Time Nick Message 00:00 Fixer 4-5 compiles, sucks 00:01 Fixer good night, have fun 00:01 Fixer VanessaE: post your confirmation if you seen this bug 00:01 VanessaE meh 00:01 VanessaE old bug. 00:10 bkeys rubenwardy: I am getting a 400 error; it seems like maybe it is having trouble parsing the JSON data? 00:11 bkeys I would think the data itself would be unstructured and you could order it however you want so long as the same fields are there 00:13 rubenwardy Make sure you don't have syntax errors in you JSON 00:14 bkeys https://da.gd/tJ6J 00:14 bkeys This is the JSON I am sending up 00:15 VanessaE bkeys: just what are you doing anyway? 00:15 bkeys VanessaE: Trying to use the minetest master server for my non minetest game 00:15 VanessaE ah 00:15 rubenwardy It's probably an abuse tbh, you're better off writing your own 00:15 rubenwardy it's not very complicated as a piece of software 00:16 bkeys I know that; but this is kind of more for personal satisfaction anyway 00:16 bkeys What part of this is specific to minetest? 00:16 bkeys It looks pretty generic 00:17 rubenwardy well, for one it checks that the address leads to a valid Minetest server by sending a Minetest hello packet 00:17 bkeys Hmm; anything beyond that? 00:19 bkeys And with the hello packet, as long as it is through the REST API the functionality itself should still be generic 00:19 bkeys If anything these seem like features that a person would want to be available in any master server 02:56 Wayward_One Rather late, but regarding the name suggestions, I really like Kuutio as well (FWIW) 03:22 paramat if game#793 is merged please merge game#1387 too 03:22 ShadowBot https://github.com/minetest/minetest_game/issues/793 -- Improve Chest appearance - opening chests. by sofar 03:22 ShadowBot https://github.com/minetest/minetest_game/issues/1387 -- Gunpowder (and tnt.burn) will trigger the on_ignite of nodes by Ferk 03:29 paramat game#1668 can have sounds corrected on merge as author is long absent 03:29 ShadowBot https://github.com/minetest/minetest_game/issues/1668 -- add snow and ice stairs and slabs by timsoftgit 03:32 paramat nore do you oppose game#1693 ? if so it can be closed (3 disapprovals) 03:32 ShadowBot https://github.com/minetest/minetest_game/issues/1693 -- Disable fire by default on servers by Ezhh 03:33 nore paramat: don't close it before the gui conf stuff is merged, please 03:33 paramat ok no problem 03:46 paramat game#1547 should be closed i feel, more opinions? 03:46 ShadowBot https://github.com/minetest/minetest_game/issues/1547 -- [suggestion] Make obsidian glass `glasslike_framed` 04:01 paramat ^ nore sofar 06:20 sofar nerzhul: got a sec? 06:41 nerzhul hi sofar , what do yuo want ? 06:41 sofar so, nore and me have been thinking of trying to see if we can make setPos() safer 06:41 nerzhul yeah ? 06:42 sofar by sending the new block immediately before the teleport setPos 06:42 sofar problem is, I have no idea where in the stack it's the easiest to do so 06:42 sofar or even what method would be the right one to use to do this 06:42 sofar I tracked lua down to setPos() in content_sao.cpp 06:43 sofar where the new pos is set and queued 06:44 sofar do I have to figure out how to call SendBlockNoLock() manually for the blockpos somehow? 06:44 sofar e.g. src/server.cpp :2290 06:45 nerzhul you want to send block if block not in range when doing setPos ? can be a nice idea, i should go to work, i will be back in 30mins 06:46 sofar nerzhul: I'll be long gone in bed 07:20 nerzhul i'm back sofar 09:42 Megaf Hi all 09:43 Megaf is dis a bug in Minetest? http://paste.debian.net/plain/929841 09:43 Megaf that's from a fresh master clone 09:46 Megaf Here's the full thing http://paste.debian.net/plain/929844 10:11 Megaf problem was here, please ignore it, thanks. 14:18 rubenwardy Slept on it - decided I really like the name Kuutio 14:21 nerzhul it's a quite good name, but not for voxel i think, but why not. And really, if we want to rename, just do it after release not before, if 0.5 is for breaking things, include name is okay 14:21 nerzhul it permit to break mods with kuutio.get_node instead of minetest.get_node 14:21 nerzhul (note: using core instead of minetest everywhere is better... i don't know why we kept this namespace) 14:30 rubenwardy nerzhul, it's not development related 14:30 rubenwardy it's basically just branding 14:30 rubenwardy and I agree it would be a good 0.5 thing 14:31 nerzhul rubenwardy: hopefully yes, it's related, it's in every mod, in every android build, in every linux package, it's also in natural referencing on search engine, it's not a simple thing to do 14:31 rubenwardy we'll be a little more Finnish-ed then 14:36 kilbith changing of name would hurt our etablished reputation 14:38 kilbith can we simply stop wasting our time on unproductive subjects 14:38 sfan5 ^ agree 14:40 Megaf +1 14:40 Fixer sfan5: here is demo of my problem, protector redo: https://imgur.com/a/JUZZi 14:41 sfan5 i don't think this is new 14:41 behalebabo it's not 14:41 sfan5 entities are not rendered if you face away from then regardless of size 14:41 sfan5 them* 14:42 Fixer sfan5: new on my setup, had no problem with like in .13 .14 .15... 14:42 Fixer i will pinpoint a commit that changed it 15:24 nerzhul i approve kilbith and sfan5 before discussings like politics, merge pr merge pr merge pr :D 15:45 red-001 could someone reopen #3050? 15:45 ShadowBot https://github.com/minetest/minetest/issues/3050 -- Shutdown progress bar not rendered if inventory is open 16:00 celeron55 changing the name could be done as a competition by having an official friendly fork and seeing how popular it gets on its own without pulling people from the existing community 16:01 celeron55 i could do that on my own without any effect on current development (other than me being busy doing that) 16:03 celeron55 i don't need any permission to try that of course but... what do people think? 16:04 red-001 well it's an intersting idea 16:07 celeron55 it's relatively easy to merge minetest into the fork if it works out and people love the name, and if the name didn't turn out to be worth anything, then just drop the fork 16:09 red-001 that's not a bad idea 16:16 Fixer i have insane suggestion 16:17 Fixer there was discussion about uselessness of locked doors, we have keys... what about using key (say protector key) for locking ANY node? 16:25 nerzhul today i coded a lua interface onto my IRC bot, and in fact Lua is great, but having too many Lua result to an increasing memory usage, not reclaimed by GC, just reused by Lua GC when needed 16:25 nerzhul then memory usage on Lua stack could not decrease if references are lua variables 16:26 nerzhul other problem, Lua copy the memory segments from/to Lua, this cause many overhead when using string, especially 16:26 nerzhul concrete example, use minetest http api, download a 1MB page, the whole memory segment of the page is copied from C++ to Lua and then when GC Pass on Lua, 1MB is not retrieved by system and kept in Lua buffer pool 16:27 nore nerzhul: so all the strings that are used somewhere in the API calls are leaked? 16:28 nerzhul pointers are invalidated and deleted, but memory is kept by Lua stack for future usage 16:28 nore oh 16:28 nore could this explain the OOMs? 16:28 nerzhul i don't think 16:28 sfan5 np 16:28 sfan5 no* 16:28 sfan5 and i don't think lua has memory leaks by default 16:28 sfan5 the gc will pick it up 16:29 nerzhul because lua memory usage is high, but it's recycled 16:29 nerzhul it's like JVM memory in fact 16:29 nore yeah, that was what I was thinking 16:29 nore so the OOMs are somewhere else 16:29 nerzhul your commited memory increase at each time you got it, and you reuse the current commited memory when GC pass 16:30 nerzhul also lua uses markspeed like java 7/8 default GC 16:30 nerzhul marksweep* 16:31 nerzhul nore, but if you Luastack take 700MB at a point, they will never be reclaimed 16:31 nore nerzhul: yes 16:31 nerzhul so having more intelligent core side call can reduce the Lua memory copy/GC overhead 16:32 nerzhul atomic calls are a pain in terms of performance 16:32 nore something I was wondering was if the lua voxel manipulators could not be a cause of some kind of leak 16:32 nerzhul it's also why voxelmanipulator is slow, you copy the memory blocks between lua and C++ 16:32 nerzhul lua leak, no, but huge memory pool yes 16:33 Fixer OOMs are strange in my case 16:33 Fixer mem usage is 20-40-60-40-30 mb... and then just OOM 16:33 nerzhul the only way to reduce memory usage for this part is, i think, i use object references which include other objects on the heap and remove them when lua reference is destroyed by gc 16:33 Fixer even on newest dreambuilder 16:34 nerzhul maybe i will look at this after this release, i have some little refactor optimizations to send as PR in the next 4 weeks, Lua performance can wait a little bit 17:08 Krock nerzhul, I think you misunderstood something in https://github.com/minetest/minetest/pull/5647#discussion_r113301039 17:08 Krock "Scripting" itself should be some kind of a glue that contains a _pointer_ for the specified scripting type 17:09 Krock overall this class is not larger than a pointer (32/64 bits) + enum/int (32/64 bits) 17:11 Krock I'm looking for a way to get access to the current present scripting API from the node metadata and inventory code - but will hopefully find an elegant way tomorrow 17:12 sfan5 int is 32-bit usually 17:12 sfan5 and "enum" is not a type 17:13 Krock but it also relies on a type 17:25 benrob0329 Hi Krock 17:25 benrob0329 Oh..opps 17:26 Fixer sfan5: pinpointed exact commit that introduced that entity render change, that is "soft node overlay" commit 17:30 nerzhul enum size is int, except in C++11 where we can say: enum Sth : uint8_t { .. }; 17:30 nerzhul it's very nice C++11 heh :D 17:40 * sofar wakes up (somewhat) 17:40 * nerzhul throw an apple to sofar 17:44 Krock merging #3497 and #5652 in 20 minutes 17:44 ShadowBot https://github.com/minetest/minetest/issues/3497 -- Allow mesh and nodeboxes to wave like plants or leaves. by sofar 17:44 ShadowBot https://github.com/minetest/minetest/issues/5652 -- Make torch digging instant again by HybridDog 17:51 sofar nerzhul: about the 'safe teleport' 17:51 nerzhul you mean the thing you talked this morning CEST ? 17:53 nerzhul Krock: sofar -> #5548 needs you 17:53 ShadowBot https://github.com/minetest/minetest/issues/5548 -- [CSM] Add event on_place_node API lua by Dumbeldor 17:53 sofar nerzhul: yes, so, sending out a block for a player getting teleported 17:53 paramat all devs, the ninja will not be here tomorrow for our PR review so i am organising it. how about we make it later at 8pm instead of 6pm UTC to be better for our USA devs? 17:54 nerzhul sofar: it seems to be an interesting idea, but it needs a delay no ? and why do you want to send mapblocks before teleporting ? to prevent waiting player for asking for mapblocks ? 17:54 nerzhul 8pm UTC ? 17:54 nerzhul it's 10pm CEST no? 17:55 sofar nerzhul: because if the map area has significantly changed at the teleport area 17:55 nerzhul UTC is your timezone paramat right ? 17:55 sofar nerzhul: the player ends up falling if everything was previously air 17:55 paramat UTC = GMT 17:56 nerzhul sofar: for example, you visited the area, it's loaded, you are very far, somebody do worldedit on it, you don't know it's modified you teleport, you are un previous mapblock, you want new mapblocks and it's not very consistent ? 17:56 nerzhul paramat: 8pm is 8pm for you ? 17:56 nerzhul you are un UKxit ? 17:56 sofar nerzhul: this is especially painful in our "itb" project 17:56 paramat i'm GMT + 1, british summer time (unfortunately) 17:56 nerzhul okay, i'm GMT +2 if i remember 17:56 sofar nerzhul: where we use vmanips to build up and tear down large amounts of blocks and teleport the player to it 17:57 sofar nerzhul: so it's almost guaranteed that the client has the wrong blocks 17:57 nerzhul sofar: objective ? arenas ? 17:57 sofar something like that, yes 17:57 nerzhul sofar: the only thing i can suggest is sending mapblocks just before sending set_pos 17:57 nerzhul in the same loop 17:57 nerzhul and only if distance pos > configured_distance 17:58 nerzhul it's the shorted way to dev it 17:58 sofar something like that 17:58 sofar but I'm lost in OOP hell 17:58 paramat ^ celeron55 nore rubenwardy sfan5 Krock is 8pm UTC ok for tomorrow's PR review? could be 7pm if better 17:58 nerzhul 10PM for me, i hope i will not be into bed :p 17:58 paramat hm maybe 7pm is better 17:59 Krock got to do on sunday. can't tell if I'll have time for it 17:59 paramat any problems with 7pm? 18:00 paramat .. UTC, too late for anyone? 18:02 Krock AFAIK 7/8 PM UTC always worked quite good for the most people 18:04 kilbith not good for Zeno` 18:08 paramat quite late in Finland though 18:08 paramat ok let's do 7pm UTC tomorrow 18:11 rubenwardy that's fine 18:12 garywhite 7pm UTC/2pm ET/12pm PT 18:26 Fixer bug detected 18:26 Fixer impossible to place a block on top of locked chest with shift-click 18:27 Fixer you are welcome 18:27 nerzhul please review #5674 (fairly trivial) 18:27 ShadowBot https://github.com/minetest/minetest/issues/5674 -- [CSM] add screenshot api lua by Dumbeldor 18:27 paramat ok. i will be less corperate, no words like 'meeting', 'item', 'agenda' or 'minutes' :] 18:29 ShadowNinja I plan to merge my getTime patch soon. 18:31 Fixer ahttps://github.com/minetest/minetest_game/issues/1715 18:35 paramat Krock ShadowNinja are you ok with game#793 being merged? 18:35 ShadowBot https://github.com/minetest/minetest_game/issues/793 -- Improve Chest appearance - opening chests. by sofar 18:36 Krock +1 on the concept but haven't tested it 18:39 ShadowNinja paramat: Looks O.K in theory, haven't looked too closely at it. 18:39 paramat ok i will merge games 793 1387 and corrected 1668 later if no-one else has 18:41 Krock the other two are fine 18:47 Krock paramat, is 5533 still in your "to test" queue? 18:48 paramat stripped sneak? yes certainly, will do a thorough review 18:48 Krock thanks a lot :) 18:57 paramat might do that tonight 18:59 Krock no rush - as long it doesn't get forgotten ;) 20:00 ShadowNinja Ehm, protection_bypass wasn't done very well. Now the protetction check boilerplate is even bigger, and it's duplicated functionality shince protecting mods already implement thoir own protection bypass privileges. 20:00 ShadowNinja Eg, my mod has the "areas" priv which allows changing areas and bypasses protection. 20:01 DS-minetest IMO having one for all is better 20:02 ShadowNinja DS-minetest: But now you've got two. I ca't remove my priv because it's needed for all the other area administration stuff. 20:03 ShadowNinja More importantly, you now need to copy/paste at least 4 lines to handle each protecting check. 20:04 DS-minetest ShadowNinja: then only use that priv for area administration stuff 20:04 DS-minetest for the other* 20:05 ShadowNinja DS-minetest: If you can change areas at will then you can already override protection, so there's no sense in making you add yourself to the area for that and having a seperate privilege. 20:06 ShadowNinja The way that I designed the privilege API basically means we have to options: make every protection mod handle protection overrides, or make every other mod handle it in their boilerplate. 20:07 ShadowNinja I think that the first option is much more reasonable. 20:11 ShadowNinja We could of course create a protection API v2 (probably not a bad idea anyways), but without doing that I think it has to be one of those two ways*. 20:11 ShadowNinja * without horrible ugly unthinkable metatable hacks. 20:28 sofar FYI I can't make tomorrow's meeting (I'll be at a birthday party) 20:30 paramat ShadowNinja i don't like protection bypass either, seems a mess 20:39 paramat .. the extra lines needed to be added everywhere 20:46 nore can't make it either (in a plane again, coming back from Sweden) 20:52 paramat ok 20:58 paramat #4524 is updated, might be ready now 20:58 ShadowBot https://github.com/minetest/minetest/issues/4524 -- Clouds api by bendeutsch 20:58 paramat i need to review the update though 21:06 sofar oh good he updated that? 21:07 sofar just nerzhul check that it's fixed the way he wanted 21:08 paramat well, i think 'cloud params' is better than 'cloud parameters', see my line comments 21:09 paramat shorter and more consistent 21:11 sofar meh 21:11 sofar fix-on-merge if you insist :P 21:23 paramat yes i'm happy to do that if the author is not 21:25 paramat or maybe fix after merge as it is a lot of changes and mistakes likely 21:25 paramat i would be happy to do that too