Minetest logo

IRC log for #minetest-dev, 2014-08-23

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:19 sapier left #minetest-dev
00:31 OldCoder joined #minetest-dev
01:28 Zeno` joined #minetest-dev
01:31 VanessaE sfan5: still awake?
03:28 zat1 joined #minetest-dev
03:32 Sokomine joined #minetest-dev
04:40 ^v joined #minetest-dev
04:56 Miner_48er joined #minetest-dev
05:48 Hunterz joined #minetest-dev
06:28 nore joined #minetest-dev
06:51 Krock joined #minetest-dev
07:05 sfan5 VanessaE: >awake >03:31 -> no, but I'm awake now
07:23 CraigyDavi` joined #minetest-dev
08:02 Hairy_Powder left #minetest-dev
08:23 Zeno` Is CONTENT_IGNORE available via Lua?
08:23 Zeno` for example: assert(minetest.get_content_id("blah") ~= minetest.CONTENT_IGNORE)   ??
08:24 Zeno` basically I want to check that the item string exists
08:26 nore Zeno`, did you try minetest.get_content_id("ignore")?
08:26 nore (you can also do assert(minetest.registered_items["blah"] ~= nil)
08:27 Zeno` I'll try both
08:27 Zeno` thanks
08:31 Zeno` I'll go with the 2nd method
08:37 PenguinDad joined #minetest-dev
08:47 Megaf joined #minetest-dev
08:47 Megaf you broke my server again
08:47 Megaf congratualations
08:47 Megaf https://github.com/minetest/minetes​t/issues/1556#issuecomment-53146903
08:53 Zeno` I didn't do that
08:54 Megaf someone did
08:54 Krock joined #minetest-dev
08:55 Megaf lucky me that I know the terrible quality of code and now a days I do a full backup before every git pull
08:56 Calinou joined #minetest-dev
08:56 Zeno` does this happen with no mods installed?
09:03 Megaf I restored my backup and it still crashing, now Im totally lost
09:08 Zeno` That code in httpfetch.cpp is a @$&*@2 mess
09:11 sapier joined #minetest-dev
09:16 Megaf server was running for several hours, then it crashed, then I backed it up and updated it, then it began to crash on start up
09:16 Megaf http://paste.debian.net/117058/
09:16 Zeno` Megaf, what happens if you add curl_parallel_limit = 8 to the .conf?
09:16 Megaf and now even restoring from the backup it will keep crashing
09:16 Zeno` or even set it to 1
09:17 Megaf Zeno`: its a server, that thing shouldnt matter anyway
09:17 Zeno` well, it's being called somehow
09:17 Zeno` I think
09:17 Zeno` just add it to the .conf to rule it out ;)
09:18 Zeno` if adding that doesn't fix the crash then at least people know to look elsewhere
09:21 sapier sorry but the stack doesn't help if you want us to track this down we need more information
09:22 Zeno` well, I'd like to rule it out anyway
09:23 Megaf sapier: so make a decent code that tells something when it crashes
09:23 Megaf insttead of useless crash info
09:24 Zeno` why g_httpfetch_thread is global is beyond me
09:24 Zeno` Megaf, you can do a backtrace using gdb
09:24 sapier gdb ./minetestserver
09:26 Megaf im installing minetest
09:26 Megaf I mean, gdb
09:28 Megaf gdb doesnt help much, does it?
09:28 Megaf http://paste.debian.net/117060/
09:28 sapier you need to do a debug build
09:29 * Megaf wishes it could shoot someone
09:29 sapier Calm down, you're using development branch things like that happen
09:30 Megaf because if I use the stable and have any problem you would say that it is old and its already fixed on the FUCLING development!
09:31 Megaf Zeno`: curl_parallel_limit = 8 doesnt help
09:31 Megaf I will try to compile without curl support
09:32 sapier dev is unstable and regressions are supposed to happen every now and then ... especially if you use a version where lots of things have been added the past 24 hours ;-)
09:40 Megaf sapier: compiled it without curl and in debug mode
09:40 Megaf terminate called after throwing an instance of 'ServerError'
09:40 Megaf what():  C++ exception
09:40 Megaf no useful info again
09:40 sapier did you run in gdb?
09:40 Megaf Program received signal SIGABRT, Aborted.
09:40 Megaf 0xb6cbff96 in ?? () from /lib/arm-linux-gnueabihf/libc.so.6
09:41 Megaf yes man, I did
09:41 sapier hmm disable luajit too
09:41 Megaf can you please tell me the cmake flag for that?
09:41 iqualfragile joined #minetest-dev
09:43 sapier -DDISABLE_LUAJIT=1 -DCMAKE_BUILD_TYPE="Debug" -DUSE_CURL=0 -DENABLE_CURL=0 .... just to be sure ;-)
09:43 sapier but I don't think curl is really involved
09:43 sapier what do you do to cause the bug?
09:44 Megaf thats the bug I reported
09:44 Megaf it just happens
09:45 sapier ugly :-(
09:46 Megaf yes it is
09:46 Megaf and theres another bug caused by the cirular saw
09:47 Megaf Im sure mods should be able to put the server down
09:47 Megaf shouldnt*
09:47 sapier how are we supposed to prevent this? mods are part of core
09:48 sapier that's the difference between mods and plugins ;-)
09:48 Megaf sapier: anyway, tried to run on my x86 machine just to run out ARCH issues, same thing
09:48 Megaf soon afeter another try with the flags you told me I will try a fresh build
09:48 sapier can you enter break __raise_exception on gdb prompt?
09:49 Megaf yes I can, after or before the run?
09:49 sapier before crash ;-)
09:50 sapier I'd buess this is a mod error, yet we need to improve this as the exception is quite useless
09:51 Megaf terminate called after throwing an instance of 'InvalidPositionException'
09:51 Megaf thats the error
09:51 sapier can you do bt and give full info?
09:52 Megaf sapier: that looks like corruped map...
09:53 Megaf I think my map got corruped on the last crash
09:56 sapier hmm very bad error message
09:56 sapier can you still post the backtrace for me to see where to add a better one?
09:58 Megaf http://paste.debian.net/117067/
10:01 Megaf celeron55: congrats, you got the worst piece of shit in the planet, thats the third map I get corruped by a fucking bug
10:01 Megaf how am I supposed to explain that to all players?
10:02 iqualfragile Megaf: do backups?
10:03 sapier calm down and help us finding what's causing it ... and stop complaining about bugs in a development version ;-)
10:05 Megaf I will try a fresh build and fres game with no mods
10:09 Megaf sapier: good news, map is not corruped
10:09 twoelk joined #minetest-dev
10:09 Megaf fresh git clone of minetest core and game works
10:10 sapier still you should do a regular backup and one prior updating to a new experimental version ;-)
10:10 kahrl joined #minetest-dev
10:12 twoelk Megaf: just out of interest, how large is your map by now?
10:12 PenguinDad Megaf: which backend do you use?
10:12 Megaf twoelk: 1.6G    ../worlds/world/map.sqlite
10:13 Megaf ^ PenguinDad
10:13 Megaf the last map I lost was more than 4 GB big
10:13 Megaf but I lost it more due to a backup software error
10:13 * Krock 's map is 700 MB and zero corruptions happened since he configured dis antivirus program well
10:16 kahrl Zeno`: uh, Megaf's crash had nothing to do with httpfetch
10:18 twoelk maybe you should burn a backup to dvd once in a while so you have a save version when the backup process corrupts a valid backup. (happened at my office, destroyed 2 years worth of backups)
10:20 Zeno` kahrl, I realised that about 5 minutes ago heh
10:21 Zeno` but I still maintain that ruling it out was a good idea
10:21 Zeno` *shrug*
10:22 Megaf twoelk: indeed, my last server was lost on the backup process and I noticed that only when rstoring a backup
10:31 sapier megaf for your bug I'd guess some mod tried to do things outside of map boundries
10:31 Megaf sapier: we suspect of travelnet
10:31 Megaf Im recompiling it in relese mode again and will run it without travelnet
10:32 iqualfragile how the f do you destroy your backups?
10:32 sapier still we need to find a way ShadowsNinjas lua exception handler not hiding all the relevant information ;)
10:32 iqualfragile just copy that shit on a btrf-formated partition on another hdd and take a snapshot afterwards
10:46 twoelk iqualfragile: destroying backups is really easy if you let it happen by an automated process and it becomes more fun if you are stupid enough to have all backups on the same physicall machine. Shit just tends to happen
10:51 Megaf sapier: now I have dozens of mods and ai need to try one by one to see whats causing the crash because minetest will just hide all info
10:52 nore Megaf, you make a dichotomy search for that :)
10:54 sapier Megaf: I'm bugging Shadowninja for months to fix this so don't tell me ;-P
10:54 Megaf ShadowNinja: common
10:56 sapier to be honest it's not easy to fix, yet we need a solution
11:01 Megaf well, it wanst like that before
11:01 Megaf someone changed something
11:01 Megaf so, just revert
11:03 sapier lol
11:03 sapier those changes aren't a single commit and have a lot of things on top of it
11:04 sapier btw without them it's not guaranteed you'd have seen an error too because error handling is broken for a long time
11:14 Megaf sapier: Zeno` kahrl twoelk VanessaE pipeworks is the mod whos causing the crash
11:15 Zeno` good find
11:16 Zeno` Megaf, also plantlife is being actively updated
11:16 Zeno` in the last few days I mean
11:19 nore Megaf: did you try updating it?
11:19 nore pipeworks had a bugfix recently
11:19 RealBadAngel Megaf, and you hanged c55 for that...
11:19 Megaf yep
11:19 Megaf someone must do something
11:20 Megaf minetest used to show more reveling error messages not long ago
11:20 Megaf if you didnt notice, its being a while since I last bothered you here
11:23 Megaf nore: yep, there when all this begun to happen
11:24 twoelk as this is software in development maybe we should have some more backup and recovery tools especially for server admins
11:25 sapier https://github.com/minetest/minetest/pull/1579 ShadowNinja this might help you finding and fixing the issue
11:26 sapier It's always a good thing to have a api fct to cause an error :-)
11:29 Megaf nore: I did a fresh git clone from pipeworks, same issue
11:30 ImQ009 joined #minetest-dev
11:30 Megaf I just find the following quite funny
11:30 Megaf DEBUG STACK FOR THREAD b28ff440:
11:30 Megaf #0  virtual void* CurlFetchThread::Thread()
11:30 Megaf what does curl have to do with pipeworks?
11:34 sapier nothing that's just a list of threads
11:40 RealBadAngel https://github.com/minetest/minetest/pull/1519/
11:40 RealBadAngel ive updated it with new version for serialize and make it compatible with older servers
11:41 RealBadAngel so its good to go imho
11:41 sapier what happens if a old client connects to this setver?
11:42 kahrl Megaf: "minetest used to show more reveling error messages not long ago" <-- I suppose before you installed luajit
11:43 Megaf I always used LuaJIT
11:43 Megaf on this server at least
11:44 RealBadAngel throw SerializationError("unsupported CF_SPECIAL_COUNT");
11:44 RealBadAngel this happens
11:44 RealBadAngel and go back to main menu
11:45 sapier on old client? really?
11:45 RealBadAngel yes
11:45 RealBadAngel this code been there before
11:46 sapier ok .. do you always send the long count or only when it's used?
11:46 RealBadAngel empty tile is also "used"
11:46 RealBadAngel so always max defined is being send
11:47 deltib joined #minetest-dev
11:47 RealBadAngel currently its 2, with change it will be 6
11:47 sapier then I suggest increasing min protocol version, no need to make a client download things if it's gonna bail out immediatly
11:48 sapier or, the better way of doing it, check client protocol version prior sending and send only two
11:48 sapier this might cause some broken faces yes but most things will most likely work
11:49 GhostDoge joined #minetest-dev
11:49 kahrl Megaf: btw, did you apply the patch that I suggested?
11:49 kahrl https://gist.github.com/kahrl/c81254227e11f09fd31f
11:51 Megaf im actually afraid of things like that
11:51 Zeno` ok, https://github.com/VanessaE​/plantlife_modpack/pull/21
11:51 Megaf afraid of messing my git thing up again
11:51 Zeno` ugh, sorry wrong @&$*@ window
11:52 Zeno` I think I need to sleep
11:52 kahrl you can always remove the patch before you do a git pull
11:52 kahrl just do git checkout src/script/common/c_internal.cpp
11:53 RealBadAngel sapier, how to check client version? an example somwhere?
11:55 Zeno` RBA, does ClientInterface::getProtocolVersion() do what you want?
11:56 Zeno` that's not really the client version though, so I guess not
11:57 GhostDoge Megaf: If minetest would use svn it would be even easier to mess up your local copy ;)
11:57 Megaf heh, I dont like SVN
11:57 RealBadAngel if i add compability for older clients i wonder if protocol version bump is needed at all
11:57 RealBadAngel version of serialization yes, its bumped but protocol?
11:58 Zeno` RemoteClient:
11:58 Zeno` u8 getMajor() { return m_version_major; }
11:58 Zeno` u8 getMinor() { return m_version_minor; }
11:58 Zeno` u8 getPatch() { return m_version_patch; }
11:58 Zeno` std::string getVersion() { return m_full_version; }
11:59 RealBadAngel isnt that engine version?
11:59 Zeno` don't hink so (from a glance)
12:00 Zeno` it's part of the RemoteClient class so if it is engine version I think it'd be in the wrong place
12:00 Zeno` better check though
12:02 RealBadAngel anyway such thing wasnt used before
12:03 RealBadAngel server doesnt care bout client's version at all
12:05 Krock joined #minetest-dev
12:09 RealBadAngel no, i was wrong it does care
12:10 RealBadAngel u16 protocol_version = m_clients.getProtocolVersion(peer_id);
12:10 RealBadAngel and that is passed to serialize
12:13 sapier net_proto_version or serialization_version are public you have direct access ... you have to at least bump serialization version, not sure about nev version too
12:14 sapier net version
12:15 RealBadAngel i think now it is ok
12:15 RealBadAngel when i get client's one i can send just 2 tiles then
12:16 sapier left #minetest-dev
12:17 Zeno` wonder what you said
12:23 RealBadAngel if (protocol_version < 24)
12:23 RealBadAngel special_count = 2;
12:23 RealBadAngel else special_count = CF_SPECIAL_COUNT;
12:24 RealBadAngel this way older client can still connect to new servers
12:25 Hunterz joined #minetest-dev
12:33 RealBadAngel ok, its updated with compability for older clients
13:07 deltib joined #minetest-dev
13:18 Jordach joined #minetest-dev
13:24 EvergreenTree joined #minetest-dev
13:30 casimir joined #minetest-dev
13:46 exio4 joined #minetest-dev
13:49 NakedFury joined #minetest-dev
13:53 RealBadAngel so, any more comments on https://github.com/minetest/minetest/pull/1519 ?
13:58 hmmmm joined #minetest-dev
14:11 Krock RealBadAngel, tested?
14:34 RealBadAngel yup, i can connect with new client to old servers
14:36 RealBadAngel hmmmm, https://github.com/minetest/minetest/pull/1519
14:39 Weedy joined #minetest-dev
14:46 VanessaE RealBadAngel: did you try it with e.g. Redcrab's old server?
14:46 VanessaE and try it multiple times
14:46 VanessaE (that server is sort of the acid test of old-new connecting ability ;) )
14:47 exio4 how is it still running?
14:47 VanessaE idk but somehow, it still is
14:48 kaeza joined #minetest-dev
15:13 eugd joined #minetest-dev
15:26 Hunterz joined #minetest-dev
15:34 Calinou joined #minetest-dev
15:35 sapier joined #minetest-dev
15:37 sapier for redcrabs server clear your media cache prior connecting ;-)
15:42 Calinou and after
15:42 sapier well if you care about those few kb yes
15:59 VanessaE seems like a "connect to redcrab" script would be easy to make :)
15:59 VanessaE mv cache cache.tmp ; minetest --go (redcrab's server address) ; rm -rf cache ; mv cache.tmp cache
15:59 VanessaE :)
16:00 VanessaE or something similar but idk if you can pass a username/pw on the command line
16:02 twoelk joined #minetest-dev
16:12 hmmmm RealBadAngel:  lgtm but I'd rather wait for celeron to comment on the update
16:27 RealBadAngel hmmmm,  i did basically what he wanted me to do
16:31 sapier RealBadAngel: Actually you did what you understood what he wanted you to do ... ;-) ... yes I'm picky now ... and I'd have understood same thing you did understand
16:33 Calinou https://github.com/celeron55/minetest/issues/373 → why does this 404?
16:33 Calinou it was about copy-paste
16:33 Calinou oh, old link
16:33 Calinou to celeron55/minetest
16:39 Tesseract joined #minetest-dev
16:43 zat joined #minetest-dev
17:13 domtron joined #minetest-dev
17:15 RealBadAngel celeron55, moving ver 6 to serialize/deserialize old would mean copying whole current code to "old" like version 5 is done, yes?
17:16 celeron55 maybe
17:17 celeron55 you don't have to do it, but try to be careful not to make serialization any less maintainable
17:18 RealBadAngel i made it that way so 6 and 7 can connect both ways
17:20 RealBadAngel but we already have one thing that was waiting to move oustide try catch
17:21 RealBadAngel waving property it is
17:22 RealBadAngel also i would like to add another one, light_source_alternative
17:22 RealBadAngel so maybe moving 6 to old would be the best solution
18:02 Megaf joined #minetest-dev
18:02 Megaf Hi all
18:02 Megaf I got a new bug for you
18:02 Megaf or at least a crash
18:02 kahrl yay :D
18:02 Megaf http://paste.debian.net/117138/
18:03 sapier hmm I'd say you can keep it but I don't think that's gonna help ;-)
18:03 kahrl "Killed"? That means something sent a SIGKILL to the minetest server
18:03 sapier why do you kill it?
18:03 Megaf not me
18:03 sapier oom?
18:03 Zefram_Fysh oom killer?
18:03 Megaf my theory is, there was a memory leak
18:04 sapier do you have valgrind?
18:04 VanessaE tried to access -- from too far means the server doesn's know where the player really is
18:04 Zefram_Fysh dmesg will tell you if it was an oom
18:04 VanessaE the server and client disagree on where the player's coords are
18:04 sapier server is right.
18:04 Megaf [3079307.408879] Out of memory: Kill process 29019 (minetestserver) score 949 or sacrifice child
18:05 sapier yes looks like a leak
18:05 Megaf see? It had to kill minetestserver, to save its child
18:05 zat Megaf: that is super dramatic
18:06 sapier oom has some heuristics what to kill usually some high memory process is killed ... but it's more complex
18:06 Megaf http://paste.debian.net/117139/
18:06 Megaf dmesg
18:08 sapier hmm if I read correct it's "only" about 510 mb
18:11 Megaf after a fresh boot of minetest server sapier
18:11 Megaf http://paste.debian.net/117140/
18:12 Megaf sapier: I got plenty of mem to handle something using “only” 510 MB
18:12 Megaf anyway, 510 MB it’s already a lot for something like minetest server
18:12 sapier true ... running valgrind to have a look wont harm
18:13 celeron55 are those numbers pages?
18:13 celeron55 509662 4kB pages would be 2GB
18:14 sapier hmm 4k? ok then I didn't read correct
18:15 celeron55 it would make sense if Megaf has 2GB of RAM
18:15 celeron55 otherwise that doesn't make much sense
18:15 Megaf yep, almost 2 GB of RAM there
18:15 VanessaE maybe he's still trying to use that arduino or bananapi or whatever it was :)
18:16 Megaf and that server runs only minetest server and lighttpd+php gci, and ssh-server
18:17 VanessaE did you update some mod(s) recently?
18:17 celeron55 it
18:18 celeron55 *it's probably a mod because the MT server has very hard time reaching 2GB in any kind of light use
18:19 celeron55 hmm wait, it's saying "1048064 pages of RAM" in there
18:19 celeron55 that means they are 2kB pages
18:19 Megaf VanessaE: all mods
18:20 Megaf VanessaE: and pipeworks will not allow my server to start
18:20 VanessaE wfm.
18:20 Megaf I called you a couple of hours ago I think
18:21 VanessaE called me?
18:21 Megaf nevermind
18:21 VanessaE O_o
18:22 sapier VanessaE: looks like somehow generation of a out of bounds block is triggered by a mod. Seems like pipeworks causes it
18:22 VanessaE sapier: how is that even possible?
18:22 VanessaE pipeworks has no lvm's or any other such code
18:22 VanessaE techic can do that though
18:22 sapier getnode?
18:22 VanessaE get_node() doesn't cause generation events.
18:23 Megaf VanessaE: I get this when I try to start my server with pipeworks enabled, http://paste.debian.net/117060/
18:23 VanessaE it'll return nil or ignore (depending on which call you use) if the block is not loaded.
18:23 VanessaE Megaf: no clue why that is without a backtrace from the engine.
18:24 Megaf the backtrace doesnt help at all
18:24 Megaf I compiled the server with debug flag and backtraced it
18:24 VanessaE Megaf: I don't mean a gdp backtrace, I mean a Lua crash backtrace
18:24 Megaf minetest will hide all info
18:24 VanessaE gdb*
18:25 sapier VanessaE: you won't get a backtrace as an exception is thrown within c++ code resulting in ShadowNinja's lua exceptionhandler catching it effectively resulting in "C++ exception occured" ;-)
18:25 VanessaE minetest does not "hide" the crashdump, it's just failing to show it due to a bug in whatever routine is handling the actual crash.  there's a difference.
18:25 VanessaE sapier: "In addition, we got a 404 while trying to show you the custom 404 page."
18:25 VanessaE :)
18:25 PenguinDad Megaf: did you try https://github.com/minetest/minetest/pull/1580
18:26 Megaf PenguinDad: I will as soon as it goes to the master branch
18:26 VanessaE Megaf: it won't go to master unless someone tests it.
18:27 kahrl Megaf: please don't complain about minetest's error message if you refuse to apply patches that fix said messages
18:27 kahrl messages*
18:27 sapier kahrl nice teamwork ;-)
18:27 Megaf ok sir, can you please help me to apply that patch?
18:28 VanessaE sapier: be that as it may, until he can provide a Lua bt, I and nore can't help him fix the crash :)
18:28 kahrl Megaf: wget https://github.com/minetes​t/minetest/pull/1580.patch
18:28 kahrl Megaf: git apply 1580.patch
18:28 VanessaE he can blame pipeworks all he wants but for all I know it could be mesecons (the two mods intermesh together nicely)
18:28 sapier VanessaE: I was talking about 1580, based uppon 1579 ;-)
18:28 VanessaE sapier: as was I
18:29 sapier I just fixed that dummy commit message, no idea why I failed to squash it prior pushing
18:32 sapier kahrl are you sure there never will be thrown a LuaError exception?
18:32 kahrl LuaError is derived from std::exception, so it will be caught as well
18:33 sapier ahh ... ok guess that will do too
18:35 kahrl anyway, good work on 1579 :)
18:35 Megaf kahrl: ok, patch applied and server recompiled, so now I just have to wait for it to crash, right?
18:35 kahrl Megaf: yep
18:35 Megaf so, should I try to enable pipeworks and make it crash on start up?
18:36 kahrl yeah, you could try that
18:37 sapier kahrl: I've been annoyed about those missing lua backtraces. Everytime I wanted to reproduce I stopped because of "nahh .. to much work" ;-)
18:37 Megaf Off Topic: BTW, VanessaE, homedecor is fantastic now a days, fantastic aditions to it. Thanks a million!
18:37 Megaf whoa!, it works! we got our old error messages!
18:37 kahrl yay :D
18:38 sapier case we merge this noone can tell "I don't know how to reproduce" ;-)
18:38 PenguinDad Megaf: throw it at us ;)
18:38 Megaf http://paste.debian.net/117142/
18:39 sapier kahrl as it is confirmed to work can you merge 1580?
18:39 kahrl sapier: sure
18:39 Megaf VanessaE: ^
18:39 sapier I don't wanna loose it like the chat overlay fix you did years ago ;-)
18:41 kahrl yeah, would be a shame :)
18:42 kahrl ok, it's in
18:43 sapier comments about 1579?
18:43 kahrl Megaf: before you do the next git pull, do a "git checkout src/script/common/c_internal.cpp" first
18:43 kahrl Megaf: in order to avoid any conflicts
18:44 Megaf Ok, doing that right now kahrl. Thanks
18:44 kahrl sapier: I would make the chat command require the "server" privilege
18:44 zat http://uppix.com/f-screensho​t_6971953f8e0dd001756f3.jpg
18:44 zat http://uppix.com/f-screensho​t_6982853f8e0e3001756f5.jpg
18:44 sapier hmm do you suggest anyone sane running debug build in production env AND having that mod installed?
18:45 sapier well it won't harm too I'm gonna add it
18:45 kahrl well, the mod will be running if you run the minimal game, even in release builds
18:45 kahrl of course you shouldn't use minimal but I'm not sure everyone knows that
18:45 sapier yes ... there a crash will happen
18:46 zat one more: http://uppix.com/f-screensho​t_7000253f8e15f001756f8.jpg
18:46 sapier guess I should catch this
18:47 VanessaE Megaf: looking....don't leave.
18:47 Megaf VanessaE: I’m all yours for the next 60 minutes
18:49 sapier the only lost blocks valgrind tells me about are shaders
18:49 VanessaE ok, this is the result of nore's rewrite of the item transport logic in pipeworks.
18:50 * Megaf looks at nore
18:50 sapier kahrl: better that way?
18:50 Megaf VanessaE: we better talk on the other channel now, right?
18:51 VanessaE yes
18:51 kahrl sapier: yes
18:51 kahrl I think it can be merged now
18:52 kahrl typo in line 22: modile
18:52 kahrl er, line 28 now
18:53 sapier fixed
18:54 sapier hopefully we'll never need it again ;-)
18:54 igor_ joined #minetest-dev
18:55 igor_ left #minetest-dev
19:01 Megaf just for notes sake, thats my git pull by the way
19:01 Megaf http://paste.debian.net/117143/
19:14 kahrl Megaf: can you paste the output of git status
19:15 kahrl (I'm confused why it would do a recursive merge and no fast forward)
19:15 Megaf kahrl: # Your branch is ahead of 'origin/master' by 24 commits.
19:15 kahrl hmm
19:16 kahrl did you apply other patches or merge other branches before?
19:16 Megaf well, there are two files that I added myself to that folder
19:16 Megaf kahrl: it says it is merging very often
19:16 Megaf when I git pull
19:20 kahrl I'm a bit surprised you haven't run into more trouble due to the same changes appearing in both minetest master and your master
19:20 kahrl but as long as it's working
19:21 kahrl I could tell you how to revert to minetest master, but then you'd lose those other changes
19:35 cg72 joined #minetest-dev
19:36 sfan5 joined #minetest-dev
19:48 domtron joined #minetest-dev
20:11 kahrl so, about weblate again
20:11 kahrl what should we do?
20:12 kahrl it appears we have the following options right now:
20:12 kahrl 1. figure out how to install it (seems somewhat involved); sfan5 said he would do the hosting
20:12 kahrl 2. request hosting from weblate
20:13 kahrl also the question is if we need to migrate some data from the old weblate server
20:14 cg72 treegen.cpp is full of crap!!!! to many for loops! is anyone aginst me cleaning it up??
20:14 sapier cleaning is fine breaking it not ;-)
20:15 cg72 sapier  i wont break it on accident lol
20:17 cg72 im testing it now, i removed 7 for loops , it only needs 4
20:23 PenguinDad cg72: I do see ~27 for loops in treegen.cpp
20:23 cg72 lol well for each tree i can do only 4 so 8 total lol
20:24 cg72 im not touching the l-system stuff
20:25 cg72 well 9 as i forgot the roots of the jungle tree lol
20:28 ^v joined #minetest-dev
20:32 pitriss Hi, please is possible to limit chat messages lenght in engine to some reasonable value?
20:33 VanessaE they already are.
20:33 VanessaE 512 chars or some such
20:33 pitriss Liberly land server was spammed by 9k+ message
20:33 VanessaE update your server then?
20:34 sapier amount of messages isn't limited
20:34 pitriss when it was introduced?
20:34 VanessaE sapier: you sure?  I could swear such a filter was added ages ago
20:34 pitriss wait i will bring here server owner
20:34 sapier no not sure but I think only length of message is limited
20:35 sapier but a mod kicking abusive users wouldn't be a big deal
20:35 sapier so I don't see this to be something to be fixed in core
20:35 sol_invictus joined #minetest-dev
20:36 pitriss VanessaE: sol_invictus is liberty land owner,.. he have more informations
20:37 VanessaE pitriss: ok.
20:39 VanessaE ShadowNinja: ^^^^^^  ?
20:39 VanessaE sol_invictus reports the server to be only a few days old.
20:40 VanessaE so either that cutoff was not added or it's broken
20:43 sapier well I can't remember about limiting number of messages
20:47 sapier and there ain't a limitation in code
20:48 sapier yet a mod doing this would be quite simple
20:50 pitriss http://pastebin.com/7ZWehiHZ
20:50 VanessaE there was such a proposal at one time (hence my pinging SN.  he was working on such)
20:51 Miner_48er joined #minetest-dev
20:51 sapier working? even a full featured mod doing this was hardly more then an hour of work
20:52 exio4 limit in the engine, iirc
20:52 VanessaE well it went along with some other patches he had in mind
20:52 sapier so if he did start he most likely would've completet it by now
20:52 sapier well we don't need this to be "in core"
20:53 VanessaE it belongs in core.
20:53 sapier why?
20:53 VanessaE every last server owner will need it.
20:53 VanessaE that's why.
20:53 sapier oh so a server owner running a private server for his two friends and himself will need it?
20:53 VanessaE yep ;0
20:53 VanessaE :)
20:53 sapier for sure ;-P
20:53 VanessaE but you know what I meant :P
20:54 sapier well if he does implement it fine but it's quite ugly to do this in core while it's almost no code to be done in lua
20:55 VanessaE http://ix.io/atr
20:55 VanessaE there it is.
20:55 VanessaE and some of the discussion that surrounded it:  http://irc.minetest.ru/minetest-dev/2014-02-20
20:55 sapier that doesn't limit the amount of messages vanessae
20:56 VanessaE sapier: it's not the amount that needed limited.
20:56 VanessaE it's the length.
20:56 VanessaE one message, 9K in length.
20:56 sapier he didn't write that
20:56 VanessaE [08-23 16:36] <pitriss> Hi, please is possible to limit chat messages lenght in engine to some reasonable value?
20:57 VanessaE "chat messages length"
20:57 VanessaE I read that as "length of a chat message"
20:57 VanessaE not the count of messages
20:57 VanessaE pitriss: clarify please?
20:58 sapier wasn't therere already some limitation? ... ok u16 ... 65k
20:58 VanessaE yeah
20:58 VanessaE 65k....  is ridiculous.
20:58 VanessaE no one's screen can hold that much
20:59 pitriss I really think there should be some limit in lenght.. and it should be handled by engine itself.. this is basic even in webcoding to limit user inputs to some reasonable value.
20:59 VanessaE even my big screens could only hold about 4k or so I guess (at a readable font size)
20:59 VanessaE pitriss: so limit the length of a single chat message?
20:59 sapier ok you guys discuss about a reasonable length first ;-) don't make it to short
20:59 pitriss yep thats good because burst of messages is not so easy to do..
21:00 sapier and don't forget clients can still send oversized messages so this will reduce load but not fix all issues
21:00 sapier burst messages are quite easy to do
21:01 sapier if someone can send a 9k message he can send 90 100byte messsages too
21:02 sapier but as I said fixing both of this issues in lua is possible by now too
21:04 pitriss eh this guy did that by typing IMO.. there are two typos in that...
21:04 sapier typing?
21:04 VanessaE well anyways, that ix.io link is the patch I was referring to.  dunno if it's still any good.
21:06 sapier discard huge messages without any action? strange behaviour who's supposed to ever find the bug why messages above 1024 don't work?
21:06 VanessaE sapier: see the chat log.  that's the same argument that was made at the time.  Why didn't anyone just take the code, make it right, and push it?
21:06 VanessaE doing this in a mod is stupid and hacky.
21:07 sapier because sometime ppl refuse to clean up the mess others do ;-) .. well shadowninja does usually complete at least those parts of his work he's interested in
21:07 sapier but looking at current pull list he's got a lot of 90% compelte things there
21:08 pitriss ok i will tell sol to try that patch.. i hope it will be still valid
21:08 VanessaE cg72: still wanna rework the tree code? :P
21:08 VanessaE pitriss: I already suggested via /msg that he throw together a mod to do the job for now
21:09 pitriss ok thank you:)
21:09 cg72 VanessaE yup its in progress why?
21:10 VanessaE cg72: see above.  you may run into .... friction....  when you are ready to get it merged :P
21:10 sapier pitriss: that patch will solve the one 9k message thing but if someone really wants to spam your server it wont really help
21:11 pitriss sapier: yeah i see..
21:11 pitriss but atleast something than such spam..
21:12 pitriss thanks for discussing it, I hope it will be fixed soon in some more system way:)
21:13 VanessaE so have the code auto-kick the user on receipt of a string that's super long?
21:13 cg72 VanessaE please clarify for me im stupid lol i dont understand what your trying to get across
21:13 VanessaE cut off and warn if it's like 256 or 512 bytes, kick the user off cold if it's say 2K or more.
21:13 pitriss yep sounds good for me
21:13 VanessaE cg72: getting patches merged into mainline is sometimes a difficult thing to do.
21:14 sapier give me about an hour and you'll have your mod
21:15 cg72 VanessaE if i can do it better it should be a no brainer for the devs, the tree code is all first commit ans shit!! ( minus the l-sys trees)  its slow and inefficient
21:15 cg72 shame on someone for wirting that code :)
21:29 Sokomine vanessae: isn't that implemented yet? clients tended to crash a lot when sending long messages in chat. which is sometimes annoying. of course not as often as it used to happen...but from time to time, i still encounter that bug
21:30 Sokomine cg72: perhaps it was coded as a temporal soulution at the time
21:31 cg72 :) Sokomine it probly was a cheap n quick code but its time to im prove it, as it seems noone ever looked at it before, hell it declares the same thing like 3 times in a row lol
21:31 Sokomine cg72: good luck improving it!
21:32 cg72 Sokomine im all dont with the same code in lua for the game so i just need to write it in c now
21:49 sapier https://gist.github.com/ano​nymous/e8d42545c4772ba9486f anyone to test ChatGuard?
22:01 iqualfragile_ joined #minetest-dev
22:09 casimir joined #minetest-dev
22:12 cg72 https://github.com/minetest/minetest_game/pull/308
22:12 cg72 all tested and works great
22:12 cg72 now for the c version of that lol
22:13 VanessaE don't bother.
22:13 VanessaE the C version is deprecated
22:13 VanessaE tree gen is supposed to be out-of-engine anyways, either done by Lua, or by a schematic, or by L-Systems
22:14 cg72 supose  to be but when will it really go away?
22:14 VanessaE idk
22:14 VanessaE now is as good a time as any
22:14 cg72 lol so its still in need of a goor rewrite rotf
22:15 cg72 well look at my cade  VanessaE and see if you like it better
22:15 cg72 https://github.com/minetest/minetest_game/pull/308  << BTW
22:17 VanessaE the code looks okay to me from a quick skim; did you by chance benchmark it?
22:19 cg72 like i know how to do that with pc stuf, if its not embedded i m lost for  a benchmark
22:19 VanessaE oh I just mean like run the code for a thousand passes or something, generating trees, and do the same with the old code
22:19 VanessaE time it with os.time() or some such
22:19 * cg72 is lost again
22:20 VanessaE or excuse me, os.clock()
22:20 VanessaE not os.time
22:21 * cg72 has no bloody idea what your telling her to do
22:21 VanessaE you do something like local t1 = os.clock() ; run your tree gen a thousand times ; print(os.clock() - t1)
22:21 VanessaE do that with your new code and do it again with the old code
22:21 cg72 ok i think
22:22 VanessaE I use the same basic idea when I benchmark a routine in e.g. plants_lib
22:22 sapier joined #minetest-dev
22:23 VanessaE record the current clock value, run the code for some heavy amount guaranteed to take forever, subtract the now-current clock value from the recording to get the elapsed time
22:23 VanessaE as long as you do the exact same "heavy amount guaranteed to take forever", e.g. generate 1000 trees, you will get a reasonably consistent report that you can say yes, my new code is faster, etc.
22:23 sapier use core.get_us_time() to do benchmarking
22:23 VanessaE sapier: ok, or use that
22:24 sapier it's way more precise
22:24 VanessaE but anything that can show that the new code is faster than the old code is good enough for this
22:24 sapier did anyone check the chatguard mod by now?
22:24 VanessaE (e.g. even if it's only precise to a tenth of a second)
22:24 sapier well I've seen time run backwards with os.clock() ;-)
22:30 VanessaE haha
22:31 VanessaE sapier: the other devs were right (whoever it was who said it, I forget).  it DOES look like it specifies time in the US. :)
22:31 VanessaE that should be renamed before it becomes too entrenched.
22:32 VanessaE core.get_clock_microseconds() or something
22:35 celeron55 get_µs_time of course! we can use unicode in 2014, right?!
22:36 cg72 celeron55 nope whats unicode?
22:37 cg72 lol
22:38 celeron55 "i just upgraded from baudot to ascii, what is this futuristic shit"
22:40 PenguinDad lol
22:41 Kray >unicode identifiers
22:41 cg72 sorry i cant get that upgrade to fit on my 5.25 floppy
22:44 cg72 sapier >>> attempt to call field 'get_us_time' (a nil value)
22:45 VanessaE what?  your build is more than 5 minutes old?!  dinosaur! :D
22:46 cg72 its 6 days old
22:46 VanessaE that's about 4 days too old for that particular call
22:46 cg72 now you tell me
23:05 Sokomine rapid progress :-)
23:06 Sokomine celeron55: any further ideas how to solve that inventory-switch-problem?
23:14 Sokomine celeron55: people continue to loose items through that bug. there just isn't any reliable way for my mod to catch it. perhaps all these...switches...where items are attached to the mouse pointer (and hidden in the slot something was taken out) could be _disallowed_ only if there's a allow_metadata_inventory_put function - and allowed elsewhere
23:14 Sokomine though that might still cause trouble with locked chests
23:16 sapier Sokomine imho we schouldn't add a workaround we have to be compatible later instead of doing a real fix. Problem is a real fix will be some work
23:16 VanessaE sapier: #minetest-->
23:16 VanessaE a question for you there.
23:26 cg72 vanessa you dont want to know the speed i get now lol
23:26 Sokomine hm. i really hope there'll be a fix
23:28 PenguinDad cg72: but I want to know that ;P
23:30 cg72 10,000 times old tree 3-3.5 sec, mine .18-.21 sec  jungletrees old 7.5, mine .30-.36
23:31 sapier nice
23:32 cg72 anyone remember how to sqush in terminal lol
23:32 cg72 squash**
23:33 PenguinDad you got calibre times now :D
23:33 sapier git rebase -i HEAD~<number>
23:33 cg72 i keep forgetting the -i lol
23:40 cg72 ok so now on th the c++ side of it
23:46 cg72 there now test and merge my code guys as its the best

| Channels | #minetest-dev index | Today | | Google Search | Plaintext