Time Nick Message 00:01 ShadowNinja Any big bugs to fix before releasing 0.4.8? I heard about the 5,2 local thing, is that major enough, and how will we fix it? 00:10 sapier 5,2 local? 00:15 sapier kahrl changeing environment usually doesn't help at all if you're on windows using msvc 00:16 sapier https://gist.github.com/kahrl/7619600 forget about this way of doing it unless you did check any single variant of setting locales (about 10 different ways to do it) 00:17 sapier I guess only sane solution for this bug is putting a "DONT COMPILE WITH GETTEXT AND MSVC" in readme 00:34 ShadowNinja Er, locale. 00:35 sapier I hate locales ;-) 00:40 hmmmm erm 00:40 hmmmm so what's the final verdict on that 00:40 sapier no I'm looking for a solution but I'm almost sure there won't be a solution to fix all glitches 00:41 sapier it's impossible to fix a issue if every single lib is messing around with it 00:43 sapier kahrls solution doesn't work as it's same code that was there before I added the fixes 00:55 hmmmm erm, so, I have two questions at this point: 00:55 hmmmm 1). how critical is this bug in practice? 00:56 hmmmm 2). who is going to add the dialog variant without the locale setting 00:56 hmmmm and I need to decide whether or not this is something that should be fixed before 0.4.8 00:56 sapier rewriting a new irrlicht dialog? 00:57 hmmmm copying the old irrlicht dialog and leaving out the setlocale call 00:57 sapier quick and dirty solution would be hide the button 00:57 hmmmm I think we ought to do that 00:57 hmmmm personally 00:57 sapier I can't even restore the locale no idea what irrlicht does there 00:58 hmmmm so last I've heard MSVC works, what about the mingw build? 00:58 sapier works with what? 00:58 hmmmm sapier, it triggers the setlocale call on dialog construction and every time an event is fired 00:59 hmmmm I mean building minetest with MSVC. I guess not with x64 + gettext, but most other configurations do work 00:59 sapier yes but even if I restore original locale after the dialog is destroyed there's still broken locale 00:59 hmmmm during the time the dialog exists 00:59 hmmmm right? 00:59 hmmmm it doesn't do anything magical to permanently break locale 01:00 sapier you're right, destructor is member of a derived class 01:01 sapier the only way to fix is either drop this feature or write a own dialog 01:01 hmmmm is modmanager usable without this dialog? 01:01 sapier yes 01:02 sapier it's local mode 01:02 sapier so online features are not affected 01:02 hmmmm alright then 01:02 hmmmm we disable the button for this release 01:03 hmmmm we fix it as best we can by implementing our own dialog and add all the other things we were meaning to, and then make the 0.4.9 christmas release 01:03 sapier when is release planed? 01:03 Fury tomorrow 01:03 hmmmm as soon as thexyz or sfan5 get around 01:03 Fury stop prolonging it or just rename to next numbering 01:04 sapier ok then disabling is only way to do so we won't find and TEST a solution for this one until tomorrow 01:04 sapier there are to many different variants to check if localitation is involved 01:04 hmmmm by the way are you aware of https://github.com/minetest/minetest/issues/1020 01:05 sapier this is exactly this issue 01:05 hmmmm oh oh oh 01:06 hmmmm that is the result of the issue 01:06 hmmmm yeah, not being able to properly parse formspec strings could do that 01:06 sapier error messages shown right after cancel are quite obvious 01:07 sapier shall I create a pull request to hide the button? 01:08 hmmmm just push it 01:16 sapier https://github.com/minetest/minetest/commit/16bd3683747e6ab86d6874b90d00a02285d68bdd pushed 01:33 ShadowNinja Alright, I don't know of any other major bugs. Has someone tested the win32 build xyz made? 01:33 ShadowNinja (On a actual Windows system, not just WINE) 03:51 hmmmm dunno 04:25 ShadowNinja I have a windows computer, but I tried running Minetest on a identical computer running Linux and it SIGSEGVd... 04:56 RealBadAngel i do have windows 04:57 RealBadAngel i made windows install to test the shaders 05:22 RealBadAngel btw makin jungle grass bigger is causing overlay glitches 05:23 RealBadAngel http://i.imgur.com/EfeGtxh.png 06:58 thexyz what? do you need a new build or something? 07:05 Miner_48er thexyz is there a version of irrlicht for cent OS? 07:05 hmmmm yes 07:05 thexyz no of course not 07:05 thexyz now stop asking silly questions 07:05 hmmmm we just want to make absolutely sure that it compiles fine on windows 07:06 thexyz okay 07:06 hmmmm so that's under MSVC, 32 bit or 64 bit? 07:06 hmmmm with or without gettext? 07:06 thexyz I've already said that FindCURL doesn't work for me 07:06 hmmmm do you do MingW builds too? 07:06 thexyz no I don't 07:06 thexyz 32 bit MSVC 07:06 thexyz is what I do 07:06 hmmmm I guess that's good enough 07:06 hmmmm gonna have to get the build system a bit more solid for 0.4.9 07:06 thexyz what do you mean? 07:06 hmmmm I am I mean, I'm going to try to set up a build environment for every variation 07:07 hmmmm so anyway 07:07 thexyz oh 07:07 Miner_48er yum install libirrlicht-dev ? 07:07 thexyz I thought about setting up buildbot 07:07 hmmmm isn't buildbot broken shit? 07:07 hmmmm lemme fix one small thing before I do something else 07:08 thexyz shall I build right now? 07:08 hmmmm no 07:08 hmmmm hold up 07:09 thexyz okay 07:10 hmmmm https://github.com/minetest/minetest/commit/068dd796f51e59953d1aff9f05510847e2bfe872 07:10 thexyz celeron55: we still have no source for android version, what shall we do? 07:14 hmmmm alright 07:14 hmmmm thexyz, build! 07:16 thexyz woof woof 07:16 hmmmm also you can update the website correct? 07:18 thexyz yes it seems 07:19 hmmmm who does the debian packaging again? 07:19 hmmmm i honestly forget 07:21 thexyz you should check the wiki 07:21 thexyz http://dev.minetest.net/Releasing_Minetest 07:22 hmmmm oh crap there's that new step 07:22 hmmmm well attempt #2 07:23 thexyz what? 07:24 hmmmm I needed to comment out -dev 07:24 thexyz okay 07:26 thexyz so in the end we still fucked up? 07:26 hmmmm yes 07:26 hmmmm wtf 07:26 hmmmm the version detection 07:26 VanessaE well, 0.4.8.1 time...... :P 07:26 thexyz well that sucks 07:27 hmmmm yeep 07:27 hmmmm where is ${VERSION_GITHASH} defined.. 07:28 hmmmm this is sapier's wacky version string detection 07:34 hmmmm alright 07:34 hmmmm got that under control 07:35 thexyz so build now? 07:35 hmmmm yup 07:39 thexyz https://github.com/minetest/minetest/releases 07:40 thexyz still building 07:41 thexyz how should we name .zip? my suggestion is minetest-0.4.8-win32-msvc.zip with minetest-0.4.8 directory inside 07:41 hmmmm that sounds about right 07:44 thexyz oh nice 07:44 thexyz now it has -dirty suffix because findcurl doesn't fucking work and I had to modify it to make it build 07:45 hmmmm I know, I hate that thing too 07:45 sfan5 hi everyone 07:45 hmmmm for now I guess you could disable it forcibly and recompile again 07:45 hmmmm cmake/Modules/GenerateVersion.cmake and remove the if (VERSION_GITHASH) part 07:45 hmmmm so it just sets the VERSION_GITHASH to the VERSION_STRING 07:45 thexyz I've removed --dirty 07:46 thexyz from git describe call 07:46 hmmmm ah that works too 07:55 sfan5 hmmmm: do you need my windows build? 07:55 thexyz sfan5: yes 07:55 thexyz add it to github release 07:55 hmmmm it's good to have along with xyz's 07:55 thexyz I've uploaded mine btw 07:56 hmmmm wow I like how git push --tags deletes all of upstream's branches 07:56 hmmmm pfffffffff 07:57 sfan5 github should fix it's z-sorting 07:58 sfan5 the progress bar is above the 'Uploading your release now...' text 07:58 sfan5 done 07:58 hmmmm alright that should take care of it 07:59 hmmmm we'll wait for pilzadam to do an ubuntu build for launchpad and find somebody to do Debian 07:59 sfan5 my build contains all DLLs so sound and leveldb works out of the box 07:59 hmmmm and then we'll start having fun pushing lots of features 08:00 thexyz hmm.. and I have no leveldb 08:00 thexyz sfan5: you don't have freetype6.dll 08:00 sfan5 woops.. the dlls only get into the .7z file 08:00 sfan5 I'll reupload 08:00 thexyz by the way, it may be a good idea to set up a clean windows installation and test stuff there 08:01 sfan5 I have a windows vm 08:01 thexyz hmmmm: why should we wait? 08:02 thexyz since the tag is already here it may be a good idea to write a forum post 08:03 hmmmm because we don't want the other guys to build from a dirty repo 08:03 sfan5 github won't accept my file :-( 08:03 sfan5 probably because the was one with the same name 08:03 hmmmm also it's way too late 08:04 hmmmm need to sleep 08:04 thexyz what do you mean by dirty repo? 08:04 hmmmm nothing, I guess they can just build at the 0.4.8 tag 08:05 thexyz i thought that's pretty common 08:05 thexyz yes i guess some sleep won't hurt.. 08:05 thexyz wat 08:05 thexyz sfan5: did you delete msvc build? 08:05 sfan5 no 08:05 sfan5 well... yes I did 08:05 thexyz why is it not here anymore then? 08:06 thexyz okay 08:06 sfan5 I wanted to delete mine 08:06 thexyz that's fine since I wanted it to appear on the left anyway 08:06 thexyz and there's apparently no other way to achieve that 08:06 sfan5 I'll reupload mine 08:07 thexyz alright 08:07 thexyz tell me when you're done 08:08 sfan5 thexyz: done 08:09 sfan5 I'll test it on Windows now 08:09 thexyz it doesn't work 08:09 thexyz can you please test it before uploading? 08:10 thexyz http://i.imgur.com/M2EWAeQ.png 08:11 sfan5 that's weird, it worked fine before 08:11 thexyz I honestly think that sapier is the one who should do mingw builds 08:11 thexyz well go fix it somehow 08:16 sfan5 I think I forgot to replace some DLLs 08:18 sfan5 yep that was it 08:23 sfan5 thexyz: fixed version uploaded 08:24 thexyz seems to work now, what irrlicht version are you using? 08:25 thexyz oh well it sorts downloads by file name 08:25 thexyz okay, I'll publish it now 08:27 thexyz https://forum.minetest.net/viewtopic.php?id=7819 08:33 VanessaE ok. my servers are updated to 0.4.8. new build goes online at the next backup cycle in a few hours. 08:34 sfan5 thexyz: 1.7.2 08:35 thexyz your checkboxes look white 08:35 sfan5 I think I should update to 1.8 sometime 08:52 celeron55 github has re-enabled it's binary upload functionality? 08:52 celeron55 that's pretty neat 08:54 VanessaE (correction, the new build is online) 08:56 VanessaE aside: ShadowNinja reported yesterday that the new lavacooling is not being logged by the rollback recorder. This may be....a problem. 08:57 VanessaE (well it ain't new but whatever) 09:00 thexyz a while ago 09:02 VanessaE oops. 10:32 PilzAdam eh, why did hmmmm add all the old branches again? 10:33 PilzAdam and why isnt minetest_game taged? 10:34 PilzAdam also why did nobody push to the stable-0.4 branches at minetest and minetest_game? 10:42 PilzAdam I updated minetest.net/download 10:43 PilzAdam so we need to do: 1) tag minetest_game 2) push to stable-0.4 branches of minetest and _game 3) remove all the additional branches hmmmm pushed 10:45 PilzAdam .... hmmmm missed doc/menu_lua_api.txt, it still says "Minetest Lua Mainmenu API Reference 0.4.6" 10:49 PilzAdam can somebody say "yes, do that"? 10:50 nore yes, do that... 10:50 nore except perhaps the branch removing things... why would you remove them? 10:50 PilzAdam thats a bit late now ;-) 10:51 PilzAdam I can ask the same thing: why would you add them? 10:51 nore wait, what are those branches exactly 10:51 nore ? 10:51 PilzAdam the old branches of c55's repo that were removed some time ago from minetest/minetest 10:55 PilzAdam ok, pushed to stable-0.4 branches, taged mt_game and removed the old branches 10:56 PilzAdam we still need to update the changelog in the dev wiki... 11:00 PilzAdam requested builds in launchpad, they should arrive in some hours 11:03 PilzAdam did I miss something? http://dev.minetest.net/Releasing_Minetest 11:57 PilzAdam I guess the feature freeze can stop when we reenable the -dev suffix in CMakeLists.txt 14:13 iqualfragile my server wont announce to servers.mineguild.de anymore 14:15 sapier did you use wireshark yet? 14:19 iqualfragile i cant its a remote server 14:20 sapier tcpdump? 14:20 iqualfragile not privileged enought 14:21 thexyz wtf is servers.mineguild.de? 14:21 iqualfragile sorry 14:21 iqualfragile net 14:21 thexyz wtf is servers.mineguild.net? 14:21 iqualfragile … 14:22 iqualfragile servers.minetest.net 14:22 iqualfragile 14:59:52: ACTION[ServerThread]: announcing to servers.minetest.net 14:22 iqualfragile nothing happened yet 14:25 sapier guy I'm planning to create a spawning framework any comments or ideas to it? 14:25 nore good idea... but will this be the base of a more general mob framework, or not? 14:26 sapier can someone change topic? 14:26 sapier it's a feature that is usefull for mobs yes but is completely unrelated as of code 14:26 nore didn't even see it... but I don't know how to do so 14:26 PilzAdam sapier, why change the topic? 14:27 sapier isn't 0.4.8 released? 14:27 PilzAdam I guess the feature freeze can stop when we reenable the -dev suffix in CMakeLists.txt 14:27 sapier thought that was meant as "I do this now" ;-) 15:16 hmmmm ?? what do you mean PilzAdam, the lua_api.txt clearly says 0.4.8 15:17 hmmmm the old branch thing, I deleted them on accident so I put them back 15:17 PilzAdam menu_lua_api.txt 15:17 hmmmm oh now there's a menu api 15:17 PilzAdam also, you havent deleted them on accident, I did that a while ago 15:17 hmmmm can we make a script to do all this nonsense 15:17 hmmmm no no no don't delete them 15:17 hmmmm why would you do that 15:18 hmmmm as for minetest_game, I don't do those things 15:18 hmmmm note that in the releasing minetest wiki page it says that one or more core devs do these steps 15:18 PilzAdam because these branches belong in celeron55's repo, and not in the upstream one 15:18 sapier can't we create a real usefull documentation? hyperlinks have been invented ages ago and we don't even them 15:20 PilzAdam hmmmm, if you release minetest then you have to do these things in minetest_game 15:20 PilzAdam minetest_game is part of the release 16:55 sapier does anyone know about the rollback out of memory issues? 16:56 ShadowNinja sapier: Yes, therefore the SQLite rewrite. 16:57 sapier why did it seem to work without sqlite rewrite before? 16:57 sapier btw I just made my little bro test 0.4.8 .... we should've fixed the modmgr stall prior releasing 0.4.8 16:58 ShadowNinja sapier: It did work, but it loaded the entire file into memory and slowly scanned it. A issue with 1GB+ rollback logs. 16:58 sapier that memory issue is what in minetest just got told about the current version? 16:58 ShadowNinja Should one line conditionals have braces? 16:59 sapier imho you always should use braces 16:59 ShadowNinja sapier: I don't understand that sectence. 16:59 sapier megaf said in minetest channel that old 0.4 branch rollback was fine 16:59 sapier while current is broken 17:00 PilzAdam rollback hasnt changed much since it was added 17:00 ShadowNinja Is it OK if I add a astyle.conf in util will SQLite rollback? I'm using it to fix the code style of that. 17:00 sapier so maybe he's remembering wrong? 17:00 ShadowNinja sapier: It only becomes an issue when the rollback log gets really big. 17:01 sapier astyle.conf for what? 17:01 ShadowNinja sapier: astyle, or artistic style, a automatic code restyler. 17:02 sapier so some special tool you use? 17:02 ShadowNinja sapier: http://astyle.sourceforge.net 17:03 ShadowNinja We could run it on all the code, but that would break every pull request. 17:04 ShadowNinja It can't remove uses of hungaruan notation and things like that of course though. 17:05 sapier if we use a tool like that we should all agree ... about astyle.conf imho we shouldn't have tool specific configs within git if they're not required for minetest build 17:06 ShadowNinja Well I think we should have it so devs can run it on their code before making a pull request. 17:06 sapier that requires coding style to be defined way more exact than it is now 17:06 sapier and I don't think coding style is a issue justifing that big changes 17:07 sapier but of course that may be only my opinion 17:07 ShadowNinja Non-existent options leave the code alone, but the official style is the Linux style. 17:07 PilzAdam a more uniform code-style would be good 17:08 sapier PA yes but would require large parts of minetest to be touched 17:09 PilzAdam sapier, thats why ShadowNinja's suggestion is good 17:09 PilzAdam ShadowNinja, can astyle check pull requests? 17:10 sapier so we don't have quality gates for checkin but style gates? ;-) shadows idea is good but imho not important enough to justify the amount of work to be done 17:11 OldCoder btw I just made my little bro test 0.4.8 .... we should've fixed the modmgr stall prior releasing 0.4.8 17:11 ShadowNinja PilzAdam: Well there isn't GitHub integration with astyle, but you can run "astyle --options=util/astyle.conf src/changedfile.cpp && git diff" 17:11 OldCoder "I command you to play this game!" 17:11 OldCoder :-) 17:12 PilzAdam ShadowNinja, you can only run it on whole files? 17:12 OldCoder So, curl media is still broken in 0.4.8, right? 17:12 OldCoder Needs to be disabled? 17:12 ShadowNinja PilzAdam: Yes, it has no diff mode that I am aware of. 17:12 PilzAdam ShadowNinja, thats bad, since your suggestion that people can run it "on their pull requests" doesnt wrok 17:12 PilzAdam *work 17:13 sapier oldcoder that's close to the words I used ;-) 17:13 OldCoder Heh 17:13 sapier no it's not broken but slow 17:13 OldCoder Oh, sorry; I thought this was the general channel 17:13 ShadowNinja PilzAdam: Well rewrites mostly, until we run it on the whole project, if we do. 17:13 OldCoder Coercion of brothers not on-topic 17:14 OldCoder sapier, Vanessa reports and I have confirmed that some textures never load if curl media is used 17:15 OldCoder httpfetch fixed it for me but I understand that httpfetch will take some time to merge 17:15 OldCoder Not slow but broken; this seems to be confirmed 17:15 sapier ok that's a different thing from modmgr stall 17:16 sapier as we're talking about automatic code checks ... what about cppchecking minetest ;-) ... I'm running a first test right now ... no idea how much errors will be found ... guess something >1k 17:25 hmmmm hmm 17:25 hmmmm guys, please make comments and suggestions on http://dev.minetest.net/Lua_Optimization_Tips 17:25 sapier os.clock is useless 17:26 ShadowNinja https://github.com/ShadowNinja/minetest/commit/6efc81ee01c63f1c781e0979b36b1a284a567887 17:26 sapier you can only find stupid bugs using it 17:26 hmmmm it works for rough estimates though 17:26 PilzAdam hmmmm, the whole site implies that the only thing that can be optimized is the vmanip usage 17:26 ShadowNinja s/site/page/ 17:26 PilzAdam s/page/article/ 17:26 sapier no it isn't even a rough estimation there's almost no sane function resulting in a delta > 0 17:26 hmmmm PilzAdam, that's sort of true, which is why I'm asking for other people to contribute to it 17:27 PilzAdam hmmmm, when using os:clock() in simple mobs every function is always 0 ms 17:27 hmmmm what I currently have there could be put under "considerations using vmanip" 17:27 sapier mobf for example results in ~5% cpu usage if using socket but 0 with os.clock 17:27 ShadowNinja s/:/./ 17:27 hmmmm PilzAdam, the resolution is quite small but it works for lua mapgens and whatever 17:28 hmmmm when we're talking hundreds of MS 17:28 sapier mapgens are in big fat laggy things 17:28 PilzAdam hmmmm, then we need to add a way to profile short functions properly 17:28 sfan5 microseconds! 17:29 sapier "Use separate variables when possible instead of tables" this is only true for simple mods doing nothing 17:29 celeron55 sapier: i've ran cppcheck on minetest... i guess sometime this year 17:29 celeron55 so you won't find the worst things anymore 8) 17:30 ShadowNinja minetest.get_timer():start():stop() 17:30 sapier hope so celeron ;-) 17:31 sapier by now most things are style issues so you're most likely right 17:31 Exio4 are there news about android's port code? 17:40 sapier nodedef.cpp 455/501 :-) who can see the error in there (there actually is one) 17:41 hmmmm 455 and 501? or 455 to 501? 17:41 sapier no both lines same error 17:42 sapier it's a quite tricky one 17:42 hmmmm something to do with that const qualifier at the end of the function definition i assume? 17:42 sapier no const is not involved 17:46 sapier this one might help finding it: [src/nodedef.cpp:455]: (error) Returning reference to temporary 17:46 hmmmm you can't chain references like that? 17:46 celeron55 it doesn't return a reference to temporary 17:46 celeron55 your checker is borked 17:47 sapier I don't think so celeron 17:47 sapier what about the [] operator? 17:49 celeron55 std::vector's [] returns a reference 17:51 sapier hmm 17:52 sapier I'm still not fully convinced cppcheck usually is more accurate then me ;-) 17:53 iqualfragile Exio4: there is an android port by some chinese rippoff company 17:53 Exio4 yes, i mean its source code 17:54 iqualfragile Exio4: go ahead and request it 17:54 iqualfragile one version is free 17:54 iqualfragile download it and you have the right to get the sourcecode 17:54 Exio4 they said they will release its code "soon" 17:54 iqualfragile but they might not give you the actual android part because minetest is lgpl and if they did it right they might not have to 17:55 celeron55 iqualfragile: i have requested it on 2013-10-30 and they said they would do it 17:55 Exio4 you could have said there are no news and it would have been faster 17:55 celeron55 but apparently they don't 17:56 iqualfragile well, then i would ask them again, pointing out that some time has passed 17:56 celeron55 why do people have to lie like that; it benefits nobody 17:56 iqualfragile and then tell them to stop using the name minetest for their ports 17:56 hmmmm ahh ! 17:56 celeron55 they have stopped using the name 17:56 hmmmm proller's development branch is what people were looking for I think 17:56 celeron55 at least mostly 17:57 Exio4 they should release changes to the source code, anyway 17:57 hmmmm there was that one other guy who was fed up with things not getting added but his fork didn't work out ever 17:57 celeron55 (no need to be so strict about that) 17:57 iqualfragile hmmmm: proller? 17:58 hmmmm yeah I was reading the forum a bit 17:58 hmmmm he has his own branch with all of his own stuff and some other pull requests that never got merged because of controversy or something other 17:58 iqualfragile well, one has to admit that he has a point 17:58 hmmmm anyway, right now since sapier's around I think now's the right time to merge the async stuff 17:59 iqualfragile merging or closing pull requests is quite slow in minetest 18:00 sapier I guess I should rebase first? 18:00 PilzAdam hmmmm, first we need to get rid of the feature freeze 18:00 PilzAdam that is: we reenable the -dev suffix 18:00 hmmmm oh 18:00 PilzAdam I can do that now if people want 18:00 hmmmm you do it 18:00 sapier I'm rebasing while you do 18:01 OldCoder Ah. MT is LGPL and not GPL. So the Android port may be legal. There has been some discussion of the matter. 18:01 sapier the port most likely is legal the usage of name isn't ;-) 18:02 hmmmm sapier, async has jsemaphore along with it, right? 18:02 sapier yes 18:02 hmmmm alright two birds with one stone i guess 18:02 sapier so if async is merged we can close that request 18:02 hmmmm what do you use that for anyway 18:02 sapier it's quite similar to events but events are non counting semaphores only 18:02 hmmmm right 18:03 celeron55 OldCoder: chances are it isn't 18:03 hmmmm we currently have too many synchronization functions 18:03 sapier it's used to distribute jobs to worker threads 18:03 sapier hmmm we only have those that any os provides and which are commonly required ;-) 18:03 hmmmm I think sometime it'd be a good idea to change events to jsemaphores of value 1 18:03 celeron55 OldCoder: we have the file listing of it and there's nothing in there suggesting LGPL would have been circumvented 18:03 OldCoder celeron55, thank you. Others, does async improve startup time or reliablity? If so, should I remain with git current as opposed to stable? sfan5 figures stable is safer but people wish fast and reliable startup. 18:03 sapier yes event's could be dropped 18:03 hmmmm like just turn Event into a bunch of #defines 18:04 sapier -' 18:04 sapier but if I remember correct they use different things on windows .. no idea why windows has different variants of semaphores 18:04 hmmmm that's true 18:05 celeron55 OldCoder: basically, minetest is in the listing as one single library (that's how you package C++ code for android), while it would be in two pieces if they would've implemented all of their additions separately 18:05 celeron55 and implementing their things in pure Lua isn't possible at the moment 18:05 OldCoder celeron55, O.K. 18:05 OldCoder Hmm 18:06 OldCoder celeron55, you said nothing to indicate that LGPL is circumvented but it sounds as though they did *not* do it correctly 18:06 celeron55 i mean they've broken the license 18:06 OldCoder Yes; so perhaps they can be pressured to release the source 18:06 sapier ok updated async request 18:06 Jordach celeron55, they've also broken the GPL too: SuperTuxCart code is in that package too 18:07 hmmmm if they made modifications to the LGPL'ed library they need to contribute back the changes 18:07 hmmmm since the android port took some code modification, we'd get that back 18:07 celeron55 yes, that's how it works 18:07 hmmmm well, I mean they need to be able to provide the source code of the LGPL'd component at all time, and it just so happens that they modified it so we can only get it to them (right??) 18:07 PilzAdam lets call stallman ;-) 18:08 hmmmm I really doubt the FSF would be happy about us trying to sue chinamen 18:08 hmmmm the response would basically be "we can't do anything about that" 18:08 Jordach hmmmm, GPL is also there, see vanessa's console: https://forum.minetest.net/viewtopic.php?pid=116881#p116881 18:09 OldCoder Can they be dealt with on the Google side? Tell them will have package pulled from the Store unless they comply? 18:09 PilzAdam whats "reasnable", CMakeLists.txt:59 18:09 hmmmm probably another sapier typo 18:09 hmmmm jeez I at least look over my commits before committing them..... usually. 18:10 sapier :-) guess I managed to get famous for my typos 18:10 sapier :-/ 18:11 sapier but that one isn't mine 18:12 sapier the tribute goes to celeron for this one 18:13 sapier ok something is missing in the updated async pull request 18:19 sapier may I merge asynchronous_events_2? 18:19 sapier https://github.com/minetest/minetest/pull/977 18:20 PilzAdam sapier, I already see a print() call there 18:20 PilzAdam and there are newlines missing at the end of files 18:20 PilzAdam and put spaces after , in function parameter lists! 18:24 sapier the newlines are in ... did all of them in jsemaphore.cpp ;-P 18:25 PilzAdam I mean in builtin/async_event.lua and async_env.lua 18:25 sapier .... sorry forgot the *joke* tag ... *irony tag* 18:27 hmmmm everything all fixed up? 18:27 PilzAdam I guess someone should test it first 18:27 sapier no still checking give me a minute 18:27 PilzAdam somebody else than sapier 18:27 sapier I hope more than one person did already check it asynchronous things are prone to contain errory 18:27 sapier -y+s 18:29 proller iqualfragile, what? 18:30 OldCoder Wish me to build and test? 18:31 sapier oldcoder it's not the missing textures issue only the mainmenu stall 18:31 proller good astyle for minetest - astyle --style=attach --indent=tab --keep-one-line-blocks --keep-one-line-statements --indent-classes --pad-oper *.cpp *.h 18:31 proller or astyle --style=attach --indent=tab --keep-one-line-blocks --keep-one-line-statements --pad-oper *.cpp *.h 18:31 OldCoder sapier, That is fine; offer stands 18:32 sapier I'm about to update pa's findings so just a couple of minutes plz 18:34 sapier ok oldcoder you can test it now 18:34 PilzAdam "ERROR: Async ENGINE step: /home/adam/Minetest/minetest/builtin/async_event.lua:14: attempt to call a nil value" 18:34 sapier great what did you do? 18:34 PilzAdam I was using the modstore 18:34 sapier what did you do EXACTLY 18:35 PilzAdam first time it completly crashed when clicking on the down button 18:35 PilzAdam then I ran it in gdb and did the same, i.e. open modstore and click on the down button 18:35 sapier can you post the error? 18:35 OldCoder sapier, You wish me to rebuild server, client, or both? 18:35 OldCoder Using git current? 18:35 PilzAdam sapier, I only got this line 2 times 18:36 sapier it's client only but just wait a moment 18:36 PilzAdam the second one has a "19:34:00: ERROR[AsyncWorkerThread_0]:" in front of it, but its exactly the same else 18:36 PilzAdam is it normal that "Initializing Asynchronous environment" is printed 4 times? 18:37 sapier yes it's printed once for each created environment 4threads --> 4 environments 18:37 sapier the error you get can only happen if there's no valid job in table 18:38 PilzAdam I can not reproduce it 18:38 sapier very strange ... I'm gonna add a check there but I don't see a reason how this could even happen 18:38 PilzAdam also the crash doesnt happen again 18:39 sapier still I'm gonna add a check 18:39 PilzAdam PANIC: unprotected error in call to Lua API ( 18:40 hmmmm do you people see why I didn't want to add this before 0.4.8 18:40 hmmmm do you freaking see why now 18:40 * OldCoder is in now rush. He sees PANIC and decides to wait :-) 18:40 OldCoder * no rush 18:40 sapier I wonder why those errors only happen on pa's machine ;-) 18:40 sapier but better there than at users machine 18:40 PilzAdam backtrace https://gist.github.com/PilzAdam/7630676 18:41 PilzAdam that happens when opening the modstore and quickly clicking the down button multiple times 18:42 sapier you're on linux true? 18:42 PilzAdam yes 18:42 PilzAdam oh, another one! 18:43 sapier you're sure you're using LATEST version? 18:43 PilzAdam your asynchronous_events_2 branch 18:43 sapier checkd out now not some weeks ago I assume? 18:44 thexyz Jordach: I don't see the GPL here, care to point me to it? 18:44 PilzAdam I rebased it against master 18:44 Jordach later 18:44 sapier I'm not concerned about master but about a old version of asynchronous_events_2 18:44 PilzAdam sapier, the other one: https://gist.github.com/PilzAdam/7630720 18:45 thexyz Jordach: no, not later, either stop saying shit or prove the shit you say 18:45 thexyz because this starts to get irritating 18:45 sapier thats same as your first one 18:45 PilzAdam sapier, well, I dont have your latest style-fix commit in, but everything else is your branch 18:45 PilzAdam it looks different! 18:46 sapier I can't reproduce any of those crashes by now 18:46 PilzAdam it happens quite randomly 18:46 PilzAdam it seems like you have to click down before the first entries load, but Im not sure about that 18:47 PilzAdam I wonder why those errors only happen on pa's machine ;-) <- I guess Im just a good tester ;-) 18:47 sapier wow you're fast ;-) 18:50 sapier can you try to remove L335 in l_async_events.cpp ... I added this based uppon shadows error handler fixes but I might have understood something wrong there 18:50 ShadowNinja nore: Can you add protection support to the screwdriver too? 18:52 sapier shadow can you have a look at l_async_events.cpp L335 if this call is really wrong? 18:54 sapier PilzAdam? does this change help? 18:54 thexyz yes, I remember VanessaE being a good tester too 18:55 thexyz but then we've figured she had a broken ram module 18:56 sapier great ... random memory errors again 18:56 VanessaE well I do tend to be good at finding obscure, corner-case bugs and issues ;) 19:01 celeron55 let's implement error correction codes for all data 19:02 celeron55 (the world would be a much better place if non-ECC RAM was illegal in consumer products...) 19:03 VanessaE PilzAdam's computer;s just getting struck by too many cosmic rays :P 19:04 sapier celeron55 blaming hardware is too easy most time it's still programmers fault 19:05 iqualfragile celeron55: yes, indeed 19:05 PilzAdam sapier, programmers are hardware too 19:05 iqualfragile sapier: broken ram is kill 19:05 VanessaE I used to have the worst issue with gcc crashing during compiles years ago. they kept blaming my RAM, even after multiple tests with memtest86. Turned out to be a weird bug in the compiler. 19:05 PilzAdam sapier, trying that now... 19:07 PilzAdam sapier, still happens 19:07 PilzAdam (I commented "lua_pop(m_LuaStack,1);" out) 19:07 sapier would've been to easy 19:20 sapier hmm I can't reproduce the crash without that line ... yet there's another bug in there 19:25 sapier PilzAdam could you provide some backtraces maybe more of them show up some common case? 19:26 sapier OldCoder if you still have time and know how to use gdb maybe you could help too ;-) 19:27 PilzAdam sapier, https://gist.github.com/PilzAdam/7631274 19:28 sapier hmm I wonder if lujit might be involved 19:30 sapier wow that was fast 19:32 sapier PA seems luajit is way more crashy then lua could you try to verify that assumption? 19:32 PilzAdam maybe later 19:35 VanessaE hm, the master server list doesn't use reverse dns? 19:37 proller Don't write player files all the time if they are not modified (PilzAdam) -- seems it not working 19:37 thexyz VanessaE: what do you mean? 19:38 VanessaE thexyz: if server_address is not set in the server's config, the master server list only shows the IP address, even if the server has a reverse DNS already set up. 19:39 thexyz oh right 19:43 proller most of reverse names is like ip-1.2.3.4.adsl.superprovider.net 19:44 proller and need double check ip->name->ip2==ip 19:44 VanessaE no matter. Easy enough to work around. 19:47 thexyz and no one sets reverse dns anyway 20:18 PilzAdam proller, what? 20:19 proller minetest still writes all player files every N minutes 20:19 Calinou as long as it doesn't write too often it isn't a problem 20:19 Calinou it is also good in case the server crashes 20:19 Calinou so that players do not lose too many items 20:20 proller i lost ~6000 player files 2 weeks ago 20:21 Calinou it's not like your server has too important data :P 20:21 PilzAdam proller, hm? thats weird 20:21 proller its because all files rewrites every N minutes 20:22 proller can test it with strace , truss or debug this place 20:23 PilzAdam wtf 20:36 PilzAdam seems like checkModified() returns true for not connected players 20:37 PilzAdam how do I check if players are connected? 20:39 sapier PilzAdam any idea if your libcurl is built against a threadsafe openssl? 20:40 proller PilzAdam, maybe store last save date ? 20:43 PilzAdam sapier, no 20:44 sapier I found out that openssl & co are not thread safe by default ... but no idea if this is (one of multiple) root cause here 20:48 PilzAdam proller, since when doesnt it work anymore? 20:48 ShadowNinja The player data shouldn't even be loaded if the player is disconnected. 20:48 PilzAdam it seems like the inventory is always seen as modified in checkModified() 20:50 proller PilzAdam, doesnt work what? 20:52 qwrwed i have a problem with minetest: it keeps crashing with the error "Some exception: "bad allocation"" 20:53 ShadowNinja qwrwed: Backtrace? Steps to reproduce? 20:54 qwrwed i use minetest 0.4.8 on VanessaE's Survival Server, and nothing in particular seems to trigger the error except for playing for a while 20:55 qwrwed so if you can play normally on that server without it crashing i don't know how you would reproduce it 20:56 ShadowNinja qwrwed: Are you out of memory? Can you get a gdb backtrace? 20:56 celeron55 that means it's running out of memory 20:56 PilzAdam seems like m_last_inventory = inventory; doesnt work 20:56 PilzAdam assert(m_last_inventory == inventory); directly under it fails.... 20:56 qwrwed i will download gdb 20:56 celeron55 qwrwed: don't yet; open some task manager or equivalent and see how memory usage goes 20:57 qwrwed alright 20:57 celeron55 if it's all used at the time that happens, it's the problem; if not, then there's something weird going on 20:58 celeron55 (in that case something weird = it tries to allocate memory according to some corrupted variable) 20:58 qwrwed started off at 20,000 K 20:58 thexyz celeron55: so what shall we do with android port? 21:00 ShadowNinja Can we get google to take it down? If so try to do that if they don't give you code by next week, or something like that. 21:00 sapier ok I guess I found a issue that might be hiding the real issue 21:00 PilzAdam does someone see bug in == or = for Inventory and InventoryList? 21:01 celeron55 thexyz: three options: 1) don't care, 2) contact the developer again, 3) contact fsf 21:02 thexyz celeron55: i'm mostly talking about the topic, leaving it on forums seems like a bad idea 21:02 celeron55 oh the topic... uh, well i'm okay with moving it to trash 21:02 thexyz tried contacting marktraceur? 21:02 celeron55 we don't support illegal stuff anyway 21:03 celeron55 marktraceur suggested contacting fsf 21:03 celeron55 (i contacted him immediately when i saw the android port :P) 21:05 celeron55 thexyz: if you prefer something, tell it 21:05 celeron55 at the moment i prefer not caring 21:05 thexyz what level of not caring? 21:06 celeron55 hide the topic in order to not promote it and do nothing else 21:06 thexyz also there was a guy who thought it's okay to post a link to cracked .apk, i've banned him 21:06 thexyz for 1 month 21:06 thexyz alright 21:07 thexyz moved 21:07 celeron55 if the android port gets popular, then we continue like real assholes 21:07 celeron55 until then, whatever as long as they don't use our name 21:07 qwrwed it crashed at 50% CPU, at about 700,000 K after 7 minutes 21:07 celeron55 qwrwed: and how much do you have free now? 21:08 celeron55 what OS are you using? 21:08 qwrwed after the crash, it stayed at 525,164 K of memory, and <10& CPU 21:08 qwrwed Windows XP SP3 21:10 qwrwed at 525000K it said it had 200000K of memory left 21:10 qwrwed so it must have used up all of the memory when it reached 700 000K 21:12 celeron55 have you disabled the windows pagefile? 21:13 celeron55 MT doesn't have any handling for running out of memory (like most software) and it completely relies on the OS swapping out stuff on disk, which usually prevents running out of memory 21:14 qwrwed how would i do that? 21:15 celeron55 dunno; if you use such an old system, you better know how it works 8) 21:15 celeron55 you can decrease the client_unload_unused_data_timeout configuration setting of minetest to make it use less memory 21:15 qwrwed i don't... I have asked for a laptop, but it doesn't look like i'll get one soon 21:16 qwrwed i'll try seeing if i can do that stuff tomorrow 21:16 celeron55 maybe something like client_unload_unused_data_timeout=200 could work for you 21:17 qwrwed does this go in minetest.conf or elsewhere? 21:17 celeron55 yes, minetest.conf 21:18 qwrwed ok, thanks 21:18 qwrwed anything else that might help before i go? 21:18 celeron55 oh also, taking 700MB of memory with default settings sounds a bit excessive though; i wonder if it's some kind of a memory leak - in that case this won't help 21:18 celeron55 but try it and see 21:19 celeron55 well, make sure to not use high-resolution texture packs if you have so little memory 21:19 qwrwed i use the default pack 21:20 qwrwed combined with the sheer amount of nodes on that server (7000+) it would just take far too long to load 21:22 sapier1 PilzAdam I fixed async thread deletion ... that one was completely messed up I guess those errors did shadow the real error 21:31 PilzAdam sapier, /home/adam/Minetest/minetest/src/jthread/pthread/jthread.cpp:49:1: warning: no return statement in function returning non-void [-Wreturn-type] 21:32 sapier ok I'm gonna fix it, win32 code is missing too I just want to know what your original error was 21:33 PilzAdam https://gist.github.com/PilzAdam/7632693 21:34 sapier can you post the other threads too? 21:34 PilzAdam how do I do that? 21:34 sapier t 21:34 sapier bt 21:34 sapier info threads shows all threads 21:35 celeron55 thread apply all bt 8) 21:36 sapier does this really work? *g* 21:36 PilzAdam https://gist.github.com/PilzAdam/7632715 21:36 celeron55 it's probably impossible to learn everything gdb can do 21:38 sapier ok 3 threads withing gnutls ... 21:39 sapier suspicious ... but could be non related too 21:46 sapier ok seems that gnutls isn't threadsafe unless initialized correctly ... does anyone do this? 21:47 sapier http://www.gnutls.org/manual/gnutls.html#Thread-safety 21:47 kahrl > can be safely divided across threads as long as a single thread accesses a single object 21:47 kahrl isn't that what happens? 21:47 kahrl I read the same kind of statement about openssl 21:49 sapier not quite sure there shouldn't be different threads involved with a single struct ... but 21:49 sapier I'm not sure if this is true for results too 21:50 sapier The random generator of the cryptographic back-end, is not thread safe and requires mutex locks 21:51 sapier but doesn't exactly match error situation true 21:51 kahrl the text states that those mutex locks are supposed to be initialized by gnutls 21:52 sapier since 2.12 ... guess noone has a more old version 21:52 celeron55 so does calling gnutpls_global_init() fix the problem? 21:52 celeron55 -p 21:53 sapier no Idea the problem doesn't happen for me only pa can reproduce it 21:53 celeron55 do you use gnutls? 21:53 sapier yes but most likely a different version 21:53 celeron55 but anyway, if so, then there is a problem, because minetest doesn't use gnutls or openssl directly 21:54 sapier exactly we have to rely on curl doing things correct 21:54 celeron55 and minetest doesn't even know what curl is using; linux distros provide gnutls, openssl and non-crypto versions of it 21:59 celeron55 wait what 21:59 sapier yes and gnutls is just a guess too it may be something completely different 21:59 celeron55 minetest doesn't call curl_global_init() 21:59 sapier it doesn't? 22:00 kahrl curl_easy_init calls curl_global_init 22:00 celeron55 it doesn't look like so (i have a slightly old version) 22:00 celeron55 okay, that explains 22:00 celeron55 https://github.com/winlibs/cURL/blob/master/lib/gtls.c#L163 22:00 kahrl but... 22:00 celeron55 and curl does that in it 22:00 kahrl > This may be lethal in multi-threaded cases 22:00 kahrl from the curl_easy_init manpage 22:01 celeron55 (that's some old version of curl that popped up in my google search) 22:01 sapier hope curl_global_init is reentrant 22:01 kahrl it's not 22:02 sapier so what happens if it's called by different threads? on different local variables 22:02 celeron55 gnutls_global_init() is not threadsafe either 22:03 kahrl sapier: the libraries might try to initialize themselves twice or more concurrently -> memory corruption? 22:03 celeron55 minetest really needs to call curl_global_init in it's main thread before anything else with curl happens 22:04 kahrl sapier: is that the branch with httpfetch added? if so it calls curl_global_init 22:04 sapier no havent rebased that branch 22:05 sapier the asynchronous_events_2 branch only, there've been other bugs in that aren't fixed in merged version by now 22:06 celeron55 PilzAdam: can you test that, assuming your build does not already call it? 22:06 PilzAdam what? where? 22:07 sapier the only thing I'm sure about that bug is it's some sort of memory corruption ... but no idea where it's from 22:09 celeron55 PilzAdam: call curl_global_init() from main() or somewhere else in the main thread before curl is used in other threads 22:10 PilzAdam that seems to work 22:10 celeron55 oh, good 22:10 sapier great :-) 22:11 PilzAdam I still get tons of 23:10:18: ERROR[AsyncWorkerThread_0]: ERROR: Async ENGINE step: /home/adam/Minetest/minetest/builtin/async_event.lua:14: assertion failed! 22:11 sapier there's one possible case I realized where this could happen ... yet it's very strange 22:12 PilzAdam mainmenu bug: when I go back from the online mod repo the game name is written in the topleft corner, although it should only be there in the singleplayer tab 22:12 sapier if a thread is finished prior it's callback is inserted to joblist but I'm not even sure this really can happen 22:13 sapier that's because someone changed background for dialogs to without realizing there aren't only create/delete world dialogs 22:13 sapier didn't realize this until you just said it too ;-) 22:13 PilzAdam backgrounds? 22:14 PilzAdam Im talking about the gamename next to the version string 22:14 sapier there was a commit fixing background modname ... shown for those dialogs too, previous dialogs did reset things like that 22:14 sapier which is handled along with background and other things 22:14 PilzAdam the game background is not shown 22:15 sapier I'll look next where shall I add the curl init code for async pull? 22:15 PilzAdam I dont see any problems with the game background in dialogs 22:16 PilzAdam I placed it at the top of main(), but I guess that isnt a good location ;-) 22:16 sapier why not? 22:16 PilzAdam because its done before command line params are handled? 22:16 PilzAdam do you want to init curl when you run --version? 22:16 sapier yes 22:18 PilzAdam I would put it somewhere in Low-level initialization 22:18 sapier I put it after debug handler 22:19 sapier wat do you use as init param? 22:20 PilzAdam CURLcode res = curl_global_init(CURL_GLOBAL_DEFAULT); assert(res == CURLE_OK); 22:20 PilzAdam thats copied from http_fetch 22:20 PilzAdam http_fetch adds it right before "Run unit tests" 22:25 sapier ok pushed the updated pull request plz test if it works for you 22:36 sapier PilzAdam mainmenu.lua L1206 needs to check for singleplayer tab too 22:36 sapier I'll fix it tomorrow 22:49 celeron55 https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_env.cpp#L666 22:49 celeron55 buuuuuuuuug 22:49 celeron55 how isn't this fixed yet? 22:49 Kray line 22:49 Kray err 22:49 Kray 22:50 Kray ah 22:50 PilzAdam celeron55, hm? that was fixed a while ago... 22:50 Exio4 because it is in the 666 line celeron55 22:50 Kray can't write that because of my sixsixsix prevention 8) 22:50 sapier :-/ Ooops 22:50 celeron55 well what the fuck 22:51 PilzAdam celeron55, https://github.com/minetest/minetest/commit/dda2071cc083db2770044db85174a8629a7598b7#diff-85ad81e8c521c268f8011f58d52cf211L667 22:51 sapier why the hell did I do that? 22:52 celeron55 umm that is the fix, yes 22:52 sapier wait I have written the fix where did it go? 22:53 celeron55 https://github.com/minetest/minetest/commit/ab433775777c4f5055bcf4d2a1cffc506c4f9961 22:53 celeron55 there it goes 22:53 PilzAdam sapier, https://github.com/minetest/minetest/commit/ab433775777c4f5055bcf4d2a1cffc506c4f9961 overwrote it 22:53 sapier argh 22:53 celeron55 go explain here: https://forum.minetest.net/viewtopic.php?id=7823 22:53 celeron55 8) 22:54 VanessaE not to say "I told you so", but.... 22:54 PilzAdam my java prof always says copy and paste is bad, but in this case it would have been good ;-) 22:54 sapier I'm gonna create a pull request instantly 22:54 PilzAdam eh, you can directly push that fix 22:55 Exio4 what about directly fixing it? 22:55 sapier ok 22:55 PilzAdam first bugix after 0.4.8! 22:55 Exio4 haha 22:55 PilzAdam VanessaE, you can say that stable is outdated again 22:55 VanessaE lol 22:55 VanessaE told you so. 22:55 VanessaE :D 22:56 celeron55 it was good for many hours! 22:56 VanessaE plol 22:57 PilzAdam its only the first ball: http://devopsreactions.tumblr.com/post/44776341195/bug-reports-after-a-new-release 22:57 sapier don't you know you shouldn't read any bugreports for at least 48h after release celeron? 22:57 VanessaE sapier: you suck :) 22:57 celeron55 that floor looks strangely similar to minetest's issue list 22:58 VanessaE lol 22:58 sapier VanessaE I wasn't even allowed to commit when this bug was readded ;-) 23:01 sapier ok fixed 23:34 Sokomine regarding the mmdb: it would be great if the order in which "texture packs" and "mods" is shown in the menu of mt could be switched. that way, the "mods" entry would be visible, because it's a shorter text than texture packs 23:35 Sokomine right now, people might not expect more menu entries behind "texture pack" 23:35 Gethiox-mobile Do you know something about "if [torch in hand] then [emit light around the player]", is it possible to implement in minetest? I think it isn't a new idea, but I can't find any topic/information about this. 23:36 Sokomine gethiox-mobile: it exists on some servers, i.e. on realtest. iirc it is called "walking light". it lags a bit when you walk on such a server which such a light source in hand; but in general, it does work 23:38 Gethiox-mobile Sokomine: Oh, I'm suprised, thanks for the information. 23:48 Exio4 it is implemented in a hacky way because it is server-sided 23:49 Exio4 those light effects should be done client-side