Time  Nick        Message
01:13 paramat     will merge #5961 #6005 in a few mins
01:13 ShadowBot   https://github.com/minetest/minetest/issues/5961 -- Mgv7 spawn search: Cope with extreme custom terrain and biome 'dust' by paramat
01:13 ShadowBot   https://github.com/minetest/minetest/issues/6005 -- Mgv7: Do not limit river generation if no floatlands by paramat
01:21 paramat     merging
01:28 paramat     complete, no more spawning waist-deep in snowblocks
06:19 nerzhul     hello :)
06:20 nerzhul     can someone take a look at #6010 & #6000 please ?
06:20 ShadowBot   https://github.com/minetest/minetest/issues/6010 -- Remove IrrlichtDevice unused pointer from ClientActiveObject class & childs by nerzhul
06:20 ShadowBot   https://github.com/minetest/minetest/issues/6000 -- C++11 cleanup on constructors by Dumbeldor
06:26 nerzhul     !tell Fixer Fixer, VanessaE i looked at the code for the mentionned crash. It can only occur if server send HudSetParam with wrong image value
06:26 ShadowBot   nerzhul: O.K.
06:29 nerzhul     !tell Fixer Fixer, VanessaE the problem is either param is wrong or value is wrong when setting this option. I can avoid the crash but not resolve the server bug :)
06:29 ShadowBot   nerzhul: O.K.
06:34 nerzhul     also can someone review #6013, it fixes #6011
06:34 ShadowBot   https://github.com/minetest/minetest/issues/6013 -- Verify HudSetParams input when hotbar textures are set by nerzhul
06:34 ShadowBot   https://github.com/minetest/minetest/issues/6011 -- Crash on connect to VE-S server
06:34 nerzhul     i added a missing control check
06:53 VanessaE    nerzhul: sounds good, though I haven't had my client actually *crash*, just the problem described by #6009
06:53 ShadowBot   https://github.com/minetest/minetest/issues/6009 -- Map stops loading, formspecs won't open, but all other server activity appears normal
06:54 nerzhul     it's strange some clients crash and some others not
06:55 nerzhul     maybe a corrupted cache ? strange
06:55 VanessaE    idk
06:55 nerzhul     but in all cases the crash fix should be avoided by checking hud input
06:55 VanessaE    right
07:06 VanessaE    I feel the need to point out that I'm running my server and client on the release commit now, and given the c++11 stuff and 0.5.x looming, I think it's best that I stay on the 0.4.x line, so I hope there will be backports of those ^^ fixes e.g. to 0.4.17
07:10 red-001     are you not able to get a c++11 compiler?
07:11 VanessaE    I probably could, but debian 8.8 doesn't come with it by default, and anyway 0.5.x is supposed to be the "break stuff" branch.  I'd much rather not break things.
08:12 nerzhul     celeron55_, can we remove this obsolete page ? http://dev.minetest.net/Dev_Log
08:13 celeron55_  yes
08:14 nerzhul     seems i can only blank it not remove :(
11:55 nerzhul     merging #6013 in ~10 mins, ty sfan5
11:55 ShadowBot   https://github.com/minetest/minetest/issues/6013 -- Verify HudSetParams input when hotbar textures are set by nerzhul
12:06 red-001     removed the links from the main page to dev log since nerzhul blanked it
12:10 nerzhul     red-001, ty
12:11 nerzhul     !tell Fixer can you pull latest master and reconnect to VE-S server, i fixed the crash, it will show you the HUD error now
12:11 ShadowBot   nerzhul: O.K.
12:11 Fixer       nerzhul: ty
12:11 nerzhul     oh you are there :p
12:24 Fixer       nerzhul: still crashes (4dc5985)
12:28 nerzhul     same backtrace ?
12:29 Fixer       nerzhul: no idea, need to compile debug one
12:30 nerzhul     here i added a check to ensure texture exists in cache before setting it to player object. Hud copy player object value and use it, then i don't see how can this be possible then
12:41 Fixer       will check backtrace in two mins
12:46 nerzhul     if it's same backtrace maybe i forgot one input or something alter the input value and i missed it. Previous commit is not lost as it ensure we transport a correct value on the network
12:49 Fixer       nerzhul: is this client-side only fix?
12:50 Fixer       got backtrace
12:52 Fixer       nerzhul: https://pastebin.com/raw/KgJzZy7a
12:54 nerzhul     erf
12:54 nerzhul     same problem it seems, are yoyu sure you pulled the change ?
12:54 nerzhul     can you drop your media cache and retry ?
12:55 nerzhul     extended debug output is better also, here i don't have local variables
13:00 Fixer       nerzhul: part of thread 1: https://pastebin.com/raw/3VMkXHav
13:03 Fixer       will try with cleared media
13:12 nerzhul     i don't see same things as before
13:12 nerzhul     oh no, it's same
13:13 nerzhul     i can only handle error by checking is texture is not nil
13:13 nerzhul     but the problem is why it's nil
13:18 nerzhul     can you add actionstream << "Hotbar image is : '" << hotbar_selected_image << "'" << std::endl;
13:18 nerzhul     just before video::ITexture *texture = tsrc->getTexture(hotbar_selected_image);
13:18 nerzhul     src/hud.cpp L 130
13:18 nerzhul     it will show you image it wanted to load
13:19 Fixer       nerzhul: will do, but first I'm trying with fresh media, no crash yet
13:19 nerzhul     maybe it was due to cache, maybe not :p
13:19 nerzhul     i push a little indent fix on this function
13:21 Fixer       nerzhul: still crash even on fresh media, will do your code
13:22 nerzhul     my code only permit to show on console the faulty file
13:23 Fixer       okay
13:30 Fixer       need few tries
13:35 Fixer       nerzhul: crash, ACTION[Main]: Hotbar image is : ''
13:35 Fixer       nerzhul: when it loaded without crash: Hotbar image is : 'gui_hotbar_selected.png'
13:36 nerzhul     okay
13:36 nerzhul     then image path is empty
13:36 nerzhul     it's the problem
13:36 nerzhul     now we need to know why it's empty
13:37 Fixer       so it sometimes works, sometimes - not
13:38 nerzhul     let me a second
13:38 nerzhul     use_hotbar_selected_image == true here
13:38 nerzhul     while we have
13:38 nerzhul     ,if (hotbar_selected_image != "")
13:38 nerzhul     use_hotbar_selected_image = tsrc->isKnownSourceImage(hotbar_selected_image);
13:38 nerzhul     it's true is isKnownSourceImage returns true before
13:39 nerzhul     oh i know !
13:39 nerzhul     erf a bool init was not good
13:39 nerzhul     (i think)
13:39 nerzhul     can you try this
13:39 nerzhul     L121 hud.h => bool use_hotbar_selected_image; => bool use_hotbar_selected_image = false;
13:40 nerzhul     i think it's just because it's not well initialized, i think i forgot it by error in one migration step
13:40 nerzhul     i read many times and reviewers too but one init failed
13:40 nerzhul     i push the fix to master as i'm sure it's the problem, bool init is not always true because it's "undef"
13:41 Fixer       will compile after you push
13:41 nerzhul     done
14:04 kilbith_    yet another warnings with gcc 7.1: https://hastebin.com/goxekojoco.txt
14:05 kilbith_    and: https://hastebin.com/tefedotaqe.txt
14:08 Fixer       nerzhul: no crash after 5 tries, I guess it is fixed
14:13 nerzhul     kilbith_, if i remember you already send that
14:13 nerzhul     can be intresting to have opened issue in the backlog
14:23 Krock       merging #5277 in 5 mins
14:23 ShadowBot   https://github.com/minetest/minetest/issues/5277 -- find_nodes_in_area: Extend maximal count to U32_MAX by SmallJoker
14:32 nerzhul     merging #6010 in ~10 mins
14:32 ShadowBot   https://github.com/minetest/minetest/issues/6010 -- Remove IrrlichtDevice unused pointer from ClientActiveObject class & childs by nerzhul
16:30 DS-minetest at game fencegate def, texture can only be a sting, right? in game_api.txt is only an example
16:33 sfan5       could also be a table
16:36 DS-minetest ok
18:01 kilbith     so I'm looking to hire a C++ developer, /msg me if interested
20:40 nerzhul     Krock, can you re-look at #6000 please ?
20:40 ShadowBot   https://github.com/minetest/minetest/issues/6000 -- C++11 cleanup on constructors by Dumbeldor
20:41 nerzhul     and i someone can take a look at #6008 too
20:41 ShadowBot   https://github.com/minetest/minetest/issues/6008 -- Fix render order of overlays by juhdanad
21:00 nerzhul     i will open a post about 0.5 news and breaking compat soon, i'm writing temp changelog inside it too
21:00 bigfoot547  +1
21:00 nerzhul     it permits to let users time to switch to stable 0.4 and do the break when we want
21:00 nerzhul     currently there isn't but some new PR are introducing that
21:06 rubenwardy  i don't really see the point of posting version news so early
21:06 nerzhul     version break compat
21:06 nerzhul     and prepare a changelog in dev wiki
21:06 rubenwardy  just post one for release candidates/freeze, and for the actual release
21:06 nerzhul     because i don't want to take hours to do it because nobody in dev teams seams to care...
21:16 nerzhul     i don't do the changelog, here is the communication about master branch state https://forum.minetest.net/viewtopic.php?f=18&t=17929
21:23 paramat     thanks for that forum post. it will be good to soon start breaking compatibility, for example the custom player collisionbox PR
21:24 nerzhul     yeah, just wait 1 or 2 weeks
21:25 juhdanad    Maybe artificial light bank can get in too?
21:25 nerzhul     juhdanad, it's a break ?
21:26 juhdanad    It has a new map block format.
21:27 nerzhul     but we have retrocompat for mapblock format, it's not possible with that case ?
21:27 nerzhul     map is special, personnaly i prefer to keep map compat as it's real data
21:28 nerzhul     break protocol and some API why not, but break data i'm not very fan, if we can keep some compat
21:28 paramat     yeah we have to support existing worlds
21:28 juhdanad    There is no retrocompat. If you try to load a world with old Minetest version, it writes that the map is too recent and download a fresh version of Minetest.
21:29 juhdanad    The same thing happened when my light bugfixes were merged, if you remember.
21:29 juhdanad    But yes, there should be a 'save copy for older version' option.
21:30 kilbith     the modding API should never, ever break
21:30 nerzhul     i mean upgrade okay, downgrade no
21:30 nerzhul     what i mean is to being unable to read old map :p
21:31 nerzhul     kilbith, i agree, i don't see why it should break it's currently quite good
21:31 juhdanad    Oh, so by break you mean bi-directional break.
21:31 nerzhul     yeah
21:32 nerzhul     wow minetest 0.4.16 release post was read 25% more than 0.4.15 on all the release time :p
21:33 juhdanad    Maybe the community is growing.
21:35 kilbith     no, the apogee of MT was in 2013-2014
21:35 kilbith     now it's declining, I rarely an amazing mod anymore
21:36 kilbith     err, *rarely see
21:37 nerzhul     there is many mods, it's difficult to create new concept, now the need is massive performance by including some common mods parts into core (npc api, maybe mapgen plugin API ?, pipeworks engine-like API ?)
21:37 nerzhul     merging #6000 in ~10 mins
21:37 ShadowBot   https://github.com/minetest/minetest/issues/6000 -- C++11 cleanup on constructors by Dumbeldor
21:41 paramat     #5981 is ready apart from small changes to docs
21:41 ShadowBot   https://github.com/minetest/minetest/issues/5981 -- Biomes/decorations: Make relative to 'water_level' setting by paramat
21:41 ShadowNinja kilbith: Google trends says there's been a decline since early 2015, but the website stats since 2015 are pretty flat.
21:41 Fixer       ShadowNinja: linux is also on decline (via google), sky is falling
21:42 kilbith     Minecraft, too, in on decline...
21:42 kilbith     *is
21:45 nerzhul     kilbith, yeah normal it's a old game too
21:48 Fixer       nerzhul: older than minetest
21:48 Fixer       8 years old
21:48 nerzhul     yeah but the game genre is on decline
21:48 Fixer       nerzhul: it is matured
21:49 Fixer       MC has established pretty good modding community, good and bad servers, prominent players, optimisations
21:49 paramat     so less popularity and interest, but it's quality and content that matter most and those continue to improve in MT
21:49 Fixer       yes
21:50 Fixer       but "Subgame Ultimate Question" still remains
21:50 Fixer       future of MTG
21:51 paramat     the MC community is massive, but also known to be less pleasant than MT's, maybe because of the mainstream popularity
21:51 Fixer       engine was improved substantially, now it is subgame turn too
21:52 Fixer       i like MT servers
21:52 nerzhul     mainstream & money
21:52 paramat     we're trying to get the focus off MTG, it is only one subgame and is understaffed, people need to make new and better ones
21:52 Fixer       sense of community
21:52 Fixer       not just wooden boxes here and there
21:52 Fixer       mt is also affected by this*
21:52 nerzhul     paramat, core too
21:53 nerzhul     these days there are not many reviews on PRs
21:53 Fixer       wuzzy already proposed to include Tutorial subgame... even that still not included
21:53 nerzhul     you comment generally, i review recent PR, Krock too, sfan5 looks are random small PR and that's all :(
21:54 Fixer       sofar will be back
21:54 Fixer       in some time
21:54 paramat     yes dev is very quiet recently, hoping it is just exams
21:55 nerzhul     Fixer, i hope yes, i don't know why
21:55 nerzhul     oh and there is Zeno` sometimes too :)
21:55 Fixer       nerzhul: just some IRL stuff, he will be back
21:55 nerzhul     paramat, i think it's that many coredevs are students
21:55 nerzhul     i know, but strangely his GH avatar disappears
21:55 paramat     yeah
21:56 nerzhul     and his contributions in june are near 0
21:56 Fixer       nerzhul: he was on irc recently, he is fine
21:56 nerzhul     i missed him erf
21:56 paramat     he blanked his avatar because someone photoshopped it in an insulting way
21:57 nerzhul     pfff stupid guys...
21:57 red-001     sofar has a bloody great internet connection
21:58 Fixer       dialup?
21:58 red-001     Fixer, I wasn't being sarcastic
21:58 red-001     he is now signed into that minetest server for days
22:00 ShadowNinja I've noticed new warnings with GCC 7.1.1, here are some of them: http://sprunge.us/AceK
22:00 nerzhul     ShadowNinja, kilbith send that today too
22:00 nerzhul     i don't why i don't see them on my gcc
22:00 kilbith     ShadowNinja: http://irc.minetest.ru/minetest-dev/2017-06-19#i_4979748
22:01 kilbith     http://irc.minetest.ru/minetest-dev/2017-06-17#i_4977656
22:01 sfan5       the updated windows buildbot also generated those warnings
22:01 nerzhul     i will push some warning fix reported by my GCC 7.1 which are not in your list
22:01 Fixer       have not seen those errors on my gcc 6.3 iirc
22:01 nerzhul     (one unused variable and one wrong order in constructor)
22:01 Fixer       and i'm using modified buildbot.sh on msys2/mingw
22:02 sfan5       with updated builbot i meant the gcc 7.1 travis now uses
22:04 nerzhul     push my warn fixes to master (not related to reported here)
22:04 ShadowNinja Looks like optimization affects those warnings, so you'll have to be building in release mode.
22:04 nerzhul     oh it's in release mode hmmm
22:06 nerzhul     oh i'm compiling with clang...
22:06 nerzhul     i forgot my clion was configured with clang
22:09 Zeno`       specified size 18446744073709551615 is very pretty to look at (http://numbermatics.com/n/18446744073709551615/)
22:09 Zeno`       mesmorising
22:10 nerzhul     sign vs unsigned
22:11 nerzhul     analyze_block has just a tiny buf
22:12 nerzhul     i propose to replace char spos[20];
22:12 nerzhul     snprintf(spos, 20, "(%2d,%2d,%2d), ", p.X, p.Y, p.Z);
22:12 nerzhul     with
22:12 nerzhul     char spos[26] = {};
22:12 nerzhul     snprintf(spos, sizeof(spos), "(%2d,%2d,%2d), ", p.X, p.Y, p.Z);
22:12 nerzhul     (note: = {} init with zero)
22:13 nerzhul     Zeno`, ShadowNinja sfan5 okay ?
22:14 Zeno`       why init with 0?
22:14 nerzhul     it's a C string, just init it permit to prevent reading non inited memory
22:14 nerzhul     (information disclosure)
22:15 ShadowNinja nerzhul: I think 23 is the max size.  You could use sizeof("(-32000,-32000,-32000)") instead (sizeof includes the NUL byte).
22:15 nerzhul     ShadowNinja, GCC said 25 is the maximum
22:16 Zeno`       it is (I counted the characters by hand)
22:16 ShadowNinja Oh, I missed the ", " at the end.
22:17 ShadowNinja So use 25 or the sizeof thing.
22:17 nerzhul     25 then
22:17 nerzhul     :)
22:17 nerzhul     char spos[25] = {};
22:17 nerzhul     snprintf(spos, sizeof(spos), "(%2d,%2d,%2d), ", p.X, p.Y, p.Z);
22:17 Zeno`       anyway I wouldn't have the = {} since the snprintf() is immediately after it, but *shrug*
22:17 ShadowNinja No need to zero it, snprintf adds a \0.
22:17 nerzhul     okay then
22:18 nerzhul     for the two other i don't find it
22:19 nerzhul     okay for pushing the snprintf buffer fix ?
22:22 Zeno`       yeah
22:22 nerzhul     done
22:23 nerzhul     i should go see you :)Ã
22:23 Zeno`       bye bye
22:33 Zeno`       The first warning is due to this: static SRP_Result H_ns(mpz_t result, SRP_HashAlgorithm alg, const unsigned char *n,
22:33 Zeno`       size_t len_n, const unsigned char *bytes, size_t len_bytes)
22:34 Zeno`       len_n + len_bytes is obviously 2*size_t
22:34 Zeno`       (potentially)
22:36 Zeno`       https://github.com/minetest/minetest/blob/master/src/util/srp.cpp#L423
23:06 paramat     game#1070 seems good for a merge
23:06 ShadowBot   https://github.com/minetest/minetest_game/issues/1070 -- Add placer's player name to tnt node metadata by jhcole
23:12 Hijiri      #5612 #5819
23:13 ShadowBot   https://github.com/minetest/minetest/issues/5612 -- Allow overriding tool capabilities through itemstack metadata by raymoo
23:13 ShadowBot   https://github.com/minetest/minetest/issues/5819 -- Fix default item callbacks to work with nil users by raymoo