always broken in different ways; come back some other day and it will have some other interesting surprise 8)
12:47 est31       :)
12:47 JakubVanek  i agree with neoascetic, documentation generated from comments above functions would be good
12:47 est31       ok ran out of ideas :)
12:47 JakubVanek  something like javadoc
12:48 celeron55   MT's lua API is defined partly in C++ and partly in Lua, so that'll take some integration work from the generator side
12:48 celeron55   it's probably impossible to generate for some parts
12:49 celeron55   and also, if the point was that we want more examples and stuff, then that's not a solution for it
12:49 celeron55   that will result in *less* examples, not *more*
12:49 celeron55   i'd ditch that idea
12:49 est31       yea
12:49 celeron55   maybe maintain a html documentation in place of lua_api.txt could workr
12:50 celeron55   with careful use of CSS it might be humanly possible to edit for people that don't want to manage the style
12:50 neoascetic  MT's wiki for developers not maintained at all, regardless it has one page per each API method
12:50 celeron55   and html supports links so that solves the issue of one huge file
12:50 est31       make it html only so that people dont have to generate it from some markdown?
12:51 celeron55   that's what i mean
12:51 celeron55   no generating of anything, ever
12:51 neoascetic  it is not question about how it should look like, but about who will maintain this documentation
12:51 celeron55   if it's in place of lua_api.txt, then it will work like lua_api.txt has worked
12:51 neoascetic  open wiki format in this case would be more handy
12:52 sfan5       i know this would make it more complicated, but why not generate HTML out of some "description"? this would solve the problem of people not wanting to work with e.g. CSS styles
12:52 celeron55   open wiki is a problem because then it's not in sync with the implementations and requiring developers to go write documentation in some other place than where they commit the code is extremely annoying
12:52 est31       we could save the double effort of having to change the doc at lua_api and the wiki
12:52 est31       also it messes up with the PR workflow
12:53 celeron55   sfan5: i think just using very simple html solves the issue better than using generators
12:54 sfan5       it probably will
12:54 neoascetic  non-generated documentation may be out of sync too
12:54 celeron55   no classes, no anything, and a CSS file that does its best to style every kind of tag properly
12:54 celeron55   but this still has the problem that there is no place for user comments
12:54 celeron55   with a wiki there is
12:55 celeron55   but given that currently the wiki is out of date at the user comment side too, i guess user comments are not important
12:56 est31       generation by javascript?
12:56 celeron55   and a special site can be created much later if needed
12:56 celeron55   est31: what do you mean
12:56 celeron55   having a html page which actually contains markdown, but also runs a javascript to turn it into html? 8)
12:57 est31       html is unchanged, doc goes to markdown file, opening the html
12:57 celeron55   you can't open other files using javascript if you open the file locally
12:57 est31       results in js converting the md to html
12:57 celeron55   at least chromium will disallow that unless you start it up with a special command line option
12:58 celeron55   s/open the file/open the html file containing the javascript/
12:58 est31       I think thats only for XMLHTTPRequest
12:59 celeron55   that's the only way for javascript to open a file for custom reading
12:59 neoascetic  may I ask? why approach with open wiki didn't work? how it is differs from html-based docs?
12:59 celeron55   neoascetic: not in sync with pull requests and the regular development flow
13:00 neoascetic  i. e. docs will lie in the repo, right?
13:00 celeron55   in my opinion still the best solution could be a wiki into which a base documentation from the code repository is updated, and which users can extend
13:00 est31       https://www.dartlang.org/articles/embedding-in-html/
13:00 celeron55   which gregorycu suggested
13:00 est31       something like that shouldnt be blocked by chrome
13:00 est31       more testing needed though
13:01 sfan5       est31: what code do you want to embed into the docs?
13:01 est31       md->html parsing code
13:02 neoascetic  github pages with jekyll may be used for this...
13:02 sfan5       we didn't want a html page that dynamically parses markdown
13:03 est31       thats only for ppl that have downloaded mt and dont want to convert ("build") it
13:03 est31       to actual html
13:03 neoascetic  yeah, md is pretty readable
13:03 est31       its an alternative to using html
13:03 celeron55   the problem is that md's links are unusable
13:04 neoascetic  I would prefer open docs in Vim instead of browser
13:04 celeron55   for most people
13:04 est31       I doubt this will result in something productive :(
13:04 neoascetic  do we have too many API methods for single page?
13:05 neoascetic  I would prefer single-page documentation
13:05 celeron55   they fit in lua_api.txt with the current verbosity, but if we want to add more information about them (version infromation, examples, whatever), the page becomes way too big
13:05 est31       I wanted to add a feature (introduced version), it was declined
13:05 celeron55   that's the base issue here
13:05 est31       yes
13:08 celeron55   or, not necessarily way too big, but way too big for many users probably, because again, md's in-page links neither work for most users
13:08 celeron55   hmm, could that still be reasonable
13:09 celeron55   it would become like tens of thousands of lines in the future years
13:09 celeron55   now it's thousands
13:11 celeron55   maybe if it would be split into like 10 main sections as separate files
13:11 est31       most packages I know have some automatic step in building docs
13:11 celeron55   then for the web it could be built into html or just viewed on github
13:11 celeron55   and it would still be relatively usable when viewed as .md files in a local repo
13:11 Zeno`       What is the new directory structure and where was it discussed?
13:12 celeron55   nobody has proposed any structure
13:12 Zeno`       #2277 suggests it has
13:12 ShadowBot   https://github.com/minetest/minetest/issues/2277 -- Main.cpp cleanup: Move ClientLauncher and InputHandlers to dedicated files by nerzhul
13:12 neoascetic  do we need this docs for previous versions?
13:12 celeron55   it would probably be something like doc/lua_api/hud.md, doc/lua_api/somethingelse.md and so on
13:12 celeron55   Zeno`: oh you refer to some other discussion
13:13 celeron55   neoascetic: users need, and developers need matching documentation when checking out a certain git version
13:14 celeron55   at least in my opinion
13:14 neoascetic  I think readthedocs in this case will be the best solution
13:15 neoascetic  http://bootstrap-datepicker.readthedocs.org/en/release/
13:15 neoascetic  for example
13:15 neoascetic  version selector in the footer
13:15 celeron55   where is that generated from
13:15 neoascetic  https://github.com/eternicode/bootstrap-datepicker/tree/master/docs
13:17 celeron55   the overall sturcture is quite like what i said (about 10 files)
13:17 celeron55   so that could be fed to readthredocs with reasonable results
13:17 celeron55   altough, it's prefectly readable on github too
13:19 neoascetic  build and uploading to readthedocs may be done in automatic way with travis-ci
13:19 est31       https://read-the-docs.readthedocs.org/en/latest/webhooks.html
13:19 celeron55   so does someone have thoughtful comments about splitting lua_api.txt to about 5-10 parts and then allowing people to make them more verbose?
13:20 neoascetic  est31 this even better
13:21 neoascetic  anyway, we need some way to keep this info synced with the code
13:21 celeron55   the way is the same as before: manually checking that whenever someone changes the lua api implementation, the commit must also update the documentation
13:21 celeron55   it has worked reasonably well in my opinion
13:22 neoascetic  great then
13:22 celeron55   only the wiki is considerably out of date
13:22 neoascetic  do we need it in this case?
13:22 neoascetic  I think Disqus comments or something else may be placed to readthedocs pages
13:23 celeron55   the API docs in the wiki probably become obsolete if this turns out well
13:24 celeron55   someone has to still draw a line in the verbosity though - i mean, rubenwardy's page is essentially a book about minetest modding
13:24 celeron55   the internal docs probably should be less verbose than that
13:24 est31       yes
13:26 celeron55   it's very easy to go more verbose; i mean, the lua api already is almost equally long if you add any kind of examples about anything
13:27 celeron55   i don't know what rubenwardy's intention is; if he's not wanting to make an actual printed book, he might want to switch to the internal docs in place of that one
13:28 neoascetic  I want docs about Engine itself be available with approach, not only lua docs
13:29 Zeno`       Why is there a rush to have #2277 merged?
13:29 ShadowBot   https://github.com/minetest/minetest/issues/2277 -- Main.cpp cleanup: Move ClientLauncher and InputHandlers to dedicated files by nerzhul
13:29 celeron55   neoascetic: we have doxygen for that
13:29 Zeno`       we discussed this yesterday and agreed that no harm could come from waiting
13:29 celeron55   it's not properly usable for the lua interface
13:30 Zeno`       It's not that I disagree with the commit (I like it)... but, I just don't understand the urgency
13:30 celeron55   and it's different in nature anyway (the other one is a public api while the other one is just internal implementation details)
13:30 neoascetic  I want something like on dev wiki with diagrams, images, etc. I guess it isn't available with doxygen
13:31 luizrpgluiz hello, how do I get the 0.4.12-dev version?
13:31 est31       I dont think its a rush zeno, its an attempt by nerzhul to make shorter PR lifetimes.
13:32 Zeno`       est31, no it's not. It's something that is, in essence, cosmetic and there has been no chance for all devs to have input
13:33 JakubVanek  git clone https://github.com/minetest/minetest.git; git checkout freeze-0.4.12
13:33 JakubVanek  ^ luizrpgluiz
13:35 Zeno`       est31, how can it hurt leaving that PR just a little bit? It doesn't add or fix anything
13:36 est31       so when it doesnt add, then its very simple?
13:36 celeron55   if you define how long it should wait for comments, then it's fine
13:36 celeron55   leaving it for an unknown time is bad
13:37 Zeno`       I said yesterday that maybe leaving it for 4 or 5 days would be good to give it a chance to be discussed
13:37 celeron55   i don't think anyone has the energy to discuss it though
13:37 Zeno`       hmm
13:37 celeron55   i at least don't want to even see it anymore (no matter if it's merged or not)
13:38 est31       how is that meant
13:39 Zeno`       Like I said, I agree with the commit.  I just don't see why it *must* be merged today
13:39 celeron55   i mean it's really just bikeshedding and if it will wait for comments, just comment to that to it and let it be; nobody wants to see the matter of whether it should wait for comments to be discussed because it's just wasting everyone's time
13:40 luizrpgluiz can someone pass me the changelog development minetest? on additions and bug fixes?
13:40 celeron55   http://dev.minetest.net/Changelog
13:40 celeron55   you should learn to use google
13:40 kilbith     luizrpgluiz: with a coffee with that ?
13:41 luizrpgluiz forgot to mention, it would be the changelog github because I can read the feature they are testing
13:43 celeron55   kilbith: lol
13:49 Zeno`       ok, if it's not worth discussing then I agree that #2277 should be merged
13:49 ShadowBot   https://github.com/minetest/minetest/issues/2277 -- Main.cpp cleanup: Move ClientLauncher and InputHandlers to dedicated files by nerzhul
13:50 Zeno`       it makes things neater
13:50 nrzkt       kahrl, yes i would want to close it, files are present no ? #1148 (sorry i went lunch)
13:50 Zeno`       and is something that should have been done already anyway.
13:50 ShadowBot   https://github.com/minetest/minetest/issues/1148 -- Please include Lithuanian translation (lt.po) from http://translate.minetest.ru/projects/minetest/core/
13:50 Zeno`       nrzkt, merge :p
13:50 luizrpgluiz how long before the official release of 0.4.12?
13:51 Zeno`       "When it's ready"? :D
13:52 celeron55   luizrpgluiz: nobody knows, it ended up being a bit of a mess
13:52 celeron55   i decided we'll just get back to development and 0.4.12 is released on the side, if it is at all
13:53 luizrpgluiz but what kind of confusion celeron55?
13:53 celeron55   some bug that i haven't cared to look into at all
13:54 nrzkt       ok Zeno` i merge it, (sorry 2 hours of IRC cross conversations to decode xD)
13:55 nrzkt       celeron55 you mean 0.4.12 will be released soon ? :)
13:56 celeron55   i don't know, i won't be doing it
13:56 JakubVanek  please wait with #2052, i'm doing last review
13:56 ShadowBot   https://github.com/minetest/minetest/issues/2052 -- Update czech translation by JakubVanek
13:56 nrzkt       Ok Jakub
13:56 nrzkt       for translations, how we merge it properly ?
13:56 celeron55   hmmmm probably has some idea about 0.4.12
13:56 nrzkt       i saw there is the update-po.sh, but does it needs to be run at each PR merge ?
13:56 celeron55   does Zeno` know anything about it?
13:56 Zeno`       About what?
13:57 JakubVanek  i already ran it and then called checkout -- to other languages
13:57 JakubVanek  ^ nrzkt (util/updatepo.sh)
13:58 nrzkt       comments for #2240, #2181 and #2285 ?
13:58 ShadowBot   https://github.com/minetest/minetest/issues/2240 -- Fix occlusion by mamoit
13:58 ShadowBot   https://github.com/minetest/minetest/issues/2181 -- Fix .zip extraction by ngosang
13:58 ShadowBot   https://github.com/minetest/minetest/issues/2285 -- No damage in creative mode
13:58 Zeno`       If you're talking about the main.cpp PR then of course I do. I refactored main() to not be 2000 lines in one function and I've also been looking at the PR for the last two days
13:58 celeron55   Zeno`: i mean 0.4.12
13:58 nrzkt       yes JakubVanek it's what i say, must i run it after merge or not ?
13:58 Zeno`       oh that
13:58 Zeno`       I think it's a Windows hold-up?
13:58 nrzkt       because when i tested your PR there was 2-4K modifications on all PO
13:59 Zeno`       I don't know. Blockmen has a PR though to fix some gettext() issues
13:59 nrzkt       Zeno` i merge 2277 now
13:59 Zeno`       I think the Windows thing was the hold up. But I don't know 100%
14:00 JakubVanek  nrzkt: yes for other languages (they're not updated), nothing will happen to czech
14:00 JakubVanek  because it already contains all strings
14:01 celeron55   it would probably be a good idea if someone would take the task of making 0.4.12 releaseable and telling when it is (and/or what needs to be done before it is)
14:01 nrzkt       tell me when your PR will be ready, i reviewed it seems very good
14:06 Zeno`       nobody is assigned to releasing 0.4.12?
14:06 celeron55   not really, unless you count hmmmm who turned quite inactive again
14:06 Zeno`       nrzkt, #2181 looks good
14:06 ShadowBot   https://github.com/minetest/minetest/issues/2181 -- Fix .zip extraction by ngosang
14:07 celeron55   i think he practically dropped it
14:07 Zeno`       I agree with PA that #2285 is a bug
14:07 ShadowBot   https://github.com/minetest/minetest/issues/2285 -- No damage in creative mode
14:08 Zeno`       I don't know about the occlusion fix... I'd need to look at the code more
14:09 Zeno`       celeron55, I don't know how the release process works. I assume it involves making "official" binaries for supported platforms?
14:09 celeron55   http://dev.minetest.net/Releasing_Minetest
14:09 celeron55   this is mostly up to date
14:10 celeron55   but more important than that process is making sure that it's ready for release
14:10 shadowzone  Good morning/afternoon/evening everbody!
14:10 celeron55   then that process is trivial to execute
14:10 nrzkt       hello shadowzone
14:11 celeron55   so the concern is that it should compile and work on all platforms where the release is going to be built for
14:11 Zeno`       I'll look at the freeze branch tomorrow and cherry pick any committed fixes I guess
14:11 Zeno`       I'm pretty sure the Windows build was the main hold up
14:12 celeron55   go ask some windows build person whether it's working properly
14:12 celeron55   i guess that'll be enough for it then
14:13 Zeno`       yeah, waiting on more feedback regarding #2295
14:13 ShadowBot   https://github.com/minetest/minetest/issues/2295 -- Fix gettext on MSVC by BlockMen
14:13 Zeno`       Krock has built the latest HEAD so his input would be nice also
14:14 celeron55   then if it seems to be ok for release, you can hilight me if you don't want to go through the release process yourself
14:14 celeron55   hmmmm is known to be able to do it properly too (i guess he can do it if he's around at the time)
14:14 shadowzone  *highlight* Sorry, I felt like I should correct that. So, pardon me.
14:15 Zeno`       I'll highlight it because I cannot seem to build with mingw (I need to use TDM-GCC for some reason)
14:15 celeron55   i don't have a working mingw setup either
14:15 shadowzone  Is Mingw windows?
14:16 celeron55   (i'd just ask some regular windows builder to make a build)
14:16 shadowzone  I have windows.
14:16 celeron55   the windows build details on that page are probably outdated anyway
14:20 Zeno`       mingw is windows and can also be cross-compiled, shadowzone
14:20 Zeno`       it just doesn't work for me and I don't know why
14:21 shadowzone  Zeno`, when I get to my desk, I could try it.
14:21 Zeno`       if you like :)
14:31 luizrpgluiz someone has issues where the link has the days of development minetest?
14:35 BlockMen    can someone confirm gettext works fine on linux with #2295?
14:35 ShadowBot   https://github.com/minetest/minetest/issues/2295 -- Fix gettext on MSVC by BlockMen
14:35 nrzkt       i test it now
14:37 BlockMen    thx
14:39 nrzkt       what do you need exactly ? it seems translations are working
14:40 BlockMen    thats all, actually. just wanted prevent breaking it the other way round
14:43 BlockMen    fine with mergin then, nrzkt?
14:44 nrzkt       seems good for me. Zeno` do you agree ?
14:51 Zeno`       BlockMen, save me some time. What was the memory leak?
14:53 BlockMen    Zeno`, not releasing tmp in wstrgettext
14:53 nrzkt       in previous version return directly
14:53 Zeno`       also... memset(buffer, 0, sizeof(buffer)); ... what is sizeof(wchar_t)?
14:54 Zeno`       I rather see this as ARRLEN (or whatever the macro in utils/ is called which should do sizeof(buffer) / sizeof(*buffer))
14:54 Zeno`       err *
14:55 Zeno`       so maybe the macro is not so good, but sizeof(wchar_t) is still important
14:55 Zeno`       actually scratch that... I'm tired :/
14:58 Zeno`       looks ok to me
14:59 Zeno`       const wchar_t* tmp = wgettext(text.c_str());      -->   const wchar_t *tmp = wgettext(text.c_str());
15:04 BlockMen    changed. when do i use this new "ok" thing?
15:04 Zeno`       and ok thing?
15:04 Zeno`       err ok thing?
15:04 est31       +1 i guess
15:05 BlockMen    some commits have now <commitmsg>. ok @<coredev>
15:05 Zeno`       oh that was a temporary mistake
15:06 BlockMen    ic. then im pushing now
15:06 Zeno`       just let me check one thing?
15:06 BlockMen    k.
15:06 nrzkt       sorry BlockMen, i used openbsd method for validation :)
15:08 fireglow    "<coredev>@"
15:09 nrzkt       yes, the method, not the syntax :p
15:09 nrzkt       :q
15:10 Zeno`       BlockMen, +1
15:10 Zeno`       go for it :p
15:12 BlockMen    done
15:25 Zeno`       can someone look at the freeze branch?
15:25 Zeno`       I'm 99.9999999999% sure I pushed properly but github is displaying a strange thing to me
15:26 nrzkt       ?
15:26 Zeno`       just see if the history looks right I guess
15:26 Zeno`       I dunno why github is wanting me to compare and pull request :/
15:27 Zeno`       very strange
15:28 Zeno`       maybe it (github) wants me to compare to master... *shrug*
15:29 Zeno`       https://github.com/minetest/minetest/tree/freeze-0.4.12      looks fine to me anyway
15:29 PilzAdam    celeron55, I have a question about the freeze branches: will the commit in the freeze branch be tagged on release? or will it be merged to stable-0.4 and then tagged? how will future freeze branches be merged to stable then, if we don't have a linear history anymore?
15:30 Zeno`       I personally think the branch should be renamed and also have tags
15:30 * Zeno`     is afk
15:31 Zeno`       dammit BlockMen! I was watching something on utube and now doing all that I've closed firefox
15:32 BlockMen    Zeno`, hmm?
15:32 BlockMen    what did i do?
15:32 Zeno`       nothing I guess. Don't worry ;) (I was joking)
15:33 nrzkt       i agree with Zeno`. Using a stable branche with tags is proper
15:33 BlockMen    Zeno`:P
15:33 nrzkt       BlockMen, Zeno`, review on #2169 please
15:33 ShadowBot   https://github.com/minetest/minetest/issues/2169 -- Minor fixes in translations by ngosang
15:35 JakubVanek  nrzkt: PR #2052 is ready
15:35 ShadowBot   https://github.com/minetest/minetest/issues/2052 -- Update czech translation by JakubVanek
15:35 nrzkt       ok, i review and merge
15:36 JakubVanek  updatepo.sh will be needed also after merging #2169
15:36 BlockMen    nrzkt, #2169 seems ok.
15:36 ShadowBot   https://github.com/minetest/minetest/issues/2169 -- Minor fixes in translations by ngosang
15:36 ShadowBot   https://github.com/minetest/minetest/issues/2169 -- Minor fixes in translations by ngosang
15:39 nrzkt       after 2169 =>  24 files changed, 8263 insertions(+), 5265 deletions(-) arg
15:39 nrzkt       argh
15:39 nrzkt       does somebody update this file sometimes ?
15:39 Zeno`       Tesseract does I think
15:40 nrzkt       i let him doing that, i only merge the PR
15:40 Zeno`       Don't take my word for it though :)
15:44 celeron55   PilzAdam: the freeze branch won't exist forever so it must be the commit on the stable branch
15:44 celeron55   and using merge commits on the stable branch is probably a good idea(?)
15:45 celeron55   this is new to me too so i'm not exactly sure
15:45 celeron55   there's probably some best way of doing itt
15:45 celeron55   it*
15:45 nrzkt       i'm using a pull from feature freeze to stable at work it works well
15:46 nrzkt       but we must be careful if we do a commit on master which must be backported to stable
15:48 nrzkt       #2052 and #2169 have been merged, thanks
15:48 ShadowBot   https://github.com/minetest/minetest/issues/2052 -- Update czech translation by JakubVanek
15:48 ShadowBot   https://github.com/minetest/minetest/issues/2169 -- Minor fixes in translations by ngosang
16:05 BlockMen    sfan5, whats your opinion on https://github.com/minetest/minetest_game/pull/420?
16:06 sfan5       looks good, BlockMen
16:19 BlockMen    sfan5, PilzAdam : fine with closing game#382 ?
16:19 ShadowBot   https://github.com/minetest/minetest_game/issues/382 -- Update nodes.lua by TheShadowZone12
16:20 sfan5       yup
16:20 PilzAdam    yes
16:20 shadowzone  Go ahead, it's been open and idle for a while, just close it.
16:41 BlockMen    PilzAdam, minetest_game maintainer rule changed: 50% present maintainers or at least 2 have to agree http://dev.minetest.net/minetest_game_Development
16:41 BlockMen    ~tell nore minetest_game maintainer rule changed: 50% present maintainers or at least 2 have to agree http://dev.minetest.net/minetest_game_Development
16:41 ShadowBot   BlockMen: O.K.
16:41 BlockMen    ~tell Tesseract minetest_game maintainer rule changed: 50% present maintainers or at least 2 have to agree http://dev.minetest.net/minetest_game_Development
16:41 ShadowBot   BlockMen: O.K.
16:46 rubenwardy  I sugges the title for #2291 is changed to "Engine API for Mobs".
16:46 ShadowBot   https://github.com/minetest/minetest/issues/2291 -- Idea : C++ Mobs.
16:47 nrzkt       done
16:47 rubenwardy  Awesome
16:48 BlockMen    PilzAdam, why? https://github.com/minetest/minetest_game/blame/master/mods/default/nodes.lua#L1101
16:48 BlockMen    those were ment to be overrideable
16:49 PilzAdam    https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L1689
16:49 PilzAdam    they are overrideable
16:50 nrzkt       rubenwardy, it's not my first priority, but when we fixed some of awaiting PR's and finished the network rework i'll look at this
16:50 PilzAdam    it's the same weird thing with lava and water viscosity: why are some random node fields stored in global variables?
16:51 PilzAdam    everything can be read from minetest.registered_nodes and everything can be overridden by minetst.overide_item
16:51 rubenwardy  Networking is way more important, and general bug catching.
16:54 BlockMen    PilzAdam, so instead overriding one global i have at least to override a whole function and the item itself.
16:54 BlockMen    sorry, but that is no improvment
16:57 PilzAdam    oh, right, it's accessed from a function and not stored in nodedef...
16:58 PilzAdam    how about storing it in the nodedef then?
16:58 rubenwardy  #2299
16:58 ShadowBot   https://github.com/minetest/minetest/issues/2299 -- Fix minetest.item_eats replace_with_item, fixes #2292 by rubenwardy
17:00 rubenwardy  ^ that should go into the feature freeze
17:02 est31       also, consider #2225
17:02 ShadowBot   https://github.com/minetest/minetest/issues/2225 -- Fix for getCraftRecipes implementation is very slow by gregorycu
17:02 BlockMen    then you would still need to override the node/item and a function, e.g. for the furnace. some things are better stored in globals
17:02 est31       not for the freeze but for 5.0
17:04 PilzAdam    rubenwardy, not a bug
17:04 PilzAdam    BlockMen, no, you would just need to override the field that holds the formspec then
17:04 PilzAdam    since the function would access that field
17:06 BlockMen    so you still need more code. also, even if you were right you broke mod compatibility
17:06 BlockMen    so i will revert it
17:07 BlockMen    or since you are maintainer aswell, feel free to do it
17:07 PilzAdam    is there a mod that is broken?
17:07 PilzAdam    also that field was not documented in game_api.txt
17:07 BlockMen    e.g. the crafting mod(pack)
17:08 BlockMen    and the game_api.txt was generally incomplete (see tree growing functions)
17:08 PilzAdam    hm... I still think it is not the correct way to do this, though
17:08 PilzAdam    also it doesn't really require more code
17:09 PilzAdam    minetst.override_item("default:chest", {formspec="lalala"})  is not a lot more than  default.chest_formspec = "lalala"
17:11 BlockMen    wut? you have to override on_construct, not  "formspec", so you have to copy everything else in this function just for the formspec
17:11 BlockMen    and furthermore, you broke an existing feature, so it doesnt matter. until we dont break compatibility in general it has to be fixed
17:12 PilzAdam    what I suggested is: don't store the formspec string in the nodedef table instead of a separate variable
17:12 PilzAdam    s/don't//
17:12 PilzAdam    on_construct would access the nodedef field instead, so other mods would only need to override this field
17:12 rubenwardy  It is when it replaces all of the stack, rather than the item which you just ate.
17:13 PilzAdam    it's a cleaner way since the formspec is a part of the chest definition
17:13 BlockMen    doesnt work for furnace, https://github.com/minetest/minetest_game/blob/master/mods/default/furnace.lua#L6
17:14 PilzAdam    for the furnace it would be a function in the nodedef then
17:14 BlockMen    then add some legacy for the canceled functions
17:15 PilzAdam    hmm... I don't see a good way to do this in a backward compatible way :-/
17:16 PilzAdam    or is there a way to make the furnace field in the chest nodedef a pointer to default.chest_formspec?
17:16 PilzAdam    Tesseract, ^ ?
17:16 Tesseract   Hmmm?
17:17 Tesseract   furnace = default.chest_formspec?
17:17 PilzAdam    that would copy it
17:18 PilzAdam    it needs to be backward compatible for mods that want to override default.chest_formspec
17:19 est31       for clarification, the version after 0.4.12 will be called 5.0?
17:20 Tesseract   You could use some setmetatable hackery, but builtin uses setmetatable, so you would have to be carefull with how you did it.
17:21 Tesseract   If mods were using an undocumented API then it's their problem if their mod breaks though.
17:22 est31       or will it be called 0.5.0
17:23 PilzAdam    est31, version numbers don't matter
17:23 est31       so its not decided yet
17:24 est31       nvm, you are right, they dont matter
17:26 BlockMen    Tesseract, _game had never a documentation, just since 0.4.10 there is a small base for it. now blaming mods using features is nonsense
17:27 PilzAdam    I agree that breaking mods is not good
17:42 Tesseract   *shrug*  Compat will take 5-20 lines of metatable code.
17:50 rubenwardy  #1843
17:50 ShadowBot   https://github.com/minetest/minetest/issues/1843 -- Add setting to customise stack max by rubenwardy
17:57 Tesseract   PilzAdam, BlockMen: >30 lines actually: http://pastebin.ubuntu.com/10191415/
18:35 nrzkt       i'll commit a fix for #1573 directly, including Calinou and Kahrl change (with a little change for french)
18:35 ShadowBot   https://github.com/minetest/minetest/issues/1573 -- Add French line to the .desktop file
19:12 Tesseract   celeron55, PilzAdam:  For the stable-0.4 non-linearity issue:  we could make stable-0.4 a tag that's assigned to the most recent freeze-0.4.x branch.
19:13 Tesseract   I think that would solve that issue.
19:13 PilzAdam    updating tags is not easy
19:13 PilzAdam    since everyone has copies in their local repos
19:14 PilzAdam    an a simple pull does not update the tags
19:14 Tesseract   git pull --tags
19:14 Tesseract   Not all that much harder.
19:14 celeron55   why does it need to be linear?
19:14 PilzAdam    in order to update tags everyone must explicitly remove the old ones (AFAIK)
19:14 Tesseract   celeron55: Because force-pulling is a pain.
19:14 celeron55   merges are made for this exact use case
19:15 celeron55   non-linear merges
19:15 Tesseract   PilzAdam: Hmmm, so same issue, just slightly different.
19:15 Tesseract   Yeah, merges might be best.
19:15 celeron55   you could of course not want to use merge for it, but it would be dumb because it's made for it
19:16 celeron55   at least this is how it's looking to me
20:21 nrzkt       has anybody experience #2300 ?
20:21 ShadowBot   https://github.com/minetest/minetest/issues/2300 -- SharedBuffer ::drop() [T = unsigned char]: Assertion (*refcount) > 0 failed.
20:27 est31       no, I had some crashes but not that one
20:28 Tesseract   nrzkt: No, but I think I've heard of that before.
21:43 nrzkt       assert((*refcount) > 0); failed if refcount <= 0, no ?
21:45 nrzkt       the bug seems simple... in drop() we have this assert and if refcount == 0 we do something, but the assert triggers if refcount == 0...
21:46 nrzkt       no. In fact this is correct because we have the refcount-- before
21:46 nrzkt       but is this possible than one thread use same sharedbuffer as another ? then if one decrement the other trigger the refcount assert ?
21:58 * Tesseract just got another player at absurd coordinates (-200000+) https://forum.minetest.net/viewtopic.php?f=10&t=7351&p=170032#p170032
22:02 est31       Tesseract, is the change to the worldedit command I did now ok for you?
22:02 est31       https://github.com/Uberi/Minetest-WorldEdit/pull/71
22:03 Tesseract   est31: -> #minetest
22:48 est31       #2301 ready for review :)
22:48 ShadowBot   https://github.com/minetest/minetest/issues/2301 -- Fix crash on passing false as value in table to table.copy(t) by est31
23:42 luizrpgluiz hi
23:49 Tesseract   luizrpgluiz: This is a serious development channel, not a casual chatting channel.  If you don't have something helpful to say don't say anything.