Time Nick Message 00:05 TBC_x you guys really argue on using i++ vs ++i? 00:08 TBC_x afaik i++ and ++i don't change meaning when used standalone 00:10 TBC_x the '++' stands for when the variable gets changed 00:13 hmmmm it makes a difference when i is a nonprimitive type 00:14 TBC_x the following will evaluate to true: i = 0; bool result = (i++ == 0 && ++i == 2) 00:28 TBC_x hmm... I see 02:54 paramat hi hmmmm you've been quiet recently, i'll prepare this for merge later #3118 also this issue surfaced game#652 i will sort this out soon 02:54 ShadowBot https://github.com/minetest/minetest_game/issues/652 -- Schematic tree trunks destroy existing nodes 02:54 ShadowBot https://github.com/minetest/minetest/issues/3118 -- Dungeongen: Remove floating stone rings by paramat 03:32 hmmmm ahh you found the problem 03:32 hmmmm nice 03:32 hmmmm if #3118 works as it's supposed to, then looks good to me 03:32 ShadowBot https://github.com/minetest/minetest/issues/3118 -- Dungeongen: Remove floating stone rings by paramat 03:34 paramat okay just need to test again 03:36 hmmmm been fighting a couple fires at work lately 03:36 hmmmm honestly i could be doing minetest right now since it's the weekend but i'm kind of on a roll with work stuff 03:36 paramat aha 05:25 paramat now merging #3118 05:25 ShadowBot https://github.com/minetest/minetest/issues/3118 -- Dungeongen: Remove floating stone rings by paramat 05:31 paramat merge complete 16:43 VanessaE hmmmm: ok now I've got those not-enough-memory errors happening in places that make no sense at all: 2015-08-29 18:33:21: ERROR[main]: Lua: OOM error from mod 'carbone_mobs' in callback luaentity_Step(): not enough memory // 2015-08-29 18:33:21: ERROR[main]: Current Lua memory usage: 856 MB 16:46 nrzkt VanessaE: this can happen everywhere, it's because the whole Lua VM take so many memory, not the current call 17:27 VanessaE great. 17:30 VanessaE nrzkt: so what can be done to fix this? 17:30 nrzkt find the consumming mod :s 17:31 T4im no way to get the memory consumption of a lua table either, it seems :/ 17:33 VanessaE nrzkt: that's next to impossible, you know, and with 24 GB of RAM I shouldn't have to worry about some mod that briefly consumes RAM from time to time. 17:35 nrzkt the problem is not the server ram amount but the Lua memory limit 17:46 T4im VanessaE: how well is that one reproduceable? is it like oom on each startup, or eventually after a while randomly? 17:47 VanessaE T4im: it has crashed several times today alone. 17:47 VanessaE and it seems to be within a few mins of startup 17:48 T4im One workaround-ish idea would be to manually pcall() mod executions such, that you get "possibly" output some location info (though that can just as well just throw the next oom error) 17:48 T4im that oom seems generally to be pcall catchable though 17:48 T4im (you might have to cover some stuff in builtin/ too) 17:49 T4im also _G dumping on error might help, if the offending mod references it globally somehow 17:51 T4im then again, that _G dumping probably won't work without acquiring more memory first :/ 18:05 VanessaE all right, this is confusing me.. how exactly do I specify an alternate install location for luajit? 18:06 VanessaE (to compile MT against it I mean) 18:10 VanessaE nevermind, figured it out. 18:11 VanessaE I'm rebuilding this against luajit 2.1.0 just to see how it behaves. 18:38 VanessaE nrzkt: so what about the info described here? http://bayesanalytic.com/access-extra-memory-from-lua-jit/ 18:42 hmmmm T4im, if we have tighter control over the GC behavior then it's theoretically possible to allocate a large enough chunk of memory from the start, say 10 MB, and then on OOM force it to become deallocated (then carry out the _G dump) 18:42 hmmmm I agree on both fronts, it's just as necessary to discover what's using up all the memory as it is to have a higher memory usage limit 18:43 hmmmm VanessaE, the problem with that link is that it requires the ffi extension which is not a part of lua 5.1 18:43 VanessaE it's part of luajit though. 18:43 VanessaE http://luajit.org/ext_ffi.html 18:43 hmmmm am aware. 18:43 VanessaE ok. 18:43 hmmmm if you want to break compatibility like that, it's your own call.. 18:44 VanessaE eh, no. :) 18:44 hmmmm not to mention that's manual memory management 18:44 hmmmm I doubt it would integrate well with minetest's existing setup 18:44 VanessaE just trying to help find a solution 18:45 hmmmm if you're starting with a fresh api like the guy here is then i'm sure you can do it 18:45 hmmmm I don't think it's too plausible to use this for minetest though 18:46 hmmmm between the two options of reducing memory usage and increasing memory capacity, the former is much more valuable 18:46 hmmmm you don't want GBs of memory to get handled by lua's GC 18:46 hmmmm and there's no telling whether or not that memory usage will simply rocket up to the new maximum capacity 18:46 hmmmm if lua is using that much memory to begin with, there's a bug somewhere 18:47 hmmmm or people didn't design things too intelligently 18:47 T4im unnecessary memory usage... but what about the space-time tradeoff kind of memory usage? we don't necessarily want to get back to slow but memory saving usage :D 18:47 hmmmm when you're talking about GBs of memory that's a false dichotomy 18:47 hmmmm iterating over so much memory isn't going to make anything faster 18:48 T4im well yea.. those nearly 900mb indicate a problem of course.. 18:48 T4im unless dreambuilder simply has become THAT big already? :D 18:48 hmmmm it's not possible 18:48 VanessaE dreambuilder is big but not THAT damn big. 18:48 hmmmm that is not a natural amount of memory usage 18:49 VanessaE the biggest memory usage of anything I can think of is biome_lib, but even THAT doesn't use that much (a few tens of megs if my estimate is correct, and then only on a 10-20-seconds-at-a-time basis, during mapgen only) 18:50 VanessaE (and the code that does THAT hasn't been touched in like 6 months or so) 18:55 VanessaE in any case, this latest bout of OOM crashes came after pushing some updates this morning.........and all those consisted of was some new skins. 18:56 VanessaE (no code updates, I'm pretty sure) 18:56 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/stats.html 18:56 VanessaE look at the memory usage graph here; that train of "saw teeth" from 12:00 to a bit over an hour ago was VE-Survival restarting repeatedly. 18:57 VanessaE sometimes with Lua itself reporting an "unprotected" OOM, sometimes with minetest reporting it alongside a backtrace. 18:57 T4im I don't take it you still have the profiler running? 18:57 VanessaE no. 18:57 VanessaE (the server went down because it's running /clearobjects now) 19:00 VanessaE http://pastebin.com/NcK64HZC 19:00 VanessaE there's an example of one of those "unprotected" restarts. never seen this before. 19:01 VanessaE times in the log are CET, so that particular crash was only about an hour and a half ago 19:02 VanessaE server was at commit ffe3f03b at the time 19:43 paramat celeron55 i hope you return to work on minetest again, what new features would encourage you to use and develop minetest again? 20:06 celeron55 it's complicated 20:08 celeron55 i don't think it's that much up to new features; i can make new features myself if i want to 20:09 celeron55 altough, i seriously lack any kind of routine for it by now in this project 20:14 celeron55 i'm not interested in any kind of first person block games at all 20:14 celeron55 there's no way to make minetest interesting because it's quite limited to being just that 20:15 rubenwardy celeron55: with client-side scripting, you could make the ability to control inputs and cameras 20:15 rubenwardy and so make an RTS camera 20:15 celeron55 minetest is stupidly far from an RTS engine 20:15 rubenwardy I suppose switching to marching cubes might be quite difficult 20:16 celeron55 the most you could do is a DOTA clone, which i find very boring too 20:17 celeron55 a thing i am interested to try to use is fixed-size worlds with all entities active at all times 20:17 celeron55 first person is fine, i don't have a reason to care about that 20:18 celeron55 i mean, i have as a separate thing, but not related to fixed-size all-active worlds 20:19 celeron55 but for example right now there is no way anyone could make me use minetest, even these features or even money or anything; i have other projects that need me way more 20:20 rubenwardy by use, do you mean to make a game in it, or to develop it further? 20:20 rubenwardy I agree that it's too limited as a game engine 20:24 celeron55 don't say that; that makes you have to define what actually is a game engine and it's unproductive 20:24 rubenwardy lol 20:25 rubenwardy that is true 20:25 celeron55 rpg maker is a very bad game engine if you want to make an fps; that doesn't make it bad though 20:25 celeron55 minetest is very bad if you want to... umm... well do things that it's not designed for 20:27 celeron55 minetest is a one-instance-per-server multiplayer block sandbox engine 20:27 celeron55 or, that's what it's designed to be 20:27 celeron55 (instance or world or however one wants to call it; instance is kind of an MMO term i guess) 20:30 paramat okay i see. i miss your influence here a little =) 20:34 celeron55 now that i think of it, i think one reason i don't care about minetest is that i have learned a lot about game design since i started 20:35 celeron55 minetest is not a good platform for many game designs at all 20:35 celeron55 it's more of a playground than a game (well duh, sandbox) and it doesn't allow me to make games as i want to 20:36 twoelk "fixed-size worlds with all entities active at all times" sounds like beeing able to use minecarts for transporting resources accross playerless streches of the map 20:36 rubenwardy I guess it's rather hard to fix that 20:36 rubenwardy twoelk, not really, that wouldn't be fixed-size :P 20:36 celeron55 twoelk: well yes, if by stretches you mean "across the 500 node wide map" 20:38 twoelk for example to have npc's involved in some sort of economy that does note relate to player presence 20:38 rubenwardy loading a chunk for a cart seems a big overkill, I'd use some sort of node network with timings 20:38 rubenwardy ie: 5 minutes between node A and B. 20:38 twoelk -> the typical warcraft enemy buildup hidden from the player 20:39 rubenwardy but I guess it wouldn't be as flexible in terms of eg: stuff falling onto rails 20:39 rubenwardy celeron55, is buildat more of what you wanted Minetest to be? 20:40 celeron55 i wouldn't want to spend time implementing stuff on some obscure level-of-detail abstractions; i'd just want the 500x500 world where everything works all the time according to the single lines of code i would have written 20:40 celeron55 rubenwardy: buildat is buildat 20:41 rubenwardy de-ja vu 20:41 rubenwardy I think I've asked that question before, and you gave the same response :P 20:41 celeron55 minetest was meant to be a survival-focused minecraft clone 20:41 celeron55 it didn't really turn out that way 20:42 celeron55 it's more like house building focused now; not really something i need or want 20:42 rubenwardy more of a builder with increasing genericalism 20:42 rubenwardy inb4 :P 20:43 celeron55 but i'm not really interested in designing the survival game i wanted to make back tehn 20:43 celeron55 then* 20:44 celeron55 i know it's probably not going to work the way i thought it could 20:44 twoelk the cart was just an example - another extreme would be a herd of npc-animals roaming the map in a sensible migrational way or a ferry line working, player present or not. If survival goes beyond the hunter gatherer level I would think this becomes more and more important to have 20:47 hmmmm you know, maybe it's not such a good idea to try to make minetest do everything 20:47 celeron55 the way i think about minetest is that it's one of my game prototypes, but one that seriously got out of hand 20:47 hmmmm i love the example with RTS maker, and it's common knowledge that games which try to do too much fall short in lots of areas 20:47 celeron55 _rpg_ maker 20:47 hmmmm i'd rather have minetest remembered as a quality piece of software that does X, and not a piece of shit that's bad at everything 20:48 hmmmm right now it's the latter 20:48 celeron55 now just define X 8) 20:48 hmmmm the only people who actually like minetest are its hardcore devotees who make mods and are generally known as part of the "community" 20:48 hmmmm X is minecraft/infiniminer-inspired sandbox games 20:48 hmmmm i realize there's a push in all sorts of directions thanks to modders 20:49 hmmmm just like there was a push to have starcraft in all sorts of awkward directions thanks to map makers 20:49 celeron55 infininiminer has fixed-size worlds and each game has a beginning and an end i think; should minetest support that then 20:49 hmmmm whenever there's a potential for user generated content, content makers will try to push the platform to do things it was never designed for 20:49 hmmmm of course it should 20:49 celeron55 or is infiniminer there "just because" 20:49 * twoelk hasn't published any mods, but loves building so that still fits sort of what has been said 20:50 hmmmm i think it might be a better idea to resist the pushes into these awkward directions rather than facilitate bad ideas 20:50 hmmmm remember there was that FPS mod? 20:50 hmmmm we do plan on making some improvements to minetest which may help make FPS mods easier 20:50 hmmmm but there's no catering involved 20:51 hmmmm i thought about it long and hard and indeed, you're right, it's a terrible idea to make minetest "generic" 20:52 twoelk so a list of goals is once again needed? 20:52 hmmmm the first step is to stop advertising minetest as a generic game engine 20:52 rubenwardy and get a good minetest_game, then 20:52 celeron55 buildat is like minetest but fairly generic, but even it is limited to multiplayer servers by design (not single-world though) 20:52 hmmmm it never was that, we're only fooling ourselves trying to make it into that, and it's wreaking havok on overall quality 20:53 twoelk so exactly which direction should be discouraged? 20:53 celeron55 but... it's not really advertised as a game engine, but as an infinite-world block sandbox game engine 20:53 hmmmm to make everything generic 20:53 celeron55 there's a large difference 20:53 hmmmm generalize things within reason 20:54 celeron55 maybe we should drop the engine anyway though 20:54 twoelk argh! strike the word infinite 20:54 hmmmm infinite can be done with multiple worlds 20:54 celeron55 it's infinite compared to all the shitty open source prototypes with a 200x200 world 20:54 twoelk yeah I can have infinite bread if I keep buying new loafs 20:54 hmmmm there's another minecraft-like game out there that has this distributed server network thing going on where users can walk off the edge of one world and enter another server transparently 20:55 hmmmm that's really cool but this sort of functionality is obviously a stretch goal 20:55 VanessaE didn't...I suggest that like, a year ago? 20:55 hmmmm suggest? 20:55 celeron55 let's define it as "31k^2-world" instead of "infinite-world", ok? 8) 20:55 hmmmm i suppose so 20:55 celeron55 "31k^3-world"* 20:55 hmmmm but a mere suggestion is a long way off 20:55 twoelk mt is large but "infinite" raises interest in the wrong direction 20:55 rubenwardy 62k? 31 is radius 20:55 celeron55 oops 20:56 paramat 'almost infinite' =P~ 20:56 celeron55 "62k^3-world"* 20:56 hmmmm just a reminder that ideas are plentiful, shit-getting-done is rare 20:56 VanessaE hmmmm: fair enough, but point is the idea came up before and no one felt it was a useful way forward. 20:56 hmmmm oh if i had a nickel for every "idea guy who needs a technical guy" situation 20:56 sfan5 mingw32 build is broken: http://sprunge.us/ZTeT 20:57 celeron55 adding that to minetest is infinitely easy; just add an api like minetest.forward_player("hmmmm", "server.net:30005") 20:57 hmmmm not saying that idea guys are worthless, but it's just not worth much if it isn't executed upon successfully 20:57 celeron55 or "redirect" or whatever 20:57 VanessaE (the "better" way instead seemed to settle on splitting the vertical dimension into multiple sub-realms) 20:57 celeron55 then just let the modders and servers make use of it 20:57 hmmmm celeron, good job defining an interface 20:57 hmmmm now how about all the hard parts 20:58 hmmmm but anyway no, what I was describing does the server connection transparently and it seamlessly integrates 20:58 VanessaE hmmmm: better idea is a teleporter block that does that. 20:58 VanessaE I guess minecraft uses "linking books" for this purpose. 20:58 celeron55 well i was thinking it would just disconnect and connect to the other place 20:58 celeron55 but i guess you were thinking of something fancier 20:58 hmmmm i.e. when a player gets within the view range of another server's nodes, it automatically connects and begins requesting blocks from that other server 8) 20:59 hmmmm no I'm not saying this is an idea, it's a feature that already exists in another minetest competitor 20:59 hmmmm instead of modding and flexibility and all that neat stuff, other people have gone into other directions 20:59 twoelk I'm all for a "limbo" ocean between servers to ease the transfer 21:00 hmmmm VanessaE: yes, a teleporter block makes it easy to do this 21:00 celeron55 whatever, the point of this discussion is not this particular feature 21:00 hmmmm from a user point of view 21:00 hmmmm but you also need the authentication 21:00 VanessaE of course. 21:00 hmmmm and it's a lot of work on Client to make this work 21:00 hmmmm err 21:01 hmmmm Client is not designed for this* there'd be a lot of code changes necessary 21:01 VanessaE I'm just saying from a gameplay standpoint; when you get to the edge of the world, frankly I think it would be better to render a "wall" or something, than to try to fetch blocks from another server. 21:01 hmmmm it's not as simple as adding some kind of api 21:01 hmmmm unless you want something utterly useless 21:01 VanessaE (think minecraft's "farlands" but simulated) 21:01 hmmmm I guess 21:01 twoelk bah, just give the main menu a tardis like ingame interface that fools the player into thinking he is not in the main menu 21:02 celeron55 well i mean, minetest's design policy was always "prototype first, think later" when i was the one who did things; and i'm sure a primitive version of this feature would have been found fun just fine without all the concerns being cared about 21:02 VanessaE heh 21:03 celeron55 i guess it's now different, but that's part of the problem 21:04 hmmmm I really dislike that design policy 21:04 hmmmm still to this day it causes a lot of problems 21:04 celeron55 i know you do, and i am not saying it would work now in this organization 21:05 celeron55 but MT wouldn't exist without it 21:05 hmmmm but then again you were designing minetest for yourself at the time 21:05 celeron55 there's no way MT would have become anything if i were to get stuck into any details 21:06 hmmmm I think you really shot yourself in the foot when scripting got added 21:06 celeron55 maybe; it was fun at the time though 21:06 hmmmm if minetest remained scriptless then your design policy would be completely acceptable 21:06 hmmmm or maybe if the number of users remained relatively small 21:08 twoelk but mt would not have been so diverse 21:09 rubenwardy I like diversity, but it's pointless if the quality's no good 21:10 celeron55 i think one of the elephants in the room when it comes to quality is formspecs 21:10 hmmmm that's getting worked on 21:10 celeron55 and because of this i am waiting for what comes of the ogre3d thing 21:11 VanessaE did zeno never release his code? 21:11 rubenwardy I was talking to an Irrlicht guy, they actually expect people to make their own GUI system :P 21:11 celeron55 i expect too, but so far not me nor anyone else has made one 21:11 twoelk does ogre3d have something better builtin? 21:12 celeron55 (i have stated multiple times that we should have a pure-lua gui system that we can additionally release as a stand-alone library) 21:13 twoelk hehe - can't we just hijack something? 21:13 celeron55 such does not exist 21:15 hmmmm I've been tossing around the idea of having a programatically generated GUI 21:15 hmmmm expose a set of lua APIs to create/whatever a GUI 21:15 hmmmm as well as an XML-based GUI description language 21:15 twoelk maybe a lua-markdown interpreter for an ingame helpsystem might be some start 21:16 hmmmm you know, like
Enter your text here... 21:16 twoelk ah, a lua-xul interpreter ;-P 21:16 VanessaE why not Qt? :) 21:16 * VanessaE hides 21:16 Sketch2 what about CEGUI? 21:16 twoelk actually I rather like XUL 21:17 Sketch2 it will do both Irrlicht and Ogre 21:18 twoelk does CEGUI allow skinning? 21:20 Sketch2 yeah, I believe so 21:27 twoelk hm.. supported platforms of CEGUI are win.mac,linux... hope that includes android and i-os 21:30 rubenwardy BSD, too 21:34 TBC_x tcl/tk? 21:35 VanessaE TBC_x: why not Xaw then ;) 21:35 paramat hmmmm okay if i now merge #3119 even though it's 1 commit behind? there are no files shared with MT's latest commit 21:36 ShadowBot https://github.com/minetest/minetest/issues/3119 -- Change ore chance in clusters by Gael-de-Sailly 21:36 TBC_x or something with tk-like api 21:36 twoelk http://sumwars.org/wiki/File:Sumwars_0.5.5_-Trading_a_bit.jpg <-that seems to be CEGUI driven then 21:37 paramat and any thoughts on #2899 ? 21:37 ShadowBot https://github.com/minetest/minetest/issues/2899 -- Allow doing something before the core mapgen and minetest.registered_on_generates are executed 21:38 paramat also is #1815 really an issue worth keeping open? 21:38 ShadowBot https://github.com/minetest/minetest/issues/1815 -- Incomplete mapblocks being saved to database 21:42 est31 I would agree with xul 21:43 est31 but we must not use javascript 21:44 est31 but xul is dying 21:44 est31 before long we would have to supply our own renderer 21:44 est31 or a xul → html converter 21:44 VanessaE est31: so I guess we'll be linking against webkit soon? ;) 21:44 VanessaE (or whatever is the new hotness) 21:44 est31 xul is gecko :) 21:45 TBC_x I don't think that html is flexible enough 21:45 VanessaE yep I know, but who in their right mind would choose that? ;) 21:46 est31 we could say we use servo, its a next generation web rendering engine 21:46 est31 much faster than gecko or webkit or blink 21:46 est31 we could have some layer of abstraction, by using CEF 21:47 est31 CEF is an API implemented both for servo and chrome 21:47 est31 TBC_x, whats more flexible in your opinion? 21:52 VanessaE all right this is pissing me off now 21:54 VanessaE these OOM errors have GOT to stop. 21:55 VanessaE /clearobjects apparently does NOT clear all objects from the world. 21:56 VanessaE (either that or my mods are just quicker than me at re-creating their missing entities before I can restart the server) 21:56 VanessaE how am I supposed to manage this? 22:00 twoelk black magic 22:01 VanessaE at this point, these fake OOM issues are the only things crashing my servers now. 22:01 est31 celeron55, did you know that rubenwardy has added a patch to customize the map limit? 22:02 est31 it would enable to make smaller maps than the current 31^3 22:02 est31 31k* 22:06 VanessaE all right I'll say it plain: am I gonna be forced to stop using luajit because of these OOM errors? 22:06 * twoelk hopes that some mapgen will soon allow to define what the map edges should look like - some barier. sheer cliffs, high mountains, an endless sea, a jungle too full of monsters to get through or some void for example 22:07 VanessaE twoelk: an endless sea makes the most sense imho 22:07 VanessaE other games tend to draw tall mountains and just prevent the player from walking to them 22:08 VanessaE (like a skybox blended into the terrain or something) 22:08 TBC_x anyone played motocross madness? 22:08 VanessaE nope. 22:08 twoelk would be the easiest to simulate travelling to another server if that ever is supported 22:10 TBC_x https://youtu.be/jLg2hRSUoL8 22:11 TBC_x I have a draft somewhere of how to implement multiworld capability 22:11 TBC_x but it requires significant architecture rework 22:12 est31 I prefer the approach thats least intrusive 22:12 est31 but multiworld is a pretty basic change 22:13 est31 so it will be pretty intrusive I guess 22:13 est31 we should also discuss how to implement Voxel Area Entities 22:13 VanessaE hmmmm: something of note: after I did that /clearobjects, the server seems to be less-inclined to OOM-crash. mean anything to you? 22:13 VanessaE it took 3 hours to run that command, btw. 22:14 twoelk on VE-basic? 22:14 VanessaE on VE-Survival. 22:14 twoelk that's a lot of map to check 22:14 TBC_x that change would also allow pretty sick single-world multi-server 22:15 VanessaE TBC_x: single-world multi-server? as multiple servers on a single world? 22:15 Sketch2 what if objects were stored in a seperate database? 22:15 TBC_x VanessaE: yes 22:16 Sketch2 then to clear objects would essentially just delete that file 22:16 paramat VanessaE, after clearobjects do you get this bug #1397 ? 22:16 ShadowBot https://github.com/minetest/minetest/issues/1397 -- Mapgen double-generating mapblocks, destroying user-built structures 22:16 paramat currently it's not certain that bug still exists 22:16 Sketch2 it would be nearly instant 22:16 TBC_x and also smooth nexus implementation 22:16 VanessaE paramat: I have never experienced that issue. 22:17 VanessaE TBC_x: I like that idea indeed. 22:17 est31 wtf is "smooth nexus" 22:17 VanessaE est31: high-availability minetest servers 22:17 paramat okay good 22:17 VanessaE or that's how I read it anyway 22:17 TBC_x nexus, an area from where you can access multiple worlds or servers 22:17 VanessaE oh 22:18 TBC_x and that could be done smoothly, so a player wouldn't even notice that he has been teleported 22:18 twoelk ah so nexus is a mainmenu in disguise 22:19 TBC_x well... If you have a server with multiple worlds... 22:19 twoelk if you do it smoothly, when do you want to load all the media? 22:19 est31 thats the problem indeed 22:20 est31 all the initialisation and block sending needs to be done very fast 22:20 est31 so that the user doesnt notice 22:20 est31 right now its a slow mess 22:21 TBC_x the best thing, all work is done server-side 22:21 est31 right now, you need to wait for media to load even if you think things have been stored in cache 22:21 est31 and should be fast 22:21 twoelk if servers have different sets of mods it will always last longer than what can be called smooth 22:21 est31 then the "initializing blocks" step 22:21 est31 damn slow as well 22:22 TBC_x one way is to load all media for all servers 22:22 TBC_x which could work if the servers aren't that diverse 22:23 VanessaE only way THAT could work is if servers submit their media to a central database that all clients try to read from 22:24 TBC_x yes 22:24 TBC_x that is in that draft, I should have mentioned it 22:24 Sketch2 I mentioned it and got yelled at for it 22:24 twoelk pff- everybody just needs to preload dreambuilder 22:24 Sketch2 I was told it causes issues for the owner of the file server 22:25 Sketch2 legal issues because they area liable for the content 22:25 VanessaE twoelk: HAH! that would only cover maybe 1/10th of all the media out there :P 22:25 TBC_x a torrent protocol would make sense here 22:26 VanessaE suddenly, minetest is distributed on a DVD 22:26 Sketch2 perhaps 22:26 VanessaE torrent would make sense, yes 22:26 Sketch2 I think the problem was that the files needed to be encrypted 22:26 VanessaE but wouldn't really be any faster for a lot of people 22:26 TBC_x yeah 22:27 VanessaE imagine having to pull 200 MB of media on first-connect 22:27 Sketch2 so that the keeper of said files would have no way of knowing what they were 22:27 paramat now merging #3119 22:27 ShadowBot https://github.com/minetest/minetest/issues/3119 -- Change ore chance in clusters by Gael-de-Sailly 22:28 twoelk having a multiplayerworld over the internet where every player can change every node is sort of crazy from the start and might actually be some sort of hybris 22:28 TBC_x MT needs a way of dynamic resource loading 22:29 est31 TBC_x, if you mean by "dynamic resource loading" that we do the "initializing blocks" step at first sight, and not at joining the game, I fully agree 22:29 est31 it would be a way to avoid the problem 22:29 est31 even the "loading media" step could be done at first sight 22:30 Sketch2 or during low points 22:30 TBC_x resources shoud be imho loaded in a similar way of mapblocks 22:30 Sketch2 where the server isn't busy 22:30 twoelk est31: btw are you aware of the auth.txt corruptions on the just test server (hoping my memory doesn't fail me and I am addressing the right dev)) 22:31 est31 twoelk, auth.txt corruptions? 22:31 VanessaE hmmmm: ok, yeah. /clearobjects definitely improved the server fault. and my client FPS is a bit higher than usual, even at the spawn. 22:32 twoelk yeah the auth.txt became useless 3 times by now which ended in a lot of nick hijacking 22:32 TBC_x loaded and cached 22:33 paramat merge complete 22:33 est31 paramat, you realize 3119 did modify noise? 22:33 est31 Ive thought this is bad, no? 22:34 paramat erm 22:34 est31 twoelk, might be a bug with saving files 22:35 twoelk https://forum.minetest.net/viewtopic.php?p=188961&sid=8ebd9cc0352bbbc3a088d475ca9192c3#p188961 22:36 paramat 3119 doesn't touch noise 22:37 est31 ore noise 22:37 est31 well, shrug 22:37 est31 you know much more bout mapgen 22:38 paramat all it does is fix a rounding bug 22:38 est31 oh 22:38 twoelk I think the auth.txt trouble on the Just-Test server began when the server updated to a version with the new password system 22:38 paramat the number of pseudorandom calls is the same 22:39 est31 and yeah, now as I read it it makes actual sense 22:39 paramat phew cool 22:40 est31 twoelk, does the just test server use latest minetest git, or does it use stable releases? 22:40 est31 and how regularly does it update? 22:40 est31 the new password system exists since months now 22:42 est31 since 8dbf68331386f5f6a40ff0f0a77f5390bdca60b9 to be precise 22:42 twoelk the problems on jt started around the beginning of august iirc 22:43 est31 does it appear after crashes/shutdown? 22:43 est31 it might be that the file is incompletely written, when minetest crashes 22:43 est31 and what are the precise symptoms? 22:45 twoelk https://forum.minetest.net/viewtopic.php?p=186537#p186537 only mention of an update :( 22:47 twoelk lag describes the corrupted auth.text 3 posts earlier 22:48 est31 "Some accounts from beginning of original file are left, also some small number of records from rest of original file is randomly added to this." 22:49 est31 hrmmm 22:53 twoelk I think I heard of similar things on other servers but Just-Test is the only one I noticed the social drama caused by this issue. Might be because it is rather hardcore survival and therefore the chat is full of accusations of people hacking accounts. 22:54 est31 it seems just test is an anarchy server, and there its more common to have accounts hijacked 22:55 est31 but I wonder how that bug occurs 22:56 est31 okay I guess celeron55 can deploy Calinou's website 22:56 est31 I haven't heard anybody speaking against 23:06 est31 hrmmm, I guess we should remove this https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_mainmenu.h#L122 what do you think Tesseract? 23:07 est31 seems to be your fault from 8f9af57314f71aae1cc77e 23:10 twoelk I wouldn't say just test is anarchy but simply rather harsh 23:14 est31 gonna push in 20 minutes: https://github.com/est31/minetest/commit/a9f5126ee295adea9711830123b45d48663e1288 23:17 est31 okay reworded, look here instead https://github.com/est31/minetest/commit/9c44aace4fce5f4b496201214d3215899f1481c3 23:34 est31 pushed 23:35 paramat game#654 23:35 ShadowBot https://github.com/minetest/minetest_game/issues/654 -- Default: Allow papyrus growing on sand group by paramat