Minetest logo

IRC log for #minetest-dev, 2015-04-01

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

All times shown according to UTC.

Time Nick Message
00:00 rickmcfarley joined #minetest-dev
00:14 hmmmm also there's a memory leak in loadPlayer(): if getPlayer returns NULL and a new player object is allocated, and the player file is not found, it returns without deallocating the player
00:15 hmmmm two bugs fixed with the same stone
00:20 hmmmm is it intentional for a player that is already found in the environment to get re-loaded on loadPlayer?
00:28 paramat joined #minetest-dev
00:31 paramat hi hmmmm please could you review #2580 ? Removing non-const references caused problems so i'm leaving that to another commit
00:31 ShadowBot https://github.com/minetest/minetest/issues/2580 -- Treegen.cpp: Add pine tree. Force place trunks. Speed optimise leaves blits by paramat
00:32 hmmmm lol 'optimise'
00:32 hmmmm your brit is showing!
00:33 paramat gosh yes!
00:33 hmmmm there should be a space after the for keyword
00:33 hmmmm the Topic: of a commit should be the component, not necessarily the specific file modified
00:33 hmmmm so like
00:34 hmmmm Treegen: Add pine tree, trunk force placement.  \n \n  Also fix some code style
00:34 hmmmm that's how I would write it
00:35 paramat yes i agree, should be 'treegen', i will also add spaces after the 'for's and 'if's
00:35 hmmmm looks good though
00:35 hmmmm spaces between operators
00:36 hmmmm so vmanip.m_area.index(pmin+p1) -> vmanip.m_area.index(pmin + p1)
00:37 hmmmm could condense the if_apple_tree thing to  vmanip.m_data[vi] = (is_apple_tree && is_apple) ? applenode : leavesnode;
00:38 hmmmm looks good
00:40 paramat okay. spaces between operators is something i prefer, c55s lack of spaces made me wonder what was expected. thanks, to work..
00:56 hmmmm oh btw i dunno if you fixed it yet but biomesdev has a failing decoration call
00:57 hmmmm dune grasses tries to spawn in an unregistered biome "forest_dunes"
00:57 hmmmm also you realize you *can* name the decoration... just add a name = "dune grasses"
01:06 paramat okay thanks will check
01:09 paramat yes that was fixed recently =)
02:04 deltib joined #minetest-dev
02:04 AnotherBrick joined #minetest-dev
02:19 Zeno` joined #minetest-dev
02:21 johnnyjoy joined #minetest-dev
02:32 paramat ugh 220 lines had missing spaces :P
02:54 hmmmm man i take it all back
02:54 hmmmm codeblocks is way better than kdevelo
02:54 hmmmm it looks way more polished and stable but it has its share of problems... i highly doubt the developers use their own IDE
02:55 hmmmm http://fpaste.org/205692/56914142/
02:56 hmmmm so what is a person like me supposed to do?  stop working on the project i actually care about to fix a tool that has big flaws?
03:05 Zeno` use QtCreator
03:05 Zeno` :p
03:06 hmmmm does QtCreator pass the ultimate test:  its code completion can handle namespace using directives
03:06 Zeno` ?
03:06 Zeno` example?
03:07 hmmmm irr::gui::IGUIElement *foo;  foo-> <code completion suggestion dialog opens here>
03:07 Zeno` yes
03:07 hmmmm now what I noticed 85% of free-as-in-freedom code editors do is fail when it's
03:07 hmmmm using namespace irr;
03:07 hmmmm gui::IGUIElement *foo;  foo-> oh dear no code completion
03:07 * Zeno` checks
03:08 hmmmm FOSS code editors are a big pile of shit
03:08 hmmmm visual studio is so, far, far, far superior to anything out there
03:08 hmmmm it just doesn't have fuckups
03:08 Zeno` works for std::
03:08 hmmmm here i feel like i'm fighting with it
03:11 Zeno` wahhhhhhhh I can't upload screenshot
03:28 WarrTab Ah, hi Zeno`
03:28 Zeno` Hi
03:28 hmmmm https://github.com/kwolekr/minetest/commit​/337e79c656a12bcf0dac4b25d5f0e021188e383a
03:28 hmmmm wooh, 4500th commit
03:29 WarrTab So, is my gui scaling worthy of a merge?
03:29 hmmmm NO
03:29 hmmmm it's horrible
03:29 hmmmm you ought to be ashamed of yourself
03:30 WarrTab It'd be kind of nice to have downloadable builds of PR's so they can get wider community testing
03:30 WarrTab Maybe instead of an "experimental features" checkbox, it could be a branch...
03:31 Zeno` LuaVoxelManip *o      no more, no less... just o
03:32 Zeno` looks like a 0 with my 3 pt Courier New font
03:32 WarrTab I figured I might try fixing the old look of fast leaves with clean transparent after guiscaling was merged, but I don't think I can maintain both patches at the same time...
03:34 hmmmm https://github.com/kwolekr/minetest/commit​/d1da014c1e716c1ff3744ed7e9bbdb9aebbe1d7d
03:34 WarrTab I guess I need a second core dev to thumbs-up it.
03:35 WarrTab hmmmm: Do you think you might have some time to review PR 2536 ?
03:35 hmmmm #2536
03:36 ShadowBot https://github.com/minetest/minetest/issues/2536 -- Anti-alias all 2D GUI Elements. by Warr1024
03:36 hmmmm ahh this PR
03:36 hmmmm i liked it when i first saw it
03:36 hmmmm let me just go through the code real quick
03:36 WarrTab Zeno's thumbs-upped it, IIRC, and I've implemented his feedback.
03:38 hmmmm heh
03:38 WarrTab Seems like the community's divided on whether to use the new look by default, though; I left the default as-is in the PR.
03:38 VanessaE there's one side-effect of your scaling/anti-aliasing code btw:
03:39 hmmmm if you're concerned about following the style, the opening bracket for functions should be on a new line
03:39 VanessaE it doesn't *quite* work on meshes (it doesn't fail, per se, but neither does it actually work like you'd expect :) )
03:39 WarrTab vanessae, are you talking about pr 2536, or the texture_min_size thing?
03:40 hmmmm did you look at it as of yesterday?
03:40 VanessaE the min-size thing.
03:40 VanessaE sorry, I'm a little tired and not entirely in-the-loop
03:40 hmmmm hrmm
03:40 hmmmm how doesn't it work?
03:40 hmmmm it seems fine for me
03:40 WarrTab I noticed that texture_min_size didn't affect some models, and I suspect it's to do with formats that don't give me an opportunity to intetcept the textures
03:41 VanessaE hmmmm: try it with a mesh node that uses a UV-mapped texture where the texture size is larger than the min threshold
03:41 Zeno` well I guess 337e79 looks pretty good :)
03:41 WarrTab oh, yesh, thst'd do it
03:41 VanessaE (only *just* larger)
03:41 VanessaE in which case it gets blurry
03:41 VanessaE it's not a huge issue
03:42 hmmmm hrmm
03:42 WarrTab It only enforces a min total resolution for the texture, but doesn't know how it'll be mapped
03:42 VanessaE WarrTab: right.  I don't expect that there's really any way to sort that out short of some kind of pre-processor that tries to evaluate the UV maps
03:42 VanessaE and that's just plain overkill
03:43 WarrTab might be, but it's beyond me, at least stm
03:43 WarrTab Ugh tabley keyboard so many typps
03:44 VanessaE yeah, not really worth screwing with - I just thought you should know about that particular corner-case.
03:44 hmmmm 2536 looks good to me
03:44 hmmmm anybody???  https://github.com/kwolekr/minetest/commit​/d1da014c1e716c1ff3744ed7e9bbdb9aebbe1d7d
03:45 WarrTab Vanessae, you could always crank texture_min_size to just below the point where smoke comes out of your gpu, if it bugs you :-)
03:45 VanessaE haha
03:47 WarrTab 64 is comfortable for me, 256 manageable
03:57 WarrTab hmmmm, Zeno`, if 2536 is all good, either one of you feel like pulling the trigger on the merge?
03:57 hmmmm sure why not
03:58 sofar ugh I didn't realize how bad minetest ran on nouveau.... kernel upgrade caused my card to fall out of the supported updated driver list :(
03:58 hmmmm sofar, ?? you mean nvidia proprietary driver?
03:59 sofar 9800gt, but 340.x driver deadlocks my gpu, and 340+ drivers no longer support the card
03:59 hmmmm yup
03:59 hmmmm you need to downgrade
03:59 sofar 340 supports the card
03:59 sofar but... locks it up
03:59 hmmmm imagine how i felt when i upgraded to 340+ and all of a sudden i couldn't use my third monitor :/
03:59 hmmmm bunch of BS
03:59 sofar I've got a 30" 2560x1600 monitor. one is enough for me.
04:00 sofar last dime I ever spent on nvidia cards, too ;^)
04:01 sofar I'll just wait until sky lake desktop boards come out next year
04:02 hmmmm ati cards are well supported these days, ironically
04:02 hmmmm i think an ati is going to be my next card
04:02 hmmmm amd, whatever
04:04 sofar I can't buy AMD, I work for intel, that would make me sick to my stomach, to say the least :)
04:04 sofar I wish they hadn't bought ATI, though
04:05 hmmmm heh
04:07 hmmmm WarrTab, minor code style issues, lines above 80 columns, and typos aside, I just tested it and it looks good!
04:08 WarrTab Yeah, I can't imagine I'll ever get a patch that large 100% right, so feel free to fix anything trivial you spot.
04:08 hmmmm I didn't because I figured that'd cause you trouble since you'd have to -f your own repository
04:09 WarrTab I suppose some people get all touchy about having their code modified, but I guess you can't really be particular when contributing to a free/open project
04:09 WarrTab Actually, once it's merged, I nuke my branches for it and rebase everything else
04:10 hmmmm ahhh
04:10 hmmmm well, five minute rule
04:10 WarrTab what's that?
04:12 hmmmm people can force push to upstream within 5 minutes of pushing in case they make a mistake
04:13 hmmmm meh, nevermind about the too-long-lines, guiFormSpecMenu.cpp is a ghetto
04:14 WarrTab Ah
04:14 hmmmm shoot
04:14 Zeno` I thought it was a bit more than 5 minutes
04:14 hmmmm I see a bigger problem
04:14 hmmmm why did I miss this
04:15 WarrTab Oh?
04:15 hmmmm the static/global variables
04:15 hmmmm 1). are you sure that's thread safe?
04:15 hmmmm 2). is there any benefit to encapsulating them into a class so state is easily resettable?
04:15 WarrTab I had an idea about a way to fix them, but it'd probably be a big rework...
04:15 hmmmm resettable/clearable
04:16 WarrTab They're reset in a sort of ghetto way right now
04:16 hmmmm do you mind if I rename them to g_ at least
04:16 WarrTab no prob
04:16 hmmmm okay
04:17 WarrTab I think in the long run, I was thinking of moving that filter into the hacky ^[ generateImage stuff, which I know at least should cover the points above.
04:17 hmmmm meh it's too late to force push
04:17 WarrTab Also, offering nnaa to mods could be nice
04:20 hmmmm https://github.com/minetest/minetest/commi​t/ddf96c7a17e45cdfb4a8060c3b0f74f17f7156df
04:20 WarrTab Thanks
04:21 VanessaE imho, you should turn the clean-transparent feature on by default, btw
04:21 VanessaE sorry for not mentioning that sooner.
04:21 hmmmm no i disabled that by default because people were crashing from it
04:21 VanessaE crashing??
04:22 hmmmm yeah
04:22 VanessaE huh.  wfm
04:22 WarrTab News to me
04:22 hmmmm and then others had some kind of huge performance regression on startup
04:22 WarrTab Huge texture packs?
04:22 VanessaE well sure there's a performance loss, but that's from the scaling
04:22 hmmmm the guy said he was using the default
04:22 VanessaE seems about the same (with the default textures) with, or without
04:22 hmmmm something must be wonky with his setup
04:22 VanessaE (using dreambuilder as usual)
04:23 hmmmm in fact I think that was today in the logs
04:23 WarrTab Maybe my next PR should be for enable_experimental_features
04:23 hmmmm ahh okay, you got that?
04:23 hmmmm that was on my todo list
04:23 hmmmm until you volunteered =]
04:24 * VanessaE shrugs...  like I said, wfm and it looks 1000x better than without.
04:24 VanessaE but with that in mind....
04:24 * VanessaE throws HDX-512 at it.
04:25 WarrTab I'll start looking into the experimental thing... *Should* be really simple...
04:25 WarrTab Can't guarantee it'll be pretty in the end, though
04:26 WarrTab Depends a bit on how defaultsettings works, i.e. whether they're loaded before actual settings and overwritten, or loaded after where missing.
04:26 WarrTab At least, the way I imagine it'll work.
04:27 hmmmm i thought about this a bit
04:27 hmmmm after g_settings is completely loaded, then if (g_settings->getBool("enable_experimental")  set_experimental_defaults();
04:28 VanessaE amazingly, it works fine, and the scaling delay from page to page in unified inventory isn't significantly longer with HDX-512 than it was with default textures.
04:28 hmmmm where that overrides the default setting (Settings::setDefault()) with the experimental features set on
04:28 VanessaE (also: offtopic: mesh nodes actually look pretty nice where HDX is being applied :-) )
04:28 WarrTab we still want explicit overrides in the conf file to override all, though...
04:29 hmmmm well then make enable_experimental an int
04:30 hmmmm 0 - disabled, 1 - set as defaults (can be overriden by the presence of explicit settings in the conf), 2 - force set settings (enables all features regardless of conf settings)
04:30 WarrTab What I was hoping for was that enabling experimental would only change thr *defaults* of some other settings, but the conf file was still supreme
04:30 hmmmm :)?
04:31 WarrTab I dunno if we would really need option 2?
04:31 hmmmm oh I thought the lack of option 2 was what you were objecting to
04:31 hmmmm if enable_experimental is true, then if you set the defaults, it'll still be overridden by an explicit setting
04:32 hmmmm did you look at how the settings work?
04:32 WarrTab not yet
04:32 hmmmm well the short version is
04:32 WarrTab Tablet is somewhat sub-optimal for coding
04:32 hmmmm for each setting there's a default setting and a user setting
04:32 hmmmm the default settings are set through Settings::setDefault() and they're all added in defaultsettings.cpp
04:32 hmmmm that goes into one std::map in the Settings object
04:33 hmmmm then there's a second std::map, with the actual user settings, that gets modified when the config file is loaded
04:33 WarrTab Ah, I see
04:33 hmmmm when you get a setting, it looks for the user setting first
04:33 WarrTab I had assumed you'd have 1 map
04:33 hmmmm if the user setting was not found (i.e. it wasn't specified in the config file) it falls back to the setting in the default map
04:34 WarrTab actually, no, it only makes sense to keep them separate
04:34 hmmmm if it's not found there, it either throws an exception or fails, depending on which variant of the function you use
04:34 WarrTab Or else you wouldn'y be able to save JUST the user settings
04:34 hmmmm yep
04:34 WarrTab That should make things simpler
04:34 WarrTab We could add a 3rd map, or just modify the defaults map
04:35 hmmmm latter
04:35 WarrTab Yeah, simpler
04:36 hmmmm hmm, Pilcrow_ was the one who had the freezing issue
04:38 hmmmm casmir was the one with the texture discoloration http://screenshot.ru/upload/images/201​5/03/23/screenshot_12123900148dad8.png
04:38 jin_xi joined #minetest-dev
04:39 Tablet_One joined #minetest-dev
04:39 hmmmm Electron_Libre was the one with the ^[combine problem:  http://pastebin.ubuntu.com/10654682/
04:39 hmmmm can't find the crash
04:40 hmmmm maybe I was making that one up
04:41 VanessaE ok casimir's thing is non-fancy leaves, I do remember seeing a comment about that
04:41 VanessaE but E_L's bug?  that's a new one on me.  I know those fonts work, I *just* tested them.
04:41 VanessaE (as in, a few mins ago when I said it worked)
04:42 VanessaE "hdf_53" indicates a signs_lib font file.
04:42 hmmmm the reason why I classify this as a clean_transparent bug is because it stopped when he disabled it
04:42 VanessaE I dunno
04:42 hmmmm oh yeah, and of course, the sign text problem
04:42 VanessaE this looks more like a serialization bug somewhere
04:42 hmmmm i forgot to list it
04:42 hmmmm well
04:42 VanessaE what sign text problem?  not that paste I guess?
04:42 hmmmm no
04:42 hmmmm the one where the text got all chopped up
04:43 hmmmm in any case, E_L's problem is readily reproducible
04:43 VanessaE oh you mean disappearing into single pixels etc?
04:43 hmmmm yeah
04:43 hmmmm next time he comes around I'll look at it closer
04:43 VanessaE oh that's since been fixed in warr's code
04:43 hmmmm ahh good
04:45 hmmmm hmm, off topic:  I wonder if the Player lock is necessary at all?
04:45 VanessaE ok now the REAL test.  HDX-256, with all those settings turned on, going to VE-Survival :D
04:45 hmmmm aren't Player modifications wrapped around Envlock?
04:45 hmmmm since they're held in the Environment
04:46 VanessaE mesh load delay is a real bitch here.....
04:46 VanessaE but other than that it seems to work fine
04:46 hmmmm everything is a massive bug we just haven't found yet
04:47 VanessaE heh
04:47 hmmmm I wonder what sort of stuff a Coverity scan would reveal
04:48 VanessaE hey!  wtf!  cheapie got rid of his big battery on Bridge Ave
04:50 VanessaE seriously though, I'm not seeing any problems.
04:50 hmmmm good stuff
04:52 WarrTab The fix for the fsst leaves issue would probably be similar to what we did with min_size
04:52 Tablet_One joined #minetest-dev
04:53 WarrTab Moving it lster in the pipeline would allow it to run *after* [noalpha
04:54 WarrTab Iirc pilcrow's freezing issue wasn't even minetest-related?
04:54 WarrTab Or was he ever able to repro it?
04:54 hmmmm shrug
04:54 WarrTab He had some horrid indexing service running in background that picked up a bunch of assets freshly downloaded into his MT cache.
04:55 paramat left #minetest-dev
04:55 hmmmm ahh
04:55 hmmmm okay, that's down to one problem then.  and he's not around so
04:55 WarrTab I think I'll tackle fast leaves before I get into enable-experimental
04:57 WarrTab I may have to do that tomorrow, though; I'm UTC -0400, and it's sleepy-time here...
05:04 WarrTab If 2536 is merged, should the pr be closed?
05:04 hmmmm yes
05:04 hmmmm hold on
05:04 Zeno` Houston I think we have a problem
05:05 Zeno` I seem to have entirely messed up my minetest repo lol
05:05 hmmmm whenever you close a PR because it was merged, paste the url of the commit as the comment
05:05 VanessaE hmmmm: you can also paste the commit ID.
05:05 VanessaE (github will link it)
05:07 hmmmm i like warr's github avatar
05:07 hmmmm it's like an angry nerd or something
05:10 Zeno` lol
05:17 Hunterz joined #minetest-dev
05:18 VanessaE heh
05:26 selat joined #minetest-dev
06:01 Zeno` Coming soon to Android, the blockbuster sequel to Angry Birds... Angry Nerds
06:03 VanessaE new!  throw pocket protectors, ink pens, and smart phones!
06:03 VanessaE :)
06:03 Zeno` but wait, there's more?
06:03 VanessaE hehe
06:07 selat joined #minetest-dev
06:21 Megaf_ joined #minetest-dev
06:26 ShadowNinja hmmmm: About the mailing list: A family matter poped up recently and therefore I've been especially busy.  I'll probably at least start with that by the end of the week though, if I remember.
06:26 ShadowNinja But I can't stay around now, so I'll be off momentarily.
06:35 hmmmm hey no prob
07:07 Calinou joined #minetest-dev
07:37 kilbith joined #minetest-dev
07:37 nrzkt joined #minetest-dev
07:38 nrzkt hey hmmmm: for the player lock copy it's simple to fix it. We need to add operator= to JMutex which does nothing, then the JMutex will no be copied and it's good
07:38 hmmmm i really wish people would stop overloading =
07:38 nrzkt i will do the modification, i already do the test on another branch because i worked with std::mutex which could not be copied because the compiler doesn't allow it
07:38 hmmmm nrzkt, what's wrong with the current fix?
07:39 nrzkt you only fix the player case, not all cases which could occur with JMutex usage
07:40 nrzkt the lock is a special case which need to overload the operator=, but i agree this should not be used too many times
07:40 hmmmm well mutexes aren't used in very many places
07:40 nrzkt i don't study each case, but in some cases it's right this is not necessary, and also the m_env lock is a pain which could be solved with fine grain locking on specific parts
07:40 hmmmm maybe there should be a macro we can insert into class definitions like DISALLOW_ASSIGNMENT_OPERATION()
07:41 nrzkt also, you talk about player memory leak. Why we keep the palyer in memory every time ? and don't disconnect it ?
07:42 hmmmm it's only a leak in an error condition
07:42 nrzkt it's very strange and that could explain the memleak pointer by VanessaE over time, if i  connect 10k players in 1 week, we have 10k Player in memory
07:43 nrzkt yes we re-use the Player object after, but it's not necessary to keep it in memory, proper disconnection with ressource desallocation must be done after a timeout, if we would keep the Player in a buffer for a time
07:43 nrzkt okay hmmmm
07:43 hmmmm #define DISALLOW_ASSIGNMENT_OPERATION(class) \
07:43 hmmmm hrm no
07:43 hmmmm #define DISALLOW_ASSIGNMENT_OPERATION(c) \
07:44 hmmmm c(const c&); \
07:44 hmmmm void operator=(const c&)
07:44 hmmmm and then in the class definition, under private, add DISALLOW_ASSIGNMENT_OPERATION(Player);
07:45 hmmmm then it should error any time somebody attempts to use it because it'd have a private access specifier
07:46 hmmmm but anyway I'm not so sure if Player actually needs a lock
07:46 hmmmm Players are part of the Environment, and the Environment is always locked while scripts are running with envlock
07:48 nrzkt i think we should keep this lock for fine grain locking, i don't thing we must keep this giant lock, if we could remove some usages in part of codes to provide a fine locking it could be good
07:49 hmmmm there's no way you can reduce the size of envlock without adding a lockless synchronization mechanism for map access
07:49 hmmmm i bet you're thinking that you could just move locks to each of the lua api, but this has been done before and it's slower than locking just once while script is executing
07:51 cd2 joined #minetest-dev
07:52 nrzkt i see
07:53 hmmmm ultimately, as part of client-side scripting, hud will theoretically be client-side only so we'd be able to remove that completely
08:50 eeew joined #minetest-dev
09:29 Megaf_ Folks, I have some serious issue with my server
09:29 Megaf_ it no longer compiles
09:29 Megaf_ since 0.12
09:30 Megaf_ take a look http://paste.debian.net/plain/164250
09:30 Megaf_ am I doing anything wrong?
09:33 kilbith -- Could NOT find Irrlicht (missing:  IRRLICHT_LIBRARY)
09:35 Megaf_ I think that is not the problem, I always compiled the server like that
09:36 Calinou -DBUILD_CLIENT=0 -DBUILD_SERVER=1
09:37 Calinou specify both, not just -DBUILD_CLIENT=0
09:37 Calinou else it compiles neither server nor client
09:37 Calinou it did this to me yesterday
09:38 Megaf_ Calinou: That's a new behaviour then
09:40 Megaf_ Calinou: Yep, that solved the issue, thanks
09:40 Calinou (-:
09:42 Megaf_ Calinou: https://github.com/minetest/minetest/commi​t/93fcab952b28f4db39d9326b83f14cbd86c0cf17
09:42 Megaf_ ShadowNinja: Why?
09:46 proller joined #minetest-dev
09:47 chchjesus joined #minetest-dev
10:01 err404 joined #minetest-dev
10:32 MinetestForFun joined #minetest-dev
10:54 Megaf_ kilbith: as you can see, the default cmake options changed, thats why I always enforce my cmake things, because I know that someday someone will just change the default options
11:11 Zeno` joined #minetest-dev
11:20 proller joined #minetest-dev
11:47 roniz joined #minetest-dev
11:51 ElectronLibre joined #minetest-dev
12:00 Zeno` #2588 needs review please
12:00 ShadowBot https://github.com/minetest/minetest/issues/2588 -- Move globals from main.cpp to more sane locations by Zeno-
12:04 nrzkt okay for me except the comment i send you on github, great job
12:07 Zeno` comment?
12:08 Zeno` oh, is that all?
12:08 nrzkt :)
12:08 Zeno` why did jenkins fail?
12:08 nrzkt http://jenkins.unix-experience.fr/job/minet​est-Android-OfficialRepository/403/console
12:09 Zeno` oh
12:09 nrzkt click on the link then console and you see
12:09 Zeno` I must have forgotten to add the new file
12:09 Zeno` I thought I did... /me checks
12:09 nrzkt hopefully jenkins/android see it :p
12:09 kilbith nrzkt, you should allow other devs to say "jenkins build please" for re-building if necessary
12:10 nrzkt the rebuild is done automaticly at each update on each pr
12:11 nrzkt and this is already the case
12:11 Zeno` hmm
12:12 Zeno` lol typo
12:17 ElectronLibre joined #minetest-dev
12:24 Zeno` jenkins is slow
12:24 Zeno` stupid butler
12:25 Zeno` hey what about this static analysis thing?
12:25 Zeno` CovertyScan
12:25 nrzkt only 1 core for each build it's a quad core machine and i have some other services
12:25 kilbith my client got crashed with an "ERROR: malformed packet read" just now
12:26 kilbith up-to-date with master :/
12:26 Zeno` seems like a LOT of high profile open source projects use it
12:26 nrzkt malformed packet read mean the algorithm want to read outside of the packet
12:26 nrzkt i need to change this stupid text to something more useful
12:26 Zeno` can that CovertyScan be added to Jenkins?
12:27 nrzkt tell me how to do it an maybe i can
12:27 kilbith nrzkt, in debug i see : 2015-03-31 21:17:29: ERROR[main]: Client::getMesh(): Mesh not found: "dddddddddddddddddddddddddddddddddddddd
12:27 nrzkt strange mesh
12:27 kilbith eh, no
12:27 kilbith it was crashed on "ERROR: malformed packet read"
12:27 kilbith and there was no mesh around anyways
12:28 Zeno` hmmm, might only "integrate" with github
12:28 Zeno` https://scan.coverity.com/
12:29 Zeno` https://scan.coverity.com/github and https://scan.coverity.com/travis_ci
12:29 nrzkt do you try it ? i can register minetest to test
12:29 Zeno` I haven't tried it yet
12:30 Zeno` It's another tool
12:30 Zeno` the more tools the better IMO
12:30 kilbith OMG big fail : http://pastie.org/10067154
12:31 kilbith (when i approached the signs_lib)
12:31 nrzkt it's not the first time i see a problem with signs_lib
12:32 kilbith in the manner on how the engine with the signs_lib, it's not inherent to signs_lib
12:32 nrzkt i don't know how does this work but this generate some errors. On which server are you ?
12:32 proller joined #minetest-dev
12:32 kilbith there has been never those errors with the signs_lib since recently
12:32 kilbith how the engine deals*
12:33 Zeno` nrzkt, I don't think jenkins is starting the android test build again
12:33 kilbith nrzkt, Megaf server (-12, 11, -165)
12:34 nrzkt Zeno`, jenkins is busy on my branch at this moment, will happen soon
12:34 nrzkt http://jenkins.unix-experience.fr/
12:34 Zeno` jenkins is a snob
12:34 nrzkt the FreeBSD worker is busy and also the Debian/Android builder
12:34 nrzkt if you have a server we could add one more builder :p
12:34 Zeno` just 'cause he is head butler he thinks he can pick and choose
12:35 nrzkt it only need an access to server and openjdk :p
12:35 Zeno` I'll setup an account on VanessaE's server. VanessaE can you give me a login please?
12:35 selat joined #minetest-dev
12:36 nrzkt kilbith i will change a little bit the error message to have more informations, maybe having the offset and the packet lenght could be good
12:36 Zeno` selat, your timer PR...
12:36 Zeno` is that still something you want merged?
12:37 Zeno` #1188
12:37 ShadowBot https://github.com/minetest/minetest/issues/1188 -- getTime refactoring by Selat
12:37 kilbith nrzkt, join the server at coords i gave you and see by yourself please
12:37 nrzkt yes, just a minute i'm at work
12:38 kilbith it happened the same thing on Zenith server, with the signs_lib
12:38 kilbith never had this crash few months ago
12:40 selat Zeno`: Probably yes... As I undestand it I should move all functions into a static class?
12:40 nrzkt many things has changed since month ago :(
12:44 nrzkt does this crash the client or only trigger an exception ?
12:44 Zeno` a static class...
12:45 Zeno` hmm
12:47 Zeno` What's the advantage of it being a static class (object of)
12:47 nrzkt ok i will push the throw message improvement
12:47 kilbith 2015-04-01 14:29:31: ERROR[main]: ClientEnvironment::addActiveObject(): id=1140 type=e: SerializationError in initialize(): deSerializeLongString: size not read:
12:48 kilbith ^ clearly a problem with an handler in the network code
12:49 nrzkt it's the same error
12:49 nrzkt you read outside packet and this trigger the addActiveObject exception
12:50 Megaf_ kilbith: weird, wasnt it working on my server?
12:50 kilbith yep
12:50 kilbith but on every server as well
12:50 nrzkt pushing the error precisions on master now
12:50 kilbith since the new networking code has been pushed....
12:51 nrzkt could you recompile with the new commit ? could be good to have more precisions on the error
12:51 kilbith ok
12:51 nrzkt we are reading outside of the packet size, it's not normal
12:52 Zeno` selat, do you mind if when I am testing it tomorrow I change some things?
12:52 Zeno` I'll leave you as author of course
12:52 nrzkt Zeno` will you push #2588 ?
12:52 ShadowBot https://github.com/minetest/minetest/issues/2588 -- Move globals from main.cpp to more sane locations by Zeno-
12:53 Zeno` nrzkt, if I get approval :)
12:53 Zeno` oh, jenkins finally got out of bed
12:53 nrzkt i approve if you update the copyright notice date on the new file :p
12:53 Zeno` nrzkt, the objects being in main.cpp made no sense at all
12:54 Zeno` I did update it didn't I?
12:54 nrzkt oh yes sorry
12:54 nrzkt outdated commit :p
12:54 nrzkt then it's okay
12:54 Zeno` Ok, I'll merge in 15 minutes or so then :)
12:55 nrzkt and objects declared in main.cpp as globals are bad, except the runners
12:56 Zeno` I'd like to get rid of more of the globals (99% of them can be removed) but it's a much larger job
13:00 selat Zeno`: Of course that's not a problem :)
13:01 Zeno` selat, ok I'll look at it tomorrow
13:02 Zeno` it just came up while I was looking at the silly globals in main.cpp :)
13:02 Zeno` so may as well tackle it at the same time
13:11 kilbith nrzkt, the big crash happened again after pulled master : http://pastie.org/10067211
13:11 kilbith exactly the same issue than : https://github.com/minetest/minetest/issues/2541
13:11 kilbith #2541
13:12 ShadowBot https://github.com/minetest/minetest/issues/2541 -- Singleplayer World Crashes - Malformed Packet Read, Server Running Different Version of Minetest.
13:12 nrzkt you don't have the serialization error here :s
13:13 nrzkt the new string is Reading outside packet (offset: <offset>, packet size: <size>)
13:14 * Zeno` doesn't feel too good
13:14 Zeno` kilbith, how to reproduce that bug?
13:15 kilbith Zeno`, when you approach some signs_lib
13:15 Zeno` any server?
13:15 kilbith any 0.4.12-dev server
13:15 nrzkt this is strange only sign lib is the problem if it's active object
13:15 nrzkt no problem with 0.4.12-stable and lesser versions ?
13:15 kilbith signs_lib is an entity
13:15 nrzkt entity, sorry :)
13:16 kilbith no problem with stable since it has not your big network rewrite at all
13:16 nrzkt what are the packet exchanged for entities ? someone knows ?
13:17 nrzkt and why only signs_lib, it's not the only entity in game, why only this mod
13:17 nrzkt entity = server active object
13:17 nrzkt LuaEntitySAO
13:17 SopaXorzTaker joined #minetest-dev
13:18 kilbith you should know that as network maintainer...
13:18 nrzkt i don't know the whole code, i'm only the tube maintainer, everybody can do what they want with the tube :)
13:18 SopaXorzTaker joined #minetest-dev
13:21 nrzkt do you have coordinates on Zenith server i don't retrieve it on this server and signs are working well atm
13:24 nrzkt Zeno` is your fast move fix correct ?
13:25 nrzkt i'm not able to be in fast_move on Zenith server whereas i have rights
13:25 Zeno` nrzkt, it works for me
13:26 Zeno` you do still need to toggle fast move on
13:26 nrzkt i know
13:26 nrzkt and this doesn't work
13:26 kilbith idk, it seems maximize the chance of crash
13:26 Zeno` weird. did on all the servers I tried
13:26 kilbith because of text-entity loading
13:27 kilbith that doesn't happen everytime but LazyJ can confirm you as well
13:27 kilbith ping him on #minetest
13:28 nrzkt i think it's a problem server side, not client side, packet is processed exactly same on client
13:28 nrzkt if packet is not complete the only problem can be from server
13:33 Zeno` nrzkt, you did remember to upload the new apk didn't you? :P
13:34 nrzkt upload where ? on github ?
13:34 Zeno` I mean to test the auto fast
13:34 Zeno` to your android
13:36 Megaf_ kilbith: nrzkt Zeno`
13:36 Megaf_ I just crashed
13:36 Megaf_ serialization, malformed packet read
13:36 Megaf_ 2015-04-01 14:35:55: ERROR[main]: A serialization error occurred:
13:36 Megaf_ 2015-04-01 14:35:55: ERROR[main]: Malformed packet read
13:36 Megaf_ 2015-04-01 14:35:55: ERROR[main]: The server is probably  running a different version of Minetest.
13:37 kilbith ^ ^ ^
13:37 Megaf_ I was flying alonside nrzkt when it crashed
13:37 Megaf_ I believe I have the same git version on client and server
13:37 nrzkt can you update with last comit too to get more information ?
13:38 nrzkt commit*
13:38 Megaf_ actually I crashed as soon as I teleported to spawn and attempted to move
13:38 Megaf_ nrzkt: server and client or just client?
13:38 Zeno` I will type /spawn
13:39 nrzkt just client
13:39 Megaf_ ok
13:41 Megaf_ downloading the latest commit
13:41 Megaf_ This one? https://github.com/minetest/minetest/commi​t/9527984dbcfc0a6cc7aa0470430cb6c3aa4103ba
13:42 nrzkt yes
13:42 nrzkt the Zeno` commit is good, previous is the useful commit
13:43 Zeno` ?
13:44 nrzkt i talk to Megaf :p
13:44 Zeno` Megaf_, teleport me to spawn hehe
13:44 Megaf_ compiling
13:45 kilbith Zeno`, /spawn
13:47 kilbith another one : http://pastie.org/10067288
13:47 Zeno` kilbith, it's not working. You have no idea how painful it is trying to play when running under valgrind hee
13:47 Megaf_ Zeno`: hold on
13:47 nrzkt everytime we see the [combine
13:47 nrzkt it's very strange this particularity
13:47 Megaf_ Zeno`: you are at spawn
13:48 Zeno` thanks
13:48 kilbith Zeno`, problem is that doesn't occurs everytime for catching it easily
13:48 kilbith but it exists for sure
13:49 kilbith Megaf_, show to Zeno` the optimal condition to reproduce the bug
13:49 kilbith ie. start at spawn and then run to the south to the bunch of signs
13:50 kilbith it happened on 3 differents servers to me
13:51 Zeno` I wonder if it's another gettext issue
13:51 nrzkt a server side gettext ?
13:51 kilbith no, not related at all
13:51 kilbith can't see how
13:51 nrzkt becase we are decoding some strings
13:52 nrzkt but i don't see why gettext could be involved in signs lib
13:52 kilbith which strings ?
13:52 Megaf_ hold on
13:52 Megaf_ my system is recovering from minetest
13:52 kilbith *cough* the letters on sign_libs are images
13:53 Megaf_ I found a way to reproduce the crash
13:53 Megaf_ turn on fly and noclip
13:53 Megaf_ fly above the clouds
13:53 Megaf_ then flight straigh up
13:53 Megaf_ do a /spawn
13:53 Megaf_ go forward
13:53 Megaf_ crash
13:54 Megaf_ the second time I actually had to kill minetest because it was using more than 2 GB of RAM and 1 GB of SWAP
13:54 kilbith Zeno`, nrzkt ^
13:55 kilbith Megaf_, grant fly and noclip to Zeno please
13:55 Megaf_ kilbith: I believe he has alrady
13:55 Megaf_ kilbith: /privs zeno there
13:55 Megaf_ by the way, I have all stuff disabled
13:56 kilbith he's not currently following your way
13:57 Megaf_ Zeno`: gettext has nothing to do with the crash
13:57 Megaf_ Zeno`: I have no gettext
13:58 Megaf_ my compilation options  -DRUN_IN_PLACE=1 -DENABLE_REDIS=0 -DENABLE_LEVELDB=0 -DENABLE_FREETYPE=1 -DUSE_FREETYPE=1 -DBUILD_SERVER=0
14:00 kilbith Zeno`, crash just now
14:00 kilbith when you have followed me to those signs
14:01 hmmmm joined #minetest-dev
14:01 kilbith for the 3rd time
14:02 kilbith http://pastie.org/10067319
14:03 Megaf_ well, mine is not crashing anymore
14:03 hmmmm oh cool you have the same bug as electron_libre
14:03 kilbith and as LazyJ as well
14:03 hmmmm kilbith:  just try setting enable_clean_transparent = false
14:03 kilbith hmmmm, #2541
14:03 ShadowBot https://github.com/minetest/minetest/issues/2541 -- Singleplayer World Crashes - Malformed Packet Read, Server Running Different Version of Minetest.
14:03 kilbith ^ same issue
14:04 kilbith Megaf_ can confirm as well
14:07 Megaf_ crashed
14:07 Megaf_ kilbith: Zeno` nrzkt http://paste.debian.net/plain/164295
14:08 Megaf_ if you keep running to a lot of places
14:09 Megaf_ run 5 seconds in one direction, then 5 seconds to another, then /spawn and move forward
14:11 Megaf_ crashed again
14:11 Megaf_ easy to reproduce
14:11 Megaf_ ^ kilbith
14:12 Megaf_ kilbith: why do you think it has to do with signs?
14:13 Zeno` very odd
14:13 Megaf_ I can't find the relation
14:13 kilbith Megaf_, not all all everything was fine before
14:13 kilbith like 2 months ago
14:13 kilbith not at all*
14:24 kilbith hmmmm, the crash occurs the same way with enable_clean_transparent = false, thus it's unrelated
14:24 hmmmm the reason why I asked is because I told E_L to do the same thing and it worked.
14:24 hmmmm i realize it is farfetched that the two things might be related but stranger things have happened
14:48 ShadowNinja Megaf_: Because if you build the client you don't need a seperate server, and building the server currently rebuilds all of the common files for the server, so it almost doubles compilation time.
15:07 kilbith the 0.5-dev branch should be closed because currently some admins use it in thinking this is the most advanced dev branch... :/
15:08 nrzkt some things must be backported to master before closing
15:13 MinetestForFun joined #minetest-dev
15:29 hmmmm instead of calling it 0.5-dev, why not make it representative of what the branch has that master doesn't
15:43 kilbith joined #minetest-dev
15:48 est31 joined #minetest-dev
15:48 fz72 joined #minetest-dev
15:54 SopaXorzTaker joined #minetest-dev
15:59 srifqi joined #minetest-dev
16:18 jin_xi joined #minetest-dev
16:30 Megaf_ ShadowNinja: Thanks, I agree.
16:50 JZTech101 joined #minetest-dev
16:52 Calinou joined #minetest-dev
16:58 leat joined #minetest-dev
17:15 Krock joined #minetest-dev
17:25 ElectronLibre joined #minetest-dev
17:29 ShadowNinja hmmmm: lists.minetest.net is up.  What lists should we use, how open should they be?
17:29 roniz_ joined #minetest-dev
17:30 neoascetic joined #minetest-dev
17:32 Calinou minetest, minetest-dev
17:32 Calinou these?
17:35 Krock minetest\src\mapgen.cpp(521): error C3861: "strcasecmp": Identifier was not found. Oh well.
17:41 Krock *creates pull request for 1 new line*
17:48 Tablet_One joined #minetest-dev
17:54 rickmcfarley joined #minetest-dev
17:57 proller joined #minetest-dev
18:17 Krock2 joined #minetest-dev
18:51 proller joined #minetest-dev
18:54 Robert_Zenz joined #minetest-dev
19:33 err404 joined #minetest-dev
19:33 paramat joined #minetest-dev
19:39 Robert_Zenz joined #minetest-dev
19:40 SudoAptGetPlay joined #minetest-dev
19:41 paramat will push #2580 once rebased and checks finish
19:41 ShadowBot https://github.com/minetest/minetest/issues/2580 -- Treegen.cpp: Add pine tree. Force place trunks. Speed optimise leaves blits by paramat
19:47 leat joined #minetest-dev
19:57 rickmcfarley joined #minetest-dev
19:57 leat joined #minetest-dev
20:21 Tablet_One joined #minetest-dev
20:33 Tablet_One joined #minetest-dev
20:40 ElectronLibre left #minetest-dev
21:12 FR^2 joined #minetest-dev
21:14 OldCoder joined #minetest-dev
21:15 DFeniks joined #minetest-dev
21:19 err404 joined #minetest-dev
21:25 Fritigern joined #minetest-dev
21:42 proller joined #minetest-dev
21:52 paramat now pushing #2580
21:52 ShadowBot https://github.com/minetest/minetest/issues/2580 -- Treegen.cpp: Add pine tree. Force place trunks. Speed optimise leaves blits by paramat
21:59 paramat done
22:01 VanessaE you should add something similar for L-systems if it doesn't already have that feature
22:01 VanessaE (an optional setting, that is)
22:03 paramat optional setting for what?
22:04 VanessaE for forcing the placement of trunks.
22:04 VanessaE (like with the regular code, in l-systems, leaves will cut through a trunk if the trunk is placed later, e.g. by a second tree)
22:05 VanessaE (of course you'd only want to force it when placing trees during a mapgen cycle, not when growing from a sapling later on)
22:05 paramat yup force placement is for all trees =)
22:06 VanessaE OH!
22:06 VanessaE but is there a way to turn it off?
22:06 VanessaE *looks at code*
22:10 paramat oh hang on, force placement is not for l-systems, sorry
22:10 VanessaE ok
22:10 VanessaE however you did do a bit of reformatting of the l-systems code
22:10 paramat yeah 'a bit' 220 lines
22:10 VanessaE I was gonna say, I don't see anything in here that jumps out at me :)
22:11 VanessaE (inb4 proll complains of same ;) )
22:11 paramat i realised last night the ugly l-system code was RBA's not c55's
22:12 VanessaE heh
22:12 paramat force placement is for the normal 3 trees, that's what i meant
22:12 VanessaE right, I gathered.
22:19 Lunatrius` joined #minetest-dev
22:51 proller joined #minetest-dev
23:07 est31 joined #minetest-dev
23:34 VargaD joined #minetest-dev
23:51 Sokomine does anyone have any ideas why mt crashes on me when i switch to singleplayer? it's latest git. multiplayer works just fine. it also seems to depend a bit on the game - minimal works, minetest_game does not. carbone, again, works. havn't found a pattern yet
23:55 est31 do you have a stacktrace Sokomine?
23:57 Sokomine nothing helpful i'm afraid...
23:58 est31 and you can reliably reproduce the crashes?
23:58 est31 I often have random crashes, either while playing or directly after starting the binary
23:59 Sokomine got it: those that work are games that have a menu/ folder with textures for the main menu next to their mods/ folder

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