Time Nick Message 00:23 VanessaE hmmmm: any progress on the VBO rebase? [fsjal.gif] 00:33 iqualfragile hmmmm: great idea, we should have hats on a regular basis 00:35 ShadowNinja hmmmm: Oh, it uses a seperate texture BTW, so it can be compatible with custon skins. 01:06 hmmmm VanessaE, I'm going to be doing that soon 01:06 hmmmm Soon (tm) 01:06 hmmmm along with the player list 01:06 hmmmm ShadowNinja, good, in that case I'll add it 01:06 hmmmm looks like I have some work to do. 01:06 hmmmm but later, I need to take a nap 01:06 VanessaE hmmmm: ok 01:07 hmmmm guys, I don't think we should have a feature freeze tomorrow 01:07 VanessaE [okay.png] 01:07 hmmmm we should adjust the length of feature freezes based on the amount of development activity perhaps 01:07 hmmmm just a week long each and every time is kinda dumb 01:07 hmmmm amount of new features really. so it should be appropriately shorter for this release because the majority of items have been making things more stable 01:08 hmmmm it's to be determined, but I feel like 2 less days is good 01:08 hmmmm so it'll enter into feature freeze on the 20th, sounds good? 01:08 VanessaE yeh 01:12 VanessaE before this freeze: 01:12 VanessaE can we please get two things done: 01:13 VanessaE 1) All model files being used in Minetest (which for now is just Sam I guess) get re-compressed into .obj or .b3d or some other format instead of .x. 01:13 VanessaE 2) Sam II becomes the standard skin. 01:18 ShadowNinja Why? https://github.com/minetest/minetest_game/blob/master/mods/default/functions.lua#L106 01:20 VanessaE ? 01:23 kaeza *globals* ;_; 01:27 ShadowNinja ^ And they do nothing. 02:10 VanessaE what are the rules regarding plain, ordinary, unadorned files served up by remote_media? 02:10 VanessaE as long as the file exists and is in the path, it's sent to the clients, right? 02:11 VanessaE e.g. this directory full of files, is there anything amiss about this? http://minetest.digitalaudioconcepts.com/creative-survival-media/ 02:44 VanessaE [12-17 21:43] 20:34:42: VERBOSE[main]: Client: Requesting remote media file "default_grass_1.png" "http://minetest.digitalaudioconcepts.com/creative-survival-media/default_grass_1.png" 02:44 VanessaE [12-17 21:43] 20:34:43: INFO[main]: Client: Received media file 297b55b087c6f90a2e71c36ffe30b05b660d083e "default_grass_1.png" mismatches actual checksum 9099a17682a8d52464300b3160b9a41e63f7e846 02:44 VanessaE um, wut? 02:45 VanessaE this is after the user cleared his cache. 02:53 ShadowNinja So can we agree to SamII(+cape)? It's been waiting far too long, something so simple should have been merged within the week. 02:54 VanessaE ShadowNinja: I'd much prefer stu's split-limb model, recompressed into .obj or .b3d or something. it'll be a smaller filesize than than the current model, and look nicer too. 02:55 VanessaE combine that with Sam II and everyone will love it. 02:55 VanessaE I've been using that for months on my server with no issues. 02:55 ShadowNinja VanessaE: Screenie and link? 02:55 VanessaE sec. 02:56 VanessaE https://forum.minetest.net/viewtopic.php?pid=86858#p86858 02:56 VanessaE the original .blend file is inside the .zip linked in the first post, I believe. 02:56 VanessaE from there, an appropriate other-than-.x file can be generated, I guess? 02:57 VanessaE I use the wield3d version of this model, but I don't think that is suitable for minetest (it's glitchy) 02:58 rambomedic how do you view .x files? blender? 02:59 VanessaE I think so, never tried to load them with it. 03:02 ShadowNinja VanessaE: That's incompatible with capes. :-| Maybe if they were merged. 03:02 rambomedic ShadowNinja: Link to samII? want to see 03:02 rambomedic and what does the current sam look like 03:03 VanessaE well no-one says one couldn't modify that ^^^ model to take a cape :P 03:03 VanessaE https://forum.minetest.net/viewtopic.php?id=6423 03:03 VanessaE Sam II ^^^^ 03:03 ShadowNinja #201 03:04 ShadowBot https://github.com/minetest/minetest_game/issues/201 03:06 rambomedic I don't like how Sam II has a mullet lol 03:09 VanessaE heh 03:19 rambomedic is RealBadAngel here 03:20 rambomedic I have some serious bug reports for his shader/parallax occlussion patch 03:20 VanessaE sometimes he is, just not at the moment 03:20 rambomedic Should I just file a github bug report 04:26 * ShadowNinja waves around https://gist.github.com/ShadowNinja/7953481 04:44 VanessaE HAY! put that away! there might be be kids present. 04:55 * us^0gb covers the eyes of his little sister 05:03 VanessaE regarding the error I posted where some users are getting consistent failures at receiving some files due to failed checksums: I wget'd several of the "failed" files from my web server and diff'ed them against the ones in minetest_game. diff says they are identical. What do I do now? 05:03 VanessaE (it always seems to be the same set of files) 05:04 VanessaE permissions are good, obviously, if I can wget them from the same base URL that the other server's media files are coming from (see above) 05:30 VanessaE found the issue. 05:30 VanessaE it was a bug in my game/server buildbot 06:27 VanessaE had to teach my buildbot roughly the order in which the server loads/serves up the media files. Sorry for the false report. :P 10:41 VanessaE So, about remote_media... I've just gotten my servers configured to the pointed where not even one single file is served up by media_FUBAR (my name for the default code ;) ).... can we please now discuss seriously about rejecting non-cURL-capable clients? 10:41 VanessaE s/pointed/point/ 10:43 VanessaE there are far too many clients out there whose builds are not cURL capable, I simply don't care about those. They are in the minority, and they are easily updated. If we reject building a client without cURL and reject connections to a server without cURL in the client, then eventually there'll be nothing left for that old code to do but handle legacy situations like if the remote_media server simply lacks a file to send. 10:43 VanessaE (e.g. some routine mod update doesn't get echoed through to the remote_media server) 10:44 Megaf kahrl, darkrose 10:44 Megaf # Maximum number of statically stored objects in a block 10:44 Megaf #max_objects_per_block = 49 10:44 Megaf What's the performance impact of that? 10:44 VanessaE hm, that didn't really come out right... "far too many" "in the minority".. non-sequitur. Anyways.. you get the point. 10:45 VanessaE I have that set to 500, Megaf. no apparent performance issue with it, but don't change it if you don't see any errors about "forcing delete" of "too many objects" in a mapblock. 10:45 VanessaE (my server needs it because pipeworks) 10:53 VanessaE could it ever be possible to "throttle" mods to consume less than a given amount of CPU (at the cost of them running visibly slower)? 10:55 proller it was already... ;) 10:56 VanessaE I mean on a configurable basis, as in I tell Minetest "all mods under the 'game' heading must, together, take no more than X% of the CPU, all mods combined under worldmods are allocated Y%", etc. 10:56 VanessaE or just some basic thing like that. 10:56 VanessaE or even "all Lua code executed by the Minetest engine, outside of the stuff included in builtin, shall take less than X% CPU" for whatever value of X I tell it. 10:57 VanessaE any leftover CPU would be usable by the engine or just left to idle. 10:57 Exio4 that is how proller made freeminer "less laggy" 10:58 Exio4 it is just making the mods lag for not lagging other things. 10:58 VanessaE better the mods should lag than the engine 10:59 VanessaE (with the exception of things that open a formspec) 11:16 proller tests broken!!1 11:16 proller http://paste.org.ru/?ofkze4 11:17 proller --disable-unittests okay 12:23 Megaf Folks, minetestserver is using only 20% of CPU and tha bw use is also low, only 12 KB/s 12:23 Megaf and client side still laggy and slow 12:23 Megaf why? 12:56 iqualfragile will the version after 0.4.9 be called 0.4.10 or 0.5.0? 12:56 iqualfragile i assume 0.4.10 13:12 specing proller: ... fixed in freeminer? 13:53 proller https://play.google.com/store/apps/details?id=com.vicono.worldcraft 14:04 specing As far as I see right now, amount of generated ore does not increase the deeper you dig 14:04 specing Is there a way to make it depth-dependent? (I'd like a mining rainbow at -30000) 14:34 Megaf /home/minetest/Minetest/_New/minetest-mapper-cpp/TileGenerator.cpp:13:22: fatal error: gdfontmb.h: No such file or directory 14:34 Megaf hm 14:36 Megaf libgd was missing 14:37 thexyz this is the true potential of minetest 14:37 thexyz a very important thing which was ignored by the devs 14:37 thexyz and that's what we get now 14:38 thexyz how many downloads are there already? 14:38 thexyz ah 100k+ 14:39 thexyz the port isn't even a quality one 15:07 kaeza Trivial patch: https://github.com/minetest/minetest/pull/1060 15:30 thexyz kaeza: this breaks everything, right? 15:43 kaeza eh? 15:43 kaeza it's only used by player:hud_change() 15:44 thexyz ah right 16:07 ShadowNinja Well, now they have four infringing apps. Although I've only checked buildcraft and worldcraft I assume they all use a single libMine.so. 16:08 thexyz yes they seem to do just that 16:08 thexyz did you `diff` it though? 16:08 ShadowNinja They're the same size. One sec... 16:12 ShadowNinja Files com.vicono.build-1/libMine.so and com.vicono.worldcraft-1/libMine.so differ 16:17 ShadowNinja Uh, there's an update for buildcraft. 16:18 ShadowNinja Still different, but the same size. 16:19 Jordach i'd suggest dropping a DMCA bomb 16:21 ShadowNinja Already done, but before this one came out. (Zombies seems new too) 16:21 ShadowNinja We have to wait on Google now. 16:22 ShadowNinja celeron55: Perhaps you should add these new apps ^. 16:23 celeron55 16:37:47 <+thexyz> a very important thing which was ignored by the devs 16:24 celeron55 don't blame me, i don't use android 16:24 sfan5 you have an old nokia phone, right? 16:24 celeron55 of course 16:25 celeron55 but i'm surprised that those guys do it and nobody internal to this community 16:26 celeron55 i know at least sfan5 has tried, and that thexyz is doing something now, so maybe there's a chance to catch up 16:26 sfan5 I tried, that's right.. I need a better device to get further first 16:27 proller Assertion failed: (mutex_lock_retval == 0), function Lock, file /usr/home/proller/minetest_test/src/jthread/pthread/jmutex.cpp, line 47. 16:27 ShadowNinja I have(and in fact am currently using) a android device. But I don't really like android dev. (Java!) 16:27 sfan5 ShadowNinja: no java involved 16:28 thexyz no, java is involved 16:28 ShadowNinja sfan5: Well we need Java to have a UI. 16:28 sfan5 yeah.. ok you're right 16:29 thexyz no, not for ui 16:29 celeron55 aanyway... 16:30 celeron55 i now recommend that people who have LGPL C++ code in the engine use this form https://support.google.com/legal/contact/lr_dmca?product=googleplay to send something like this: http://c55.me/random/2013-12/dmca-2013-12-11_20-59-07.png 16:30 celeron55 google hasn't responded to my notice so maybe they care if we have like 5 of these 16:31 celeron55 just tack every minetest-based app they have uploaded (links to google play pages) to the last part in that image 16:31 celeron55 (at the time i sent that there were only two) 16:32 specing google gets several tens of millions of DMCAs per month, no wonder they haven't responded 17:09 Jordach can i grab a bunch of links in a random pastebin 17:10 kahrl if the dmca gets through, does the money earned from advertising on the android app go to us? 17:10 kahrl like the content id system on youtube? 17:11 Jordach it'll usually be pulled down and destroyed 17:15 celeron55 kahrl: ehm... i wouldn't think so 17:15 celeron55 i don't really believe that can be taken down at all in the first place; google is just hopeless in handling stuff like this in a meaningful way 17:15 ShadowNinja kahrl: I think not, Google doesn't have as much control over Android advertising as it does over YouTube. 17:50 VanessaE just wheat and cotton 17:55 VanessaE papyrus, maybe. jungle grass has been proposed before. cacti? no. 17:55 PilzAdam grass is handled by the flowers mod; and papyrus and cacti by default 18:00 VanessaE um, wut? 18:00 VanessaE terminate called after throwing an instance of 'ServerError' 18:00 VanessaE what(): ServerError: LuaError: allow_metadata_inventory_put should return a number 18:00 VanessaE stack traceback: 18:00 VanessaE Program received signal SIGABRT, Aborted. 18:00 celeron55 oh one of those "lol it's like you'd have any use for a traceback" errors 18:02 VanessaE I've got this loaded up under gdb at the moment and I *THINK* it's a debug build. want anything from it? 18:03 PilzAdam a Lua error is not traceable in gdb AFAIK 18:03 VanessaE PilzAdam: no, perhaps not, but maybe the cause of the engine failing to produce the backtrace is :) 18:03 PilzAdam well then do bt all 18:04 VanessaE http://pastebin.ubuntu.com/6595514/ 18:04 ShadowNinja It's because the stack has unwound by the time the LuaError is thrown. 18:08 VanessaE need anything else from it before I restart it? 18:09 ShadowNinja You're using non-JIT Lua? 18:09 VanessaE JIT. 18:10 ShadowNinja Hmmm, usually it eats the error message and replaces it with "C++ exception". 18:13 celeron55 ShadowNinja: it depends on which way the call goes 18:13 celeron55 ShadowNinja: that always happens if it's lua calling C++ 18:13 celeron55 but what happens here is C++ calls Lua and gets a bad result 18:13 ShadowNinja Ah, I see. 18:14 ShadowNinja Then it *should* have a traceback, I remember adding a errorhandler there. 18:17 celeron55 lua isn't returning an error but just a value that C++ doesn't expect 18:18 ShadowNinja Hmmm, see the bottom of this: http://luajit.org/ext_c_api.html 18:18 celeron55 and no data about who has registered that callback is stored in minetest... umm... well actually, just printing out what node name it belongs to would be all the help someone would need 18:19 celeron55 ShadowNinja: that's not relevant here 18:20 ShadowNinja celeron55: What do you mean? Doing that should fix the error message eating. (With a slight performance hit) 18:21 celeron55 no exceptions are happening in VanessaE's error 18:22 ShadowNinja Oh, relevant to Vanessa's error. 18:22 VanessaE leave it to me to find the obscure crashes ;) 18:25 PilzAdam VanessaE, its not really obscure, just a mod which cant read the docs 18:26 VanessaE PilzAdam: non-sequitur. mod crashes -> server produces a backtrace and dies. end of discussion. 18:28 kahrl VanessaE: what do you mean? the backtrace you have in your paste is perfectly accurate 18:28 VanessaE kahrl: um, it doesn't identify the mod and file that crashed. 18:29 kahrl the error happened after the lua function had already returned 18:29 kahrl so the backtrace must by definition be empty 18:39 celeron55 it should log the node name though 18:40 celeron55 that way debugging it becomes easy 18:42 kahrl already working on it 18:44 kahrl https://gist.github.com/kahrl/8027566 19:10 ShadowNinja kahrl: Looks good. Although it looks like you didn't add messages to all of them. 19:10 kahrl the lower ones were only scriptError() 19:10 kahrl I changed them for symmetry 19:12 ShadowNinja kahrl: How does https://gist.github.com/ShadowNinja/7953481 look? 19:15 kahrl ShadowNinja: code looks fine 19:15 kahrl I don't have an opinion on whether it should be added to the API 19:17 kahrl } else { root = Json::nullValue; 19:17 kahrl shouldn't this raise an error instead for types like LUA_TFUNCTION? 19:19 ShadowNinja Hmmm, yes. I'll change that. 19:19 ShadowNinja Does LuaJIT define something so we can know if we are compiled against it as opposed to regular Lua? 19:22 kahrl maybe #ifdef LUA_JDIR 19:25 kahrl what should be more future-proof though is adding USE_LUAJIT to config.h 19:30 * ShadowNinja watches EVERYTHING re-compile after adding CMAKE_USE_LUAJIT. 19:32 VanessaE you ran without LuaJIT before? O.o; 19:33 ShadowNinja VanessaE: No, I added a #define so that some code is only compiled when LuaJIT is enabled. 19:35 ShadowNinja \o/ It works! The only issue is that it prints "LuaError: " twice. 19:36 ShadowNinja How does this look kahrl? http://ix.io/9rq 19:38 kahrl catch (...) seems a bit unsafe... 19:38 ShadowNinja It has a slight performance hit unfourtunately. 19:40 ShadowNinja I copied that from the example. The previous two should catch just about anything anyway. 19:40 kahrl iirc this code even catches access violations on windows 19:40 kahrl and std::bad_alloc obviously 19:40 kahrl I might be misremembering 19:41 ShadowNinja Maybe just catch ServerErrors? 19:42 ShadowNinja Or even just LuaErrors. 19:43 kahrl that sounds better, but I don't know which exception classes can occur 19:44 ShadowNinja Shouldn't LuaErrors be derived from ServerErrors rather than directly from std::exception? 19:44 kahrl hmm, yes 19:46 kahrl I wonder if there's any place in minetest that wants to catch ServerErrors but not LuaErrors 19:50 proller http://paste.org.ru/?ofkze4 19:51 proller something wrong with new threads 19:53 kahrl proller: is that in fm? 19:53 proller in fresh git mt too 19:53 proller paster from mt 20:00 kahrl I don't see anything that could cause this in the SimpleThread->JThread commit 20:21 artur99 who is Mitchell Ward 20:22 artur99 ? 20:22 artur99 !lastseen oldcoder 20:22 rubenwardy That pinged me :( 20:23 rubenwardy I set it so "ward" pings me, and "ruben" does too 20:27 ShadowNinja artur99: ,,(seen OldCoder) 20:27 ShadowBot ShadowNinja: OldCoder was last seen in #minetest-dev 2 days, 22 hours, 7 minutes, and 9 seconds ago: VanessaE, may I PM? 20:27 artur99 thanks 20:27 ShadowNinja Well this balooned: https://gist.github.com/ShadowNinja/8029329 What do you think kahrl? 20:28 kahrl wow 20:29 specing What hash does minetest use for passwords? 20:29 specing It looks kinda ... short 20:29 ShadowNinja specing: SHA1 salted with the player name 20:29 kahrl ahh 20:29 kahrl don't include server.h in c_types.h 20:29 * kahrl weeps for the compile times 20:30 ShadowNinja kahrl: Yes, that didn't seem good, but how else am I to get ServerError? 20:30 kahrl dunno, maybe move ServerError (and LuaError?) to exceptions.h 20:30 ShadowNinja Hmmm, that would work... 20:31 kahrl actually not LuaError since lua.h shouldn't be included everywhere 20:32 artur99 so, does anybody know who is Mitchell Ward? He made some errors in translation 20:32 kahrl instead of CMAKE_USE_LUAJIT use USE_LUAJIT 20:32 artur99 i think he used Google Tr. 20:32 kahrl which should be defined in config.h similar to the other USE_ variables 20:32 iqualfragile artur99: might be rubenwardy 20:32 rubenwardy no 20:33 iqualfragile keh 20:33 artur99 thanks 20:33 iqualfragile you ere andrew ward 20:33 rubenwardy aahh!!! 20:33 rubenwardy I assume you know that from my copyright notices :P 20:35 rubenwardy or github profile 20:36 artur99 on reslults on google about him :( 20:36 artur99 https://www.google.ro/search?q=%22Mitchell+Ward%22+minetest 20:36 iqualfragile rubenwardy: yes, and i even know your email 20:37 rubenwardy Security night mare 20:38 E4xoi 2-c 20:38 E4xoi oops, sorry :P 20:38 PilzAdam I see, very engine dev related stuff going on here 20:39 kahrl seems commercial entities get relatively quick treatment of their takedown requests by google... http://www.reddit.com/r/duolingo/comments/1t0usb/fake_duolingo_in_google_play/ 20:49 artur99 in the guide: http://weblate.readthedocs.org/en/latest/usage.html i can see that can subscrie in a lot of projects, but in my profile, i can only select minetest 20:49 artur99 http://1.1.1.2/bmi/weblate.readthedocs.org/en/latest/_images/profile-subscriptions.png 20:51 artur99 ooohhh! that's the general use of weblate :D sorry 20:54 ShadowNinja kahrl: I named it CMAKE_USE_LUAJIT because that's what the other USE_* macros were named. 20:59 kahrl ShadowNinja: see config.h 20:59 kahrl the only usage of the CMAKE_USE_* macros is to define the corresponding USE_* macro 21:01 ShadowNinja kahrl: Better? https://gist.github.com/ShadowNinja/8029329 21:04 ShadowNinja Changed to use USE_LUAJIT. 21:05 kahrl I'm not too fond of macros that define classes 21:06 kahrl they confuse static analysis tools and things like doxygen, which some folks like to us 21:07 ShadowNinja Hmmm, well we don't use doxygen, and that is a lot of repetition... 21:07 kahrl I personally don't use those so it wouldn't affect me, but you have to keep that in mind 21:08 kahrl well... I guess the constructors could be shortened to 1 line each 21:08 kahrl but does it really matter? I mean who looks at exceptions.h that often 21:08 sapier shadow how do those exceptions show backtrace? 21:10 ShadowNinja sapier: If you throw a exception inside a Lua -> C++ call you get a traceback. 21:10 ShadowNinja (It's collected by the srciptError() after pcall() fails) 21:12 sapier Good idea 21:12 sapier you don't like branches ShadowNinja do you? 21:13 ShadowNinja I didn't bother too because this wasn't that big of a patch. 21:13 sapier come on ShadowNinja "not a big patch" isn't something with 3 digit lines changed ;-) 21:14 ShadowNinja sapier: Well, it's grown... 21:14 sapier you know you can switch to a branch AFTER editing files? ;-) 21:15 ShadowNinja Yes, I do. 21:15 sapier plz don't make it a habit ... adding this big things as gits always feels like someone trying to do things below radar ;-) 21:16 sapier yet I fully support those fixes, correct error handling for lua(jit) is absolutely necessary 21:27 ShadowNinja It now has every exception written out explictly(But shortened a bit). Approval? 21:29 sapier does this harm on pure lua? 21:29 ShadowNinja sapier, kahrl: ^ (Gist is up-to-date) 21:29 ShadowNinja sapier: No, it isn't compiled in in Lua. 21:30 sapier why? could this help against those instant crashes I experience in lua too? 21:30 ShadowNinja sapier: LUAJIT_MODE_WRAPCFUNC is LuaJIT-only. 21:31 sapier :-( no generic method possible? 21:31 ShadowNinja sapier: Lua doesn't seem to have this issue. See http://luajit.org/ext_c_api.html#mode_wrapcfunc 21:32 sapier so we don't know why lua errors can cause a direct core crash? 21:32 sapier ok but that's not related to this issue ... maybe you can have a look at that one next ;-) 21:34 ShadowNinja sapier: Are you talking about the SIGABRT crashes? Those are caused by some functions calling Lua functions without catching LuaErrors. 21:35 sapier should be fixed next ;-) ... imho target should be no lua error crashes core but results in proper error handling 21:35 ShadowNinja sapier: Well, do you approve this patch? 21:35 sapier your fix seems to be fine merge it 21:36 sapier another question it's already 18th what about feature freeze for christmas release? 21:38 sapier darkrose hmmmm kahrl PilzAdam sfan5 Shadow thexyz your oppinion about feature freeze for christmas release is requested! 21:38 sapier sorry ShadowNinja ... victim of ShadowBot 21:39 sfan5 feature freeze, now? 21:39 sfan5 I wanted one feature in the christmas release.. but I forgot 21:39 sapier if we want a christmas release it's time to do it NOW 21:40 sapier and if we want it for new years eve we need to do it this weekend too 21:40 sfan5 how about we add a mod that sometimes generates christmas trees with presents below? 21:40 ShadowNinja sapier: How about we get the christmas texture and model in first. ;-) 21:41 sapier if anyone of you has that mod ready to merge? ;-) 21:41 sfan5 I can do the code.. maybe a bit textures 21:41 PilzAdam last time we scheduled a release it was only 2 months delayed 21:41 sfan5 s/a bit/some/ 21:42 sapier we don't have to do this everytime ;-) 21:43 sapier despite of the download issues vanessae is annoyed of current state should be quite marture 21:43 sapier compared to last release of course ;-) 21:44 sapier and we need to fix the modstore hangs prior users get too annoyed by it 21:44 proller thread crashes must be solved before release... 21:44 sapier can you give more details proller? 21:44 proller compiling last.... maybe it freebsd-only problem 21:44 sapier what happened? 21:46 sapier atm I only know about vanessae's crashes, but she assumes those to be mod bugs, guess shadows latest fixes can give more details about this 21:47 proller http://paste.org.ru/?pq06ai 21:48 proller cmake . -DRUN_IN_PLACE=1 -DENABLE_LEVELDB=1 -DENABLE_FREETYPE:BOOL=YES -DCMAKE_BUILD_TYPE=Debug 21:49 ShadowNinja kahrl: Push your offending-nodemane patch. :-) 21:49 sapier --disable-unittests 21:49 proller ./minetest --disable-unittests - works ok, but sometimes crashes with same problem 21:49 proller like on disconnecting 21:50 sapier pthread_kill isn't something you do without issues 21:50 sapier that's a design flaw of jthread, no idea why this didn't happen before 21:51 sapier I assume this crash is result of pthread_kill beeing called before, unless you see any case where this isn't true I'd not consider this a critical bug 21:54 proller if i press esc on connecting... -> http://paste.org.ru/?3zqlmn 21:54 kahrl ShadowNinja: heh, it already classifies as "rebase needed" 21:54 ShadowNinja :-) 21:55 sapier ok that's odd proller 21:55 proller minetest.write_json - coool!!111 21:56 proller all mods need to use it instead serialize ;) 21:56 ShadowNinja :-) 21:56 sapier minetes.write_docx even more cool adds microsoft integration ;-P 21:56 proller sapier, 21:56 ShadowNinja proller: Actually I profiled it and it's a bit slower than minetest.serialize. JSON is also more limited than Lua tables. 21:56 sapier minetest.write_odt for munich administration 21:57 sapier and of course minetest.xml for same things as json 21:57 sapier and mintetest.write_pdfa for archiving reasons 21:57 ShadowNinja sapier: Yes, then we can finally have a in-game web browser. :-) 21:57 proller i have 15 seconds step on my arm, no problem with bit slower ;_ 21:58 sapier you know my oppinion ONE tool for one thing 21:58 proller with emended web server we can serve status page, or map 21:58 sapier we can't even serve our maps 21:58 proller sapier, okay, lets throw .serialize 21:59 kahrl seriazlie? 21:59 sapier replace it by json and face modders rage uppon breaking all their saved data 21:59 proller its modders problem ;) 22:00 sapier proller sometimes it's good you're that far away 22:00 sapier btw I can't reproduce your esc on connect crash 22:01 sapier there have to be some environmental things you haven't told ... or it's really a freebsd issue 22:01 sapier -free 22:03 proller 90% - freebsd, also only my freebsd have crashes on bind pull 22:03 ShadowNinja celeron55, thexyz: Can you change the icon for the Minetest organization to the Minetet logo? 22:07 kahrl wait, did anyone actually give the final agreement to merge write_json? 22:07 sapier I didn't 22:07 sapier I'm against it 22:09 kahrl well... I guess I said the code was ok, and proller said he liked adding such an api 22:10 kahrl still, I wouldn't have merged it being that controversial 22:10 sapier ok I guess I really need to write xml writer 22:10 sapier is as good as json writer and does exactly same thing ... as serialize does 22:11 sapier by now noone was able to tell me a single benefit of json to serialize 22:11 proller its readable by any lang 22:11 proller and human 22:11 sapier it's not readable 22:11 sapier its as crap as serialized 22:11 proller readable. 22:11 proller serialize 50% crappiest 22:12 proller with [0] = 22:12 sapier by the way we added that damn json reader because YOU wanted to replace the core internal json parsing ... you didn't do anything by now 22:13 sapier to me all that json crap is your(maybe shadow too) personal prefered dataformat no more no less 22:14 sapier and if we keep on adding personal prefered dataformats we will have a great converter in about 2 or three years ... but not a game 22:14 proller http://servers.minetest.net/ uses same json with in game list 22:14 sapier but not in LUA 22:15 sapier noone uses json in lua 22:15 proller why? 22:15 sapier because it's useless 22:15 sapier if you save data you save it from lua 22:15 sapier and in 99.9% of cases you need it in lua again 22:16 sapier and for saving huge data json is as useless as any text format 22:17 proller hmm.. saving map into json... good idea! 22:18 E4xoi add a csv exporter when you are at that 22:18 sapier *kicks prollers ass .... sorry but this was absolutely necessary* 22:19 iqualfragile proller: an hero… good idea! 22:19 proller also sull json key-value protocol! 22:19 proller and we can write compatible game in browsers! 22:19 sapier and maybe replace c++ by java 22:20 proller by php 22:20 sapier don't forget about the minetest.autoSendToNSA() 22:21 E4xoi haha 22:21 kahrl "mapblock": {"flags": 42, "nodes": {"x": 0, "y": 0, "z": 0, "name": "default:chest", "param1": 0, "param2": 3, "metadata": {"strings": {}, "inventory": {...}}, ...} 22:21 kahrl I think that's pretty efficient 22:21 proller super! 22:22 sapier of course that call reads all disks enables all audio and video recording devices as well as uses wlan bluetooth and all other interfaces to find other accessible devices 22:22 proller [not a joke] but for formspec - its ok 22:22 proller kahrl, efficient in minetest channel? ahahhaaaa 22:22 sapier formspec? LOL have you seen the suggestion in wiki? 22:23 proller sapier, i wrote this suggestion 22:23 ShadowNinja JSON obviously isn't for big data like a map. But it's good for small to medium piesc of data. 22:23 sapier forspec is crap but replacing a readable formspec line by a twice as long not readable json line is best example why NOT do it 22:23 proller lets replace it with html 22:24 ShadowNinja sapier: Check the Formspec json page, I tweaked it and it's much more readable. 22:24 sapier is as useless as json 22:24 sapier if you want to do formspecs in json lua xml or what other full featured language you need to provide a gui tool to create forms 22:25 sapier shadow no matter how you tweak it it's still twice the size it was before and therefore will be hard to read for more complex forms, and I'm not talking about the desired full featured gui 22:25 ShadowNinja No we don't. You can use external tools thuogh. (Like rubenwardy's nodebox maker for nodeboxes) 22:26 sapier if you don't do it json formspecs will be even worse then current 22:27 sapier and json wasn't even invented to show forms 22:27 proller and you can make small online tool in html+js for creating json 22:27 sapier I wont do anything like that 22:28 ShadowNinja sapier: We don't provide a formspec making tool for the current format. And JSON can be read and written in just about any language, so that would be relatively easy. 22:28 sapier current format is READABLE 22:28 proller wat? 22:28 sapier with one exception any element fits in a single line 22:29 ShadowNinja sapier: So cramming everything on one line makes ir readable? 22:31 sapier comparing it to exploding a simple 9 character definition to 33 characters .... YES 22:32 sapier and by the way just replacing formspec syntax by same broken json syntax is more then useless 22:33 sapier if you really want to redo formspecs do a propper redesign fix those issues formspec is limited by 22:33 ShadowNinja sapier: It may be longer, but it allows you to add optional parameters and the like. It also allows you to easily manipulate it as you can convert it to a Lua table. 22:34 sapier we don't have any optional parameters in formspec 22:35 ShadowNinja local x = minetest.parse_json(player:get_formspec()) x[2].width = 5 player:set_formspec(minetest.write_json(x)) -- Nearly impossible with a formspec string. 22:36 ShadowNinja sapier: We do actually. 22:36 sapier can you tell me a single case where this is usefull? 22:37 sapier first of all in general you don't have any idea what get_formspec will return 22:37 ShadowNinja sapier: Modifying a player's inventory without copying what you think it is and rewriting it. 22:37 ShadowNinja sapier: That's a hypothetical function, IDK if it exists. 22:37 sapier so you cannot modify anything in there without risk to break it completely or have no effect at all 22:38 ShadowNinja You would of course add more checks and such, that's just an example. 22:38 sapier so your're adding hypothetical functions for sake of functions????????? 22:38 kahrl how many modders would actually add more checks? 22:38 ShadowNinja ... 22:39 ShadowNinja I'm trying to explain it's use. You might use meta:get_string("formspec") or something similar in a real mod. 22:39 sapier converting formspec to json provides almost no benefit to formspec but adds maintenance, binds resources and doesn't give the form freedom we actually want to have 22:40 ShadowNinja sapier: Actually, do you have a better idea for a formspec serialization format? 22:40 sapier we cannot drop old formspec format for years so we need to convert it or add a second parser 22:40 sapier there's no reason to switch the serialization format 22:40 sapier if we want to redo formspecs we need to redesign it from scratch 22:41 sapier or use some already existing form description language and write a interpreter 22:41 ShadowNinja sapier: And what format would we use? 22:42 sapier I don't care about the format as long as it's a prooven format and not a self designed one like formspec ... no difference between original formspec xml formspec json formspec or soemthing else 22:42 sapier and I don't consider forms to be a actual issue in minetest atm 22:43 ShadowNinja Ok, so even you sugest JSON. XML is verl slow and complicated to process. 22:43 sapier no i don't suggest json 22:43 sapier all gui languages I know have some xml style 22:43 sapier but if you come up with a proven solution I don't care about the language 22:44 ShadowNinja XML == slow and complicated. You can't simply convert it to a ua table. 22:44 proller xml is error of 2000-2010 years 22:44 sapier and json will be error of 2010-2020 22:44 proller no 22:45 sapier a text format for data transfer ... whoever designet that one things a20 gate was a good idea too 22:45 proller you can gzip and get binary format 22:45 ShadowNinja Messagepack would also do, it can be converted to a Lua table pretty easily. 22:46 sapier so great we use cpu power to build a text and then cpu power to compress it decompress it restore binary data ..... YEAH power companys love the designers 22:46 ShadowNinja And zlib compression ought to help, not that formspecs will be all that big anyways. 22:46 iqualfragile yay, lets start a compression-algo discussion 22:47 kaeza lzip! 22:47 kaeza compresses down to 0 bytes! 22:47 iqualfragile inb4 zpaq 22:47 iqualfragile inb4 >/dev/null 22:47 sapier back to feature freeze discussion 22:48 ShadowNinja Well, whoever writes it can decide what format to use. 22:48 iqualfragile ShadowNinja: bad idea 22:48 sapier I'm gonna ask proller and shadow in half a year about a mod to use json ... let's have a look if anyone except them will use it ... if they even use it 22:48 sapier you can write whatever you want if it doesn't get merged Shadow 22:49 sapier and believe me I'll make a lot more trouble then this time if you break formspec by merging some crap like you merged json writer 22:51 proller .ha rulez! 22:51 sapier and I will revert the change if you avoid a open discussion for this point by using a gist instead of pull request 22:53 ShadowNinja ... 22:53 sapier I'm not kidding ... I really dislike your gist style merging of big changes 23:03 sapier ShadowNinja does write_json really crash on writing non serializable data? 23:04 ShadowNinja sapier: Yes, it throws a LuaError if the table can not be converted to JSON. 23:05 sapier you know this is inconsistent to other serialization methods we have? 23:06 sapier it should just return nil in this case 23:07 ShadowNinja Hmmm, a try-catch block ought to do. 23:08 ShadowNinja My server still isn't showing up on the serverlist... 23:08 sapier this is one of the reasons what pulls are for ... the more people see it the more likely bugs like this one are recognized 23:10 rambomedic Do you think we could get that sonic ether guy to port his shaders to minecraft? 23:10 proller less do, more speak! 23:11 sapier proller I guess some time you will realize software development is more talking then coding 23:11 hmmmm sapier, I mentioned yesterday that we should shorten the length of the feature freeze because there isn't nearly as much stuff this release 23:11 hmmmm so I want to enter feature freeze on Saturday if that's fine 23:11 hmmmm still releasing on christmas 23:11 sapier 3 days? isn't this a little bit too short? 23:12 hmmmm it'll be 5 days 23:12 sapier ok ok christmas is 24th in germany 23:12 proller and fix crash! 23:12 proller btw need to test in windows 23:13 sapier proller that's up to you I can't reproduce it in linux with the information you provided 23:13 sapier and crash on pthread kill ... yea can happen 23:14 sapier hmmm I assume you meant saturday 0.00am? 23:16 ShadowNinja sapier: Good? http://ix.io/9rE 23:16 sapier should fix the issue yes 23:19 sapier why didn't we add that json thingy by updating serialize? e.g. minetest.setialize(something) == minetest.serialize(something,"default") and minetest.serialize(something,"json") returns json? 23:19 sapier this way if next time someone wants csv we don't need to add another set of api calls 23:50 rambomedic can somebody please confirm and merge my pull request for minetest_game? 23:50 rambomedic It will greatly remedy the amount of time it takes to connect to a server 23:52 sapier can you post a link here rambomedic? 23:53 rambomedic https://github.com/minetest/minetest_game/pull/220 23:53 rambomedic maybe not *greatly* remedy 23:54 rambomedic but atleast it'll cut a third off the time it takes to load textures 23:54 sapier for most servers this won't help very much as those use custom textures 23:55 rambomedic I see. 23:56 sapier and due to minetest internal limitations reducing size from 300 to 190 bytes won't make much difference 23:56 sapier still I support merging it, but I usually don't work at minetest game 23:57 rambomedic *kilobytes 23:58 rambomedic And what is that limitation (curious)? 23:58 sapier textures are sent each texture within a packet ... max packet size is 512 bytes (for reasons noone can really explain) 23:59 sapier it might help for curl enabled servers, but those usually don't suffer that much