Time  Nick         Message
02:20 hmmmm        so who actually has ops here
14:55 celeron55    hmmmm: completely based on the time of that happening: europeans
14:55 celeron55    8)
15:06 Garmine      Hi! I am very sorry for the dis/connect spam, I've fixed the issue.
16:44 Selat        What about this https://github.com/minetest/minetest/pull/1159 and this https://github.com/minetest/minetest/pull/1167?
17:54 ShadowNinja  celeron55: Well, the flooding that you mentioned happened (although unintentionally).  :-)  We should have a few more ops.
17:55 celeron55    my point was, that was at the time all europeans were sleeping, and because nobody reacted at all, probably ops are europeans
17:55 celeron55    not that i would have noticed even if i was awake, i don't have any activity indication on join/part/quit
17:57 celeron55    nobody seemed to be here anyway so whatever
18:21 ShadowNinja  celeron55: Yes, you should have ops from other parts of the world.
18:23 PilzAdam     or ops that dont sleep
18:25 ShadowNinja  That works too.  :-)
18:36 celeron55    i think those are called bots
19:13 Selat        I think in wiki or readme you should specify average time of ignoring pulls. It can be very useful for newcomers.
19:18 proller      3-12 months
19:46 ShadowNinja  celeron55: Hmmm, well ShadowBot's here, I haven't added slow-join/quit-flood protection though, yet...
20:39 ShadowNinja  celeron55: I added floppy connection banning to ShadowBot.
20:40 sapier       ~ tell xyz SIGBUS is a unaligned access on dereferenciating a calculated lua_Number pointer in marshall
20:40 ShadowBot    sapier: O.K.
20:41 sapier       to everyone else, never try to access unaligned memory with non char datatype on android ;-)
20:45 sapier       #1159 is fine #1167 ... I think it's fine too but I don't have time to test it myself now maybe someone else can confirm it's fine
20:45 ShadowBot    https://github.com/minetest/minetest/issues/1159 -- Pass arguments by reference by Selat
20:45 ShadowBot    https://github.com/minetest/minetest/issues/1167 -- Remove goto from Environment::removePlayer by Selat
20:48 ShadowNinja  sapier: 1167 seems like it should work, but I would do it slightly differently (about 2 lienes shorter and a bit simpler)
20:49 ShadowNinja  I'll test and tweak it.
20:49 sapier       even more short? :-)
20:50 ShadowNinja  sapier: Yes, keep the ++i in the for loop, remove the else clause, and use erase(i);
20:51 sapier       I'm carefull with ++i i++ as this is one of the locations I know I make mistakes so test it good ;-)
20:54 sapier       I wonder if there's a way to catch those unaligned memory accesses without getting a issue for each occurence
20:56 ShadowNinja  sapier: Like so: http://ix.io/b3E
20:58 ShadowNinja  That goto was really odd.  The loop would have removed any other players even without it.
20:58 kahrl        ShadowNinja, you can't do that
20:58 sapier       good question ... I just don't know :-) that's why I avoid iderators for doing things like that ... maybe someone knows if this is correct or not
20:59 kahrl        you'll use i after it has been destroyed
20:59 ShadowNinja  kahrl: Huh?  Where?
20:59 kahrl        you have to say i = m_players.erase(i) and skip the ++i afterwards
20:59 ShadowNinja  kahrl: Ah, OK.
21:00 sapier       that's the thing called "fun with iterators" ... I hate it ;-)
21:03 ShadowNinja  This should do it: http://ix.io/b3F
21:03 ShadowNinja  I'll push it in a few minutes.
21:04 kahrl        looks good
21:05 ShadowNinja  Hmmm: /home/owen/minetest/minetest-dev/src/script/cpp_api/s_base.cpp:73: ScriptApiBase::ScriptApiBase(): Assertion 'm_luastack' failed.
21:05 ShadowNinja  When running in valgrind ^.
21:05 ShadowNinja  (Race condition)
21:05 ShadowNinja  +?
21:08 kahrl        what other threads are accessing m_luastack at the same time?
21:09 sapier       none
21:09 sapier       it's not allowed at all
21:10 sapier       if someone tries this is a bug
21:11 sapier       but this assertion doesn't tell about a concurrent acces but about a failed creation ... no idea how that's possible
21:11 kahrl        perhaps http://stackoverflow.com/questions/20858911/lua-open-returns-null-using-luajit?
21:12 kahrl        or something similar
21:12 kahrl        out of memory?
21:13 sapier       yes but It's long time I've seen a out of memory in a non resource leak scenario
21:13 ShadowNinja  m_luastack = luaL_newstate();\nassert(m_luastack);
21:13 sapier       the assertion is correct there, so why do we fail to create a lua stack is the real question
21:14 ShadowNinja  Returns the new state, or NULL if there is a memory allocation error.
21:15 sapier       memory allocation errory should be as rare as water in desert nowadays ;)
21:15 ShadowNinja  That happens consistently though.
21:15 sapier       we're not allocating GB of memory but just a few KB
21:15 sapier       unless there's a bug in there ... is this limited to a special luajit version?
21:16 ShadowNinja  http://pastebin.ubuntu.com/7081500/
21:17 ShadowNinja  luajit -v --> LuaJIT 2.0.2 -- Copyright (C) 2005-2013 Mike Pall. http://luajit.org/
21:18 sapier       is there a chance we can drop luaJIT support? that piece of code always causes bugs
21:18 sapier       and those are very hard to track down
21:20 ShadowNinja  sapier: No.
21:20 VanessaE     no.  fucking.  way.
21:20 ShadowNinja  sapier: And you don't know that it's LuaJIT-only.
21:21 sapier       I haven't heared about any similar bug in lua ;-) ... but I was kidding anyway ... yet luajit causes a lot of trouble
21:22 sapier       obviously is way from beeing as marture as lua is
21:25 sapier       does @arch mean he's on arch linux?
21:27 ShadowNinja  sapier: Yep.
21:28 ShadowNinja  non-JIT Lua does work.
21:28 sapier       in this case I'll not even try to find this one without way more information ... in 9/10 cases it's related to some of archs bleeding edge versions
21:30 sapier       I don't know why people always think "new is better"
22:00 sapier       http://animalsmod.comuf.com/downloads/Minetest-debug.apk CURL support
22:16 ShadowNinja  sapier: You approve #1159?  I have it ready to push.
22:16 ShadowBot    https://github.com/minetest/minetest/issues/1159 -- Pass arguments by reference by Selat
22:19 sapier       #1159 is fine merge it
22:19 ShadowBot    https://github.com/minetest/minetest/issues/1159 -- Pass arguments by reference by Selat
22:27 VanessaE     So, status report:  sapier's latest build works on my tablet, finally! :D
22:29 ShadowNinja  sapier: #1141 Removes -Wall from debug, but not release.  And why?
22:29 ShadowBot    https://github.com/minetest/minetest/issues/1141 -- Fix warnings shown by clang by sapier
22:31 sapier       because -Wall warns about things like C99 usage too
22:32 sapier       you don't see the real errors in hundreds of bogus ones ... and we can't fix them
22:32 sapier       those -W things I added should be same as -Wall except of those bogus errors
22:36 ShadowNinja  sapier: Well, we aren't using C99...
22:38 sapier       we are
22:38 sapier       we use some datatypes specified in C99 quite often
22:44 ShadowNinja  sapier: I mean, we use C++.  If there are C99-only features being used they should be removed.  What datatypes?
22:46 sapier       IRC some long datatype not beeing specified until C99 or c++11
22:47 sapier       I think it's "long long" but not sure about it
22:47 sapier       you can find out quite quick if you reenable -Wall and compile
22:49 ShadowNinja  sapier: Seems like the only things using that are database*, things like SQLite that are in C99, and JSONCPP, which uses an #ifdef _MSVC_VER.
22:49 ShadowNinja  Oh, and serialize,*
22:50 sapier       ahh ok but that's as bad too as we can't modify sqlite
22:50 ShadowNinja  Also irrlichttypes.
22:50 sapier       too much ;-/
22:50 ShadowNinja  sapier: But SQLite3 is a C99 library.
22:50 ShadowNinja  sapier: No, I'll fix it...
22:51 sapier       ok I wont stop you of course :-)
22:52 ShadowNinja  Oh, irrlichttypes uses a #ifdef _MSC_VER too.
22:52 ShadowNinja  (MSVC doesn't have stdint.h?!)
22:53 sapier       maybe in recent versions but not the older ones ... don't ask me as of what version ;-)
22:53 sapier       but I guess even there it's called something like "ms_special_int.h"
22:55 ShadowNinja  Hmmm, getBlockAsInteger returns a long long, aka s64, shouldn't it be u64?
22:56 sapier       if it's called AsInteger I'd not think so
22:56 sapier       but of course that name could be wrong too I don't know
23:04 ShadowNinja  I've had bad experiences with signed numbers and binary operators...
23:04 sapier       you could ask celeron I guess this code is written by him
23:05 ShadowNinja  Possibly, or whoever added the alternative backends.
23:05 sapier       oh it's not in engine but database backends?
23:06 sapier       the backends aren't by celeron
23:06 ShadowNinja  Anyway, it's 64 bits and the position will only take up 46 bits of that, so it can't overflow to signed.
23:06 ShadowNinja  s/signed/negative/
23:07 ShadowNinja  sapier: All long longs removed, with the exception of SQLite3 and things in #ifdef _MSC_VER: http://ix.io/b3I
23:08 sapier       you could try to use different flags for subdirs?
23:08 ShadowNinja  sapier: SQLite3 should be compiled as C99.
23:09 ShadowNinja  The included SQLite3 is almost never used and far out of date though.
23:10 sapier       time to update but that's something different
23:11 RealBadAngel hi
23:12 RealBadAngel ive updated #1117 with all the latest additions (no water surface shaders there) and bugfixes. its ready to be merged and i would like to end this stage.
23:12 ShadowBot    https://github.com/minetest/minetest/issues/1117 -- Normal maps generation on the fly. by RealBadAngel
23:13 sapier       ok lets check and try to merge it till end of coming weekend
23:14 VanessaE     yes, get that damn thing merged already
23:20 RealBadAngel it would be good if folks will try this
23:36 ShadowNinja  sapier: Fine if I push those long long fixes?
23:36 sapier       yes ... hope it's really all
23:36 sapier       it's useless if it isn't :-(
23:41 RealBadAngel btw, whats the status of #1096 ?
23:41 ShadowBot    https://github.com/minetest/minetest/issues/1096 -- Add circuit simulator. by Selat
23:44 ShadowNinja  RealBadAngel: I believe it is simply too non-generic.  Just about the only similar mod is mesecons, and even mesecons might not use it because of issues with auto-rotation and the like.
23:44 ShadowNinja  s/similar mod/mad that could use it/
23:59 RealBadAngel even if mesecons will be the only mod that will use  it, still worthy imho
23:59 RealBadAngel should be lotsa faster than a mod