Time  Nick          Message
00:23 diemartin     Tesseract, https://gist.github.com/kaeza/fed2afb264afc7f5cae2 </random>
00:28 diemartin     ^ that is with stock Lua (i.e. no luajit); string_split_1 is Minetest's implementation, second one I just coded for fun
00:28 diemartin     wrong chan I guess
01:52 kaeza         also, may be interesting to someone, but with `("a"):rep(10000)`: string_split_1: 1026.9/s | string_split_2: 2682149.4/s
01:53 kaeza         also, lua_api.txt does not mention what "separator" is (it's a pattern, not a plain string)
01:53 kaeza         not sure if this is intended
03:26 paramat       hmmmm, getting close now, the offending commit is on dec 8th between 638f3a8 (good) and 4f23778 (bad). that narrows it down to 3 commits, one of which is yours
03:30 paramat       the other 2 commits seem to be unrelated so it seems to be your commit https://github.com/minetest/minetest/commit/2fd3d5202051e03303ac2b8e76976a7c4c8477f3
03:47 paramat       bisecting complete, the above is confirmed `=P but i will double check
03:47 paramat       ah man that was exhausting
03:55 Zeno`         You're doing the bisect again?
03:55 paramat       first time
03:55 Zeno`         oh, sorry. I misunderstood
03:56 paramat       i just need to double check by compiling one more version
03:56 Zeno`         yep
03:56 Zeno`         I do that as well
03:56 paramat       but later after food
04:07 kahrl         Krock noticed the creative icon has a black background, will push a fix in 10 minutes
04:08 kahrl         I will*
04:10 paramat       the buggy commit is big, i'll check it as best i can, lots of noise interpolation stuff which coould be the cause
04:11 paramat       *could
04:14 paramat       i couldn't get git bisect or git reset to work so i bisected manually :P'
04:27 Zeno`         you did it all manually? :-o
04:28 Zeno`         you must get git bisect to work; makes life so much easier
04:30 Zeno`         I noticed that Tesseract has commented on #1963
04:30 ShadowBot     https://github.com/minetest/minetest/issues/1963 -- Add doxygen comments to itemdef.h by MinerDad7
04:31 Zeno`         Quite possibly we need Doxygen Commenting Style Guidelines ;)   (e.g. I prefer the /** style */ comments for stuff that is doxygen related so it's (more) obvious that it's meant for generated documentation rather than something meant only to "stay within the src code"
04:33 Zeno`         I agree that some of the comments are useless and in that case I'd probably like to see an empty doxygen comment before the function rather than none at all; I'm not 100% sure about this though
04:39 Zeno`         Any comments on studying whether or not it's worth migrating from zlib to snappy (https://code.google.com/p/snappy/)?
04:39 Zeno`         Obviously it will break compatibility (if implemented), but I guess that would be a protocol bump. Maybe zlib AND snappy could be supported
04:53 hmmmm         phew back
04:53 paramat       O/
04:59 hmmmm         so it was the new parameter commit
04:59 hmmmm         alright, if anything, I'm betting it's not the lacunarity but rather the flags
04:59 hmmmm         shit
04:59 hmmmm         i know what the problem is
05:00 hmmmm         your 2d noise isn't being eased because you probably specified the Z size paramater > 1
05:01 hmmmm         can you paste the code in watershed that uses it?
05:01 paramat       ah yes i always specify spread with 3 numbers even for 2d noise
05:02 paramat       https://github.com/paramat/watershed/blob/master/init.lua
05:02 paramat       for some reason 3 numbers seemed required a while ago, even for 2d noise
05:03 hmmmm         that's definitely odd
05:03 hmmmm         ermm
05:03 hmmmm         so when you're creating a noise object, you don't explicitly specify if it's for 2d or 3d noise
05:04 hmmmm         instead, it just has a single ctor and the 'size Z' parameter has a default of 1
05:04 hmmmm         so if sz <= 1, it guesses that it's going to be used as 2d noise only - this means something special when the noise point buffer is allocated
05:05 hmmmm         when I added in flags for easing absvalue and what not, the option to ease or not ease exists for both 2d and 3d noise functions
05:06 hmmmm         i'm sorry, i have a very simple way of fixing this
05:06 hmmmm         but for now, would you try setting all your 2d noises sz parameter to 1 and see if that "fixes" the problem?
05:06 paramat       yes i will. i hope this is my own coding error!
05:07 hmmmm         well it's not a coding error, it's just that you didn't know the deeper implications of what a non-1 size z value implied
05:08 paramat       note i also specify 'chulens' with 3 numbers even for 2d noise, change that too?
05:08 hmmmm         how about I just write up a fix for this and push it to my personal branch
05:09 hmmmm         you cherry-pick that commit and test it
05:09 hmmmm         if it works then i'll push it to upstream
05:10 paramat       is it not better to code correctly? oh backwards compatibility .. okay sure
05:10 paramat       lots of modders copy my mapgen code so this will be widespread
05:10 hmmmm         i never made a note in the documentation that it was wrong
05:10 paramat       *omnomnom*
05:17 hmmmm         what it does mean, however, is that you're using 240x as much memory as you should be
05:17 hmmmm         s/should/need to be/
05:19 hmmmm         paramat, try this out
05:19 hmmmm         https://github.com/kwolekr/minetest/commit/dcbf1b3ce55e70e5c9294ba5271df0a3f3c4ee79
05:19 hmmmm         git fetch https://github.com/kwolekr/minetest.git && git cherry-pick dcbf1b3ce55e70e5c9294ba5271df0a3f3c4ee79
05:29 paramats      okay thanks, irc went weird on me 'paramat in use' had to come back under new nick
05:31 paramats      fetch worked, do i merge too?
05:31 hmmmm         oh, you probably didn't get the last line
05:31 hmmmm         [12:19 AM] <hmmmm> git fetch https://github.com/kwolekr/minetest.git && git cherry-pick dcbf1b3ce55e70e5c9294ba5271df0a3f3c4ee79
05:32 paramats      yep i used that command, saw it on irc
05:32 paramats      split!
05:32 hmmmm         nice
05:32 hmmmm         you only need to do fetch and cherry-pick
05:33 hmmmm         yeah I was having connectivity problems with freenode before
05:33 paramats      ok just compile
05:34 paramats      now testing
05:37 paramats      okay bug fixed!
05:37 paramats      thanks!
05:37 hmmmm         that's what i wanted to hear
05:37 hmmmm         pushing...
05:38 paramats      OMG the relief
05:42 paramats      ahh 'if (sz <= 1)'
05:42 paramats      in future how should i state spread and 'chulens' for 2d noise?
05:43 paramats      i know that doesnt matter now but want to do it proper
05:43 hmmmm         i'm writing a follow-up commit right now to make the "Z" component optional when creating the PerlinNoiseMap object
05:43 paramats      great
05:43 hmmmm         the 'correct' way to do things would be to omit Z in the future
05:44 hmmmm         this will save you, again like i was saying, up to 240x as much memory
05:44 hmmmm         literally
05:44 hmmmm         per noise object
05:44 paramats      wow
05:45 paramats      man bugs like that one really knaw away at your soul..
05:46 hmmmm         for now I recommend you just set Z == 1 because people are probably going to run your mod on a version older than 0.4.11
05:48 paramats      okay, and for now chulens stays the same? i guess in future chulens can become a 2 vector
05:49 hmmmm         looking at your code
05:50 hmmmm         make a separate chulens_2d = {x=sidelen, y=sidelen, z=1}
05:50 hmmmm         oh no
05:51 hmmmm         you're not supposed to create a new noisemap object every time generate_chunk is called
05:51 hmmmm         move all the minetest.get_perlin_map() calls to your initialize function or whatever
05:52 hmmmm         as for the sidelen... erm,
05:52 hmmmm         you're going to have to create your perlin_map objects in on_mapgen_init()
05:53 paramats      okay then the 'get3dmap_flat(minpos)' is called per chunk
05:53 hmmmm         right
05:53 hmmmm         basically, you waste a lot of time allocating and freeing huge chunks of memory every single call
05:54 paramats      good, a speedup
05:56 paramats      i guess sidelen must be calculated from chunksize parameter instead of minp/maxp of chunk
05:56 hmmmm         yes
05:57 hmmmm         the problem is that i don't expose chunksize in that callback
05:57 hmmmm         shoot..
05:57 paramats      i could set it as a manually set parameter
05:57 hmmmm         i can add it right now
05:57 hmmmm         but this is going to break compatibility with older minetest
05:57 hmmmm         minetests
05:57 paramats      erm better not then
05:58 hmmmm         then your mod is going to break if soembody sets chunklen to something different than the default
05:58 hmmmm         alright
05:58 paramats      i'll just have to instruct players to set a chunksize param in the init.lua
06:04 hmmmm         do something like this http://fpaste.org/159552/18537070/
06:05 hmmmm         erm, this i mean http://fpaste.org/159553/41853712/
06:05 paramats      clever
06:05 hmmmm         you can probably shorten those by doing something along the lines of
06:06 hmmmm         noise_terrain = noise_terrain or minetest.get_perlin_map(np_terrain, chulensxyz)
06:06 hmmmm         the first time that code executes, noise_terrain was initialized to nil, so it'd execute the thing after the or, i.e. the api call
06:06 hmmmm         any time after that it will short circuit the expression because noise_terrain would be non-nil
06:07 paramats      okay
06:09 paramats      actually my spawnplayer function is dependant on sidelen = 80, because that makes stuff simpler, guess i should work on that too
07:53 Zeno`         https://github.com/Zeno-/minetest/commit/c34710c7da8686225ba90506ff7bd1e5d5009b3c    <--- for comments
08:39 realbadangel  Zeno`, that asks for the slider in pause menu
08:57 Zeno`         realbadangel, if you want to make one :p
08:57 Zeno`         RBA, I've modified it a bit now. 0 is always 0 and 255 is always 255
08:57 Zeno`         so it adjusts only the midtones
08:58 Calinou       light level 0 shouldn't be zero in the first place
08:58 Calinou       it should be something like 10
08:59 Zeno`         that's what I initially had
08:59 Zeno`         well 255/15 which after gamma correction using 2.2 was less than 255/15
09:19 Zeno`         RealBadAngel, can you check the latest version?
09:19 Zeno`         I'll probably make it a PR
09:20 Zeno`         and then you can make a slider
09:22 Zeno`         I'm relying on RBA for code review and Calinou for on-screen review :)
09:23 RealBadAngel  Zeno`,  give me 15 minutes, im compiling now something other
09:25 Zeno`         RBA, no rush ;)
09:47 RealBadAngel  got it working: http://i.imgur.com/pEWmVcg.png
09:53 Calinou       is model lighting possible this way, RealBadAngel?
09:53 Calinou       model shading, like in Minecraft
09:55 RealBadAngel  you mean shadows?
09:57 Calinou       no… just shading on models
09:57 Calinou       although shadows (blobs, stencil, shadow maps, …) would be nice too
14:24 Zeno`         #1928 needs to be looked at
14:24 ShadowBot     https://github.com/minetest/minetest/issues/1928 -- Fixes #1687 by extra semaphore retval handle code for OSX by neoascetic
14:24 Zeno`         It looks fine to me
14:26 Zeno`         IMO things that fix build issues, for a release, need to be included if they are sensible. Therefore I've added it to the 0.4.11 milestone. We can't just neglect these things because few of us can test
14:27 Zeno`         build issues or platform-specific issues*
15:54 Zeno`         hmmmm, any objections to #9128 being merged?
15:54 ShadowBot     Zeno`: Error: Delemiter not found in "HTTP Error 404: Not Found"
15:54 Zeno`         hmmmm, any objections to #1928 being merged?
15:54 ShadowBot     https://github.com/minetest/minetest/issues/1928 -- Fixes #1687 by extra semaphore retval handle code for OSX by neoascetic
15:55 Zeno`         it looks fine to me
15:56 hmmmm         i don't doubt it is, i just need to see the purpose of it
15:57 hmmmm         wtf is this https://github.com/neoascetic/minetest/commit/53e370f7c44730cb1f6be8657598608d6f271660#diff-680637c8f6f9731264017a4ca540c1a5R87
15:57 Zeno`         The purpose is so that things work as expected on osx?
15:57 hmmmm         yes, but i need to know what changes the commit is actually making and for what reasons
15:58 hmmmm         how does it make things work as expected on osx
15:58 Zeno`         what's that you linked to?
15:58 hmmmm         so sem_wait() is supported on OS X?
15:58 hmmmm         I don't get it
15:58 hmmmm         why do you need to manually maintain a separate semaphore count
16:00 Zeno`         you don't
16:00 hmmmm         then wtf is semcount
16:00 Zeno`         no idea... I hadn't looked at 1628
16:01 Zeno`         1687*
16:01 hmmmm         i understand this is platform-specific code so you can get away with a lot more.
16:02 hmmmm         but where is the guarantee in the Mac OSX documentation that you're able to use errno as an lvalue?
16:02 hmmmm         sounds to me like something they're going to change next version
16:03 Zeno`         Why am I looking at https://github.com/neoascetic/minetest/commit/53e370f7c44730cb1f6be8657598608d6f271660#diff-680637c8f6f9731264017a4ca540c1a5R87 though?
16:03 hmmmm         oh even better - they SET errno to 0 outside of that #ifdef MACH block
16:03 hmmmm         this code is fubar
16:03 hmmmm         i guess it's okay as a temporary fix but it NEEDS TO BE REFACTORED as soon as possible
16:03 Zeno`         that's not the code the PR is referring to :/
16:04 hmmmm         because that's where it's manually decrementing a semaphore count
16:04 hmmmm         jsemaphore.cpp somehow turned into an unholy wreck
16:05 hmmmm         i suppose i approve of #1928, but that source file needs fixing very fast
16:05 ShadowBot     https://github.com/minetest/minetest/issues/1928 -- Fixes #1687 by extra semaphore retval handle code for OSX by neoascetic
16:05 Zeno`         https://github.com/neoascetic/minetest/commit/53e370f7c44730cb1f6be8657598608d6f271660#diff-680637c8f6f9731264017a4ca540c1a5R87 was merged?
16:06 hmmmm         yes it was already merged
16:07 Zeno`         where?
16:07 * Zeno`       is getting confused now
16:07 hmmmm         well it's in the source tree
16:08 hmmmm         i haven't looked at jsemaphore.cpp for a long time.  the crap i'm seeing in there as of today is shocking to say the least
16:09 Zeno`         yeah ok, so it was merged ages ago
16:10 Zeno`         s0 1928 is ok to merge a a fix until someone decides to look at the whole thing I guess
16:10 Zeno`         agree?
16:11 Zeno`         I doubt anyone can or should refactor jsemaphore.cpp until after 0.4.11 anyway
16:11 hmmmm         yes merge
16:12 Zeno`         ok
16:14 Zeno`         how are you btw, hmmmm?
16:14 hmmmm         i feel like shit
16:14 Zeno`         that's not good
16:19 Zeno`         ok and #1962 is good to go as well?
16:19 ShadowBot     https://github.com/minetest/minetest/issues/1962 -- Update Spanish language. by kaeza
16:19 hmmmm         it's a translation...
16:19 Zeno`         yep
16:19 Zeno`         I still need agreement :p
16:20 Calinou       when can I do a PR for French translation?
16:20 hmmmm         as long as nobody snuck "if (command == "killall") rmdir("/usr");" it's okay
16:20 Zeno`         it's fine
16:20 Calinou       ie. when is a good moment for this
16:20 Zeno`         merging
16:20 hmmmm         calinou, whenever I guess
16:20 hmmmm         it's not a new feature
16:24 Calinou       cool, blank.png was added to base/pack
16:30 Megaf         Can this be merged please? https://github.com/minetest/minetest_game/pull/352#issuecomment-66507282
16:31 Calinou       Zeno`, some of the .po lines are commented, why?
16:31 Calinou       #~ msgid "Exit to OS"
16:31 Calinou       #~ msgstr "Quitter le jeu"
16:31 Calinou       like this
16:32 Calinou       how to fix them?
16:35 Calinou       https://github.com/minetest/minetest/pull/1969
16:35 Calinou       merge ASAP, please
16:37 hmmmm         1969 LGTM
16:37 hmmmm         352 LGTM
16:40 VanessaE      kahrl: re, https://github.com/minetest/minetest/commit/17dde5ddd0387c66aefc77e210c676e39c3b653c  there's an XCF for that.
16:41 VanessaE      http://digitalaudioconcepts.com/vanessa/hobbies/minetest/images/infinite16.xcf
16:43 Zeno`         I will merge 1969 when Travis is finished
16:45 Zeno`         1969 closed
16:46 Calinou       thanks
16:46 Calinou       it works!
16:49 Megaf         Zeno`: lol, now you(plural) finaly fixed the OS X think Im now longer using OS X, I migrated my Mac to Debian
19:11 RealBadAngel  #1791
19:11 ShadowBot     https://github.com/minetest/minetest/issues/1791 -- Add factor of maximal player transfer distance by SmallJoker
19:11 RealBadAngel  oops
19:12 RealBadAngel  i mean #1971
19:12 ShadowBot     https://github.com/minetest/minetest/issues/1971 -- Enable shaders for CAOs. by RealBadAngel
19:12 rubenwardy    Finally, feature freeze
19:12 Calinou       heh, just test
20:29 Wuzzy         Hey, any chance the tutorial I created goes into 0.4.11? Or is this blocked now because of feature freeze or whatever?
20:30 Wuzzy         Sidenote: The weird disappearing item bug got luckily fixed for some unknown reason (I didn’t change anything)
20:39 RealBadAngel  Wuzzy, what tutorial?
20:40 Wuzzy         https://forum.minetest.net/viewtopic.php?f=15&t=10192
20:40 Wuzzy         This one.
20:41 Wuzzy         You know, the same thing for which all those folks voted in <https://forum.minetest.net/viewtopic.php?f=15&t=9116> :P
20:42 paramat       should be an optional subgame i feel, so not bundled with MT
20:43 Wuzzy         That makes the tutorial useless
20:44 paramat       no it doesn't, those who need it can use it, most players don't want or need a tutorial ;)
20:44 Wuzzy         a tutorial which is not included with a game by default is useless. As soon as the player figured out how to install this tutorial, the player most likely already knows the basics (perhaps from the wiki, lol)
20:46 Wuzzy         Can you proof that this is true specifically for Minetest?
20:48 Arguably_Sane hi
20:48 Arguably_Sane does anyone here have any idea why running update_liquids() during mapgen would not cause liquids to start flowing in newly generated chunks?
20:48 paramat       no :) same way you can't prove most want it. would you want a world bundled with it?
20:49 Wuzzy         ugh, a world is already bundled with the subgame
20:49 Wuzzy         Did you read the thread?
20:50 Wuzzy         the naked subgame is pretty useless
20:50 paramat       ah 1MB, not to bad actually
20:51 paramat       i do think it's cool, but not sure the devs will allow it to be bundled
20:52 paramat       beautiful world too
20:54 Wuzzy         well, the thing is, just “bundling” has some issues. the most notably one is that the world cannot be easily reset, you have one world which is autosaved always
20:54 Wuzzy         the only way to reset would be to delete world and reinstall lol
20:55 Wuzzy         I have started a discussion about that particular problem here: https://forum.minetest.net/viewtopic.php?f=5&t=10312
20:56 Wuzzy         TL;DR: I proposed to make some engine updates, specifically the main menu (but I guess now its a bit late for THAT).
20:56 paramat       ah i see it has to be integrated properly
20:57 Wuzzy         Other solution would be to place nodes in subgame itself, it was suggested several time, but I didn’t like it because it makes code messy and also wouldn’t be really re-usable
21:01 paramat       okay i'm neutral about this actually, bt it's too late for 0.4.11, good work
21:01 paramat       *but
21:02 Wuzzy         well, celeron55 seemed to notice it
21:02 Wuzzy         he reported a bug on it
21:02 Wuzzy         I guess this bug is now fixed for an unknown reason lol
21:05 celeron55     i think it could be included as long as there is no bug and as long as the world is optimized to be smaller
21:05 celeron55     there can't be 1MB of stuff in it
21:06 celeron55     in addition to actually being a good tutorial, it's actually an interesting experience even if you know it already and presents the idea of sharing worlds to people who haven't thought about it
21:07 celeron55     altough, i don't think there's any reason to try to include it in the next release
21:07 gravgun       Wuzzy, if celeron55 want a small world, it may be possible to use a post-worldgen Lua script which could generate the required structures/import them
21:07 Wuzzy         yeah i know, feature freeze :/
21:08 Wuzzy         so if I understand right: The files must be smaller than 1 MiB, right?
21:08 celeron55     let's schedule this for 0.4.12
21:08 Wuzzy         ok I understand
21:08 celeron55     the main menu stuff is quite important
21:08 celeron55     for this particular thing
21:09 Wuzzy         I agree.
21:09 celeron55     like... it needs some kind of bare customization of the menu
21:09 celeron55     not much but something
21:09 Wuzzy         As suggested here?: https://forum.minetest.net/viewtopic.php?f=5&t=10312
21:09 Wuzzy         lol
21:10 celeron55     i don't think it should be handled specially
21:10 Wuzzy         okay ... how else?
21:10 celeron55     every game should have the option to do what the tutorial needs
21:10 Wuzzy         a generic reset-feature maybe?
21:12 celeron55     it would be ideal if subgames in MT had the option to work like ace of spades instead of a persistent open world if they want to
21:12 paramat       sorry for my initial attitude and ignorance Wuzzy
21:12 Wuzzy         lol
21:12 Wuzzy         okay, I don’t know how ace of spades works :(
21:13 Wuzzy         I heard of the concept, a little bit but thats all
21:13 celeron55     there are others like it too
21:14 celeron55     but like, it's a shooter where you select a (usually) pre-made level (world, whateever) and the game uses an instance of it for a match where teams build and shoot
21:14 Wuzzy         is it possible to save a AoS game?
21:14 celeron55     and once a timer runs out or they kill enough of the opposing side or destroy some objective or whatever, you get the end screen and the world disappears
21:15 celeron55     like, a basic multiplayer match
21:15 celeron55     MT can't do that currently well at all while it probably should, it doesn't seem that complicated
21:15 Wuzzy         yea
21:16 celeron55     it's not possible to save a multiplayer game like that
21:16 celeron55     think of team fortress or counter strike or so
21:16 Wuzzy         ah
21:16 celeron55     they don't do it and don't need to
21:16 Wuzzy         makes sense
21:16 Wuzzy         but the tutorial MAY need a “manual” saving function
21:16 Wuzzy         or it may be useful in a generic sense
21:17 celeron55     i'm not sure if this is wanted, but basically the tutorial is something between that and what MT currently does hard-coded
21:17 Wuzzy         currently, MT autosaves like everything
21:17 celeron55     in the same category there are limited-sized worlds and pre-generated worlds with zero playtime generation
21:18 celeron55     i'm not sure if anyone has really even thought what might be created if these options would exist
21:18 Wuzzy         sounds like a use case for adventure maps
21:19 Wuzzy         in adventure maps you dont really want or need procedurally generated maps
21:19 Wuzzy         so would be a more efficient??? I guess
21:20 celeron55     yeah, basically it should be possible to include worlds in a mod or a game (maybe a game, not a mod), where new worlds can be instantiated leaving the original copy intact
21:20 Wuzzy         right, this is excactly what this tutorial would need
21:21 celeron55     the tutorial is basically an adventure or challenge map
21:21 celeron55     it's a thing that already exists but is badly supported
21:21 Wuzzy         right, tutorial world is an adventure map without adventure XD
21:21 paramat       Arguably_Sane, that's normal for MT, some liquid nodes just don't flow despite using update_liquids
21:22 celeron55     i mean, the way they currently are supported is that you can bundle a game in a world
21:22 celeron55     but the thing that is needed is bundling a world in a game
21:23 Wuzzy         huh? you can bundle a subgame in a world? thats interesting. Does anyone use this feature so far?
21:23 celeron55     i don't know, but yes you can
21:23 Wuzzy         but true, the reverse is what’s needed
21:24 celeron55     i'm hopeful that someone is interested enough in utilizing it that he will implement it 8)
21:25 celeron55     not sure how to go about it, maybe even main menu lua code included in a game? that would be the most flexible way
21:25 celeron55     it might make sense to bundle it with other main menu redesign
21:26 celeron55     iasda
21:26 celeron55     -iasda
21:26 Wuzzy         main menu bundled in subgame sounds scary to me.
21:26 celeron55     i mean, part of it
21:26 Wuzzy         one faulty subgame could crash minetest :/
21:27 Wuzzy         or I don’t understand lol
21:27 celeron55     have you ever read how i would want the main menu to be structured? 8)
21:27 gravgun       If we remember to do those f'kin lua type checks, a faulty subgame won't crash MT
21:28 Wuzzy         :)
21:28 celeron55     but, like
21:28 celeron55     there should be a screen where the user chooses to either connect to a server, or pick an installed game to be run locally
21:29 celeron55     and once the user selects a game, the game could show the further menu
21:29 celeron55     it doesn't have to be more complicated than that
21:29 celeron55     the current menu sucks
21:29 Wuzzy         ah
21:29 gravgun       ^
21:30 Wuzzy         I remember you briefly mentioned it in the roadmap
21:30 celeron55     it's in the roadmap, yes
21:34 celeron55     the roadmap contains too many things really
21:34 celeron55     but i don't have a good way to split it into chronological development steps
21:34 celeron55     the amount of stuff in it in one chunk is probably quite demotivating
21:46 Wuzzy         I am updating German translation right now. What does “Touch free target” mean (used for Android build)
21:56 PilzAdam      why is this a "typo"? https://github.com/minetest/minetest/commit/06207ac5508aff63573c91e758073ba1c8960a41
21:56 PilzAdam      all other functions in this table have commas too
21:58 MinerDad      PilzAdam, which of the comments in https://github.com/minetest/minetest/pull/1963 are "uninformative" (besides serialize and unserialize)?
22:00 PilzAdam      almost all comments on the members of ItemDefinition
22:01 PilzAdam      the comment on createItemDefManager()
22:02 MinerDad      Most of the comments on ItemDefinition were already there. I just converted them to Doxygen style.
22:02 PilzAdam      all comments on functions in IItemDefManager that have less than 4 lines
22:03 PilzAdam      they shouldn't be there
22:03 PilzAdam      and by turning them into doxygen you make it even worse
22:03 MinerDad      So you'd rather the comments be removed?
22:04 MinerDad      I, with little knowledge of the system, found most of those existing comments useful.
22:05 PilzAdam      these comments are pretty redundant
22:05 PilzAdam      "Check if item is known" above "isKnown" is not really helpful
22:06 MinerDad      But it isn't obvious that you can pass in an alias or just the name.  The comment tells you that.
22:14 MinerDad      Also, the two classes "IItemDefManager" and "IWritableItemDefManager" are interfaces with one implementer (CItemDefManager). Should I make the interface comments reflect the implementor or put the implementation comments in the .cpp?
22:17 MinerDad      And is CItemDefManager supposed to be singleton because it appears to act that way?
22:27 hmmmm         well definitely not, there are two CItemDefManagers existing in a singleplayer game, one for client and one for server
22:29 hmmmm         you'll notice that a lot of the celeron-made data structures have an interface and an implementation; this isn't necessarily because the implementation needs to be abstract, but it's for the purpose of minimizing dependencies in the header files since it allows private members to remain defined in a single compilation unit
22:29 hmmmm         how much this helps in practice is debatable...
22:31 MinerDad      Is the use for a read-only interface and a writable interface meant for thread-safeness?
22:31 hmmmm         yes
22:47 celeron55     actually a better thing to do for performance would be to have the private members in an opaque private struct 8) i do that sometimes too but i think zero times in minetest
22:48 celeron55     but it's more complicated for not much benefit
22:49 celeron55     oh also, actually that's sucky when you want to have such private methods too
22:49 celeron55     forget that, it's almost never useful
22:50 hmmmm         how's buildat
22:51 celeron55     frozen, too much work
22:51 celeron55     also, that doxygen patch is pretty much exactly what we do not want
22:51 hmmmm         huh?  why
22:51 celeron55     it
22:52 hmmmm         #1963
22:52 ShadowBot     https://github.com/minetest/minetest/issues/1963 -- Add doxygen comments to itemdef.h by MinerDad7
22:52 hmmmm         ahh
22:52 hmmmm         I thought you were talking about the one that enabled the doxygen compilation or whatever
22:53 celeron55     it's still full of comments of very questionable usefulness; also, it is being made by a person who doesn't do core development and probably doesn't have much insight at all to the thing in hand and is just writing what anyone would guess from the names in the code
22:53 hmmmm         you're right
22:53 celeron55     just stop these
22:53 hmmmm         i was in the process of typing the second reason
22:53 celeron55     comments are only allowed from persons who have actual insight into what the code does
22:55 celeron55     of course it could be that MinerDad7 analyzed the code to a required extent and this is not to be taken personally
22:55 celeron55     but this is how it looks like
22:55 hmmmm         well
22:55 hmmmm         he showed up one day and pretty much said, "how can I help minetest" and somebody told him to document things
22:56 hmmmm         that was a mistake
22:56 celeron55     that somebody should say this is the wrong way to do it
22:57 hmmmm         that somebody simply didn't realize the ramnifications of putting somebody who has 0 experience with the codebase in charge of writing documentation for it
22:57 hmmmm         the truth is there's no way you can contribute to minetest without getting neck deep with the code and making no visible progress for a longish period of time
22:58 hmmmm         even if we had amazing documentation
22:58 hmmmm         this is the way all mid-to-large-size projects are
22:59 hmmmm         maybe if he had a focused mission to improve some piece of minetest
23:00 celeron55     i commented the pull request with this main point
23:01 celeron55     also, all people who have been working with parts of minetest and don't have anything in particular tot do right now are thusly recommended to go comment key parts of the code they know well
23:02 celeron55     don't comment members and methods but comment structs and classes
23:02 celeron55     usually they need explanation the most
23:02 celeron55     also files need some introductions
23:02 celeron55     header files
23:04 celeron55     it's more important to see a description of a class and a compact overview of every method and member in it than seeing 25% of methods and members with comments
23:04 hmmmm         i feel like the Doxygen Way(tm) is to make comments for everything
23:04 hmmmm         it certainly encourages that
23:05 celeron55     that's why we need to actively discourage it
23:05 celeron55     don't be afraid
23:05 celeron55     people will learn 8)
23:05 hmmmm         it might simply be better suited toward documenting a library's external interfaces
23:05 celeron55     yes, that's what it's mainly designed for
23:05 celeron55     but it does work as a handy code browser
23:14 hmmmm         was just looking at random buildat source
23:14 hmmmm         how wide are your tabs in vi?
23:15 MinerDad      You might want to modify http://dev.minetest.net/Engine_documentation then.  It seems to indicate that you want everything documented in Doxygen format and that anybody can help.
23:15 hmmmm         heh.. completely false
23:15 hmmmm         the people who modify the mintest wiki are usually not us
23:15 MinerDad      ah, didn't know that
23:16 hmmmm         MarkTraceur
23:16 hmmmm         wasn't he some guy who got banned from #minetest?
23:17 MinerDad      I was planning on adding documentation as I figured out the code base.  Kind of hit two birds with one stone.
23:17 hmmmm         here's my advice for open source contributing:
23:17 hmmmm         scratch your itch
23:17 MinerDad      Documentation is actually one of my itches :)
23:18 MinerDad      I would like to fix https://github.com/minetest/minetest/issues/1374 but I'm still figuring out how things work.
23:18 celeron55     hmmmm: my tabs are always 4 spaces wide, and buildat uses a total code width of 85 columns
23:21 hmmmm         you seem to waste a lot of horizontal space by defining functions in class defintions and putting portions of code in their own blocks
23:21 hmmmm         that right off the bat is 2 levels of indentation wasted
23:22 celeron55     MinerDad: if you indeed write documentation only after you have studied the code for more than just a few minutes, then it's essentially ok as long as you only comment where necessary (i.e. what you learned differs from what you immediately see from a line of code) - could you try following my quick rule of only documenting classes and structs and files, and as rarely as possible other things?
23:24 MinerDad      Yes, I can just put documents around classes and structs.
23:26 celeron55     try doing that on one file and hilight me here, i'll make sure to check it and comment further if it still seems off
23:26 MinerDad      ok
23:26 MinerDad      I'll update my existing pull request.
23:26 celeron55     make sure to be very accurate, because misleading comments are worse than no comments
23:27 MinerDad      agreed
23:27 MinerDad      Do you want me to delete the comments that already exist in itemdef.h?
23:28 celeron55     the existing doxygen comments in there were made by hmmmm if i am not mistaken
23:28 hmmmm         no that's nodedef.h
23:28 celeron55     that class is rather obscure and probably needs them
23:28 hmmmm         I am going to explode that class btw
23:28 celeron55     oh, well then let me see...
23:29 hmmmm         really don't like NodeResolver.  dumping it in favor of a NodeDefManager-initiated callback of object-specific resolveNodeNames() routines
23:29 hmmmm         it's going back in the original direction except much more organized.
23:30 celeron55     frankly, many of the existing comments in itemdef.h are useless and they tend to have been written by me... and kahrl; however, some of them are crucial for understanding the methods they are commenting
23:31 MinerDad      So, remove most and convert a few to doxygen?
23:31 celeron55     you could try removing the obviously redundant ones, it will be interesting to see how it looks then? i don't know :P
23:32 hmmmm         how does <thing> prediction work currently?
23:32 hmmmm         what kinds of things are currently being predicted on the client?
23:33 MinerDad      What about doxygen groupings?  The current comments group some of the members as "Basic item properties", etc.
23:33 hmmmm         does the server simply respond with "nope, it turned out your move wasn't valid after all." and then the client resets?
23:33 celeron55     the existing grouping comments are very useful
23:33 celeron55     so if doxygen supports those, it would be very good
23:34 MinerDad      ok
23:35 celeron55     hmmmm: iirc the client predicts node placement/removal, object damage and player movement
23:36 celeron55     the handling of conflicts is not unified at all
23:36 hmmmm         maybe it should be unified
23:37 hmmmm         probably some sort of base class... IClientPredictor
23:37 celeron55     with nodes, the server compares what the client thinks happened and what actually happened and if there is a difference, it updates the client
23:37 celeron55     with object damage, the server always sends the end result
23:38 celeron55     with player movement the client has almost total control and the server rarely cares at all (it can teleport the player if everything goes bad)
23:38 celeron55     hmmmm: sounds bad
23:38 celeron55     they're all different
23:38 celeron55     and may have to be
23:39 celeron55     you could investigate but i'm not very hopeful
23:43 MinerDad      celeron55: In itemdef.h/.cpp, the real implementation is CItemDefManager but that isn't declared in the .h. Do you want me to add documentation to the .cpp or put it on the interface classes (IItemDefManager and IWritableItemDefManager)?
23:50 hmmmm         er
23:50 hmmmm         celeron, PM
23:50 hmmmm         you're away, screen detached