Time Nick Message 00:49 Hijiri I use emacs 06:41 sofar nerzhul: I wrote some code to do CSM fuzzing 06:41 sofar interested? 06:42 sofar I just fuzzed the crap out of my server :D 06:54 nerzhul i will go to work, but show me ? :p 06:54 nerzhul what did you fuzz exactly ? 06:54 nerzhul CSM is here for a such case 11:26 Megaf #6476 11:26 ShadowBot https://github.com/minetest/minetest/issues/6476 -- Debian Stretch 11:26 Megaf +possible close? 16:29 paramat Krock rubenwardy please can you review game#1881 ? =) 8) =) it's almost identical to the map mod codewise, but see my last comment 16:29 ShadowBot https://github.com/minetest/minetest_game/issues/1881 -- Add binoculars mod with binoculars item to allow zoom by paramat 16:46 Krock needs more smilies but ok 16:49 sapier Hello, I wonder why the upscaling stuff was integrated to plain drawing method and not as separate draw function. Does anyone know? 16:50 Krock lots of dupe code? Dunno 16:50 Krock paramat, works fine 16:51 sapier actually now about 75% of code is upscaling ;-) and the rest is about 10 lines ;-) 16:52 sapier I stumbled uppon it because somehow all 3d modes got broken but whatever broke them doesn't seem to be in drawing cod 16:52 sapier e 16:57 Krock sapier, fix is on the way #6253 16:57 ShadowBot https://github.com/minetest/minetest/issues/6253 -- Rewrite rendering engine by numberZero 16:57 sapier krock are you sure because the code which was within rendering engine doesn't seem to cause it 16:59 sapier well at least I should talk to numberzero because he's working at exactly same code I intend to touch 16:59 Krock That code area isn't my speciality but the PR improves the 3d modes overall 17:00 sapier Yes I'm looking at it right now looks promising 17:03 sapier let's see if it already works again on his branch I wouldn't have to find out what broke it in this case 17:07 sapier yes it does, so forget about the previous comments ;-) 17:10 paramat thanks 17:11 paramat sapier is it possible you can add a joystick control to android? controls really need improving 17:12 paramat sapier see https://github.com/minetest/minetest/issues/1750#issuecomment-331305403 and linked discussion 17:13 sapier everything is possible but I don't even have a working android environment any longer 17:13 paramat oh ok heh 17:14 sapier atm my primary focus is making minetest vr ready :-) 17:14 sapier that's why I'm looking at the draw/3d code too atm 17:14 paramat it looks like lordfingle can help us with android in early 2018 17:15 paramat he has his own android fork and seems capable 17:15 sapier that'd be good I'm quite sure there's a lot to improve in android. I didn't look at it for quite some time but I guess it hasn't progressed a lot since the time I added the rudimentary basic controls? 17:16 sapier btw I was surprised to see gamepad support in minetest and it even works (mostly) :( 17:16 sapier :-) 17:21 paramat indeed android is neglected, many issues, that controls issue is 3 years old 17:22 sapier well android is closely linked to fixing formspec that's why I gave up ... did anyone change at that issue? 17:22 sapier anything not anyone of course 17:23 paramat not sure but i expect formspec is still an issue 17:23 sapier well I'm still hoping to avoid it completely in vr :) 17:24 rubenwardy the main problem on android is even before you open formspecs 17:25 rubenwardy both movement and special things (fly/use/fast/etc) 17:25 rubenwardy formspecs are still a big issue though 17:26 rubenwardy but not the main issue on Android 17:26 rubenwardy also the build system really really really sucks 17:26 sapier rubenwardy: yes you mention another problem ppl expect minetest on android to be exactly same as on pc 17:26 sapier that's not gonna work 17:26 Fixer also, fdroid still has 0.4.15, don't ignore this 17:26 rubenwardy other apps do it fine 17:27 rubenwardy people use other apps as the official is bad in controls 17:27 sapier rubenwardy: nope I don't know of any desktop app ported to mobile without changes 17:27 sapier and those mobile apps ported to desktop usually work but are stupid ... see windows 8 ;-) 17:27 rubenwardy other apps -> other minetest-based android apps 17:28 sapier rubenwardy: I don't say our controls aren't bad ;-) 17:28 sapier there's lot of room for improvement but to get it good you'll have to do some changes to what it's like (compared to desktop) 17:28 Fixer rubenwardy: what the problem with controls? i've played with original and ... controls are controls 17:29 sapier well at least c++11 solves the 0 NULL null discussion for what I see :) 17:31 paramat proller does freeminer have better android controls than MT? 17:32 paramat (we can use their code now) 17:32 sapier did minetest switch to gpl? 17:32 rubenwardy no 17:32 rubenwardy proller switched to LGPL 17:32 rubenwardy or at least gave permission 17:33 sapier still we'd have to look if that code was added either from him or while it was alread lgpl 17:33 paramat fm doesn't have a joystick movement pad 17:33 Fixer proller switched to death fork 17:33 Fixer ._. 17:33 Fixer dead* 17:33 sapier sad 17:34 sapier even if I personally don't like proller I still respect his knowledge he did at least trigger some good things 17:37 Fixer i would rather see him commiting to original 17:37 Fixer the way mt dev was before recent time was very elitist-cancerous 17:37 sapier yes once he did 17:38 sapier well a lot of big egos ;-) (including mine for sure too ;-)) 17:39 Fixer it was also very closed and bu*rocratic 17:40 lhofhansl Any further comments on #6483? 17:40 ShadowBot https://github.com/minetest/minetest/issues/6483 -- Retrieve extra speed-directed, narrow view cone to load blocks the player will encounter soon. by lhofhansl 17:42 sapier well most important is there seem to be more ppl working at it atm 17:44 Fixer except MTG, it needs even more people 17:46 sapier well not all things are better "#pragma once" 17:49 sapier even if all (major) compilers support it it's still non standard and may always be changed or have different meaning 17:50 sapier guess someone told it's c++11 ;-P 17:50 nerzhul sapier, except if you use non supported compiler which doesn't support C++11, you will not have this problem 17:51 sapier nerzhul: so you say it's c++11 ? ;-) 17:54 sapier well I found the voting ... obviously you know it's non standard 17:58 paramat nerzhul or anyone ok with #6483 ? 17:58 ShadowBot https://github.com/minetest/minetest/issues/6483 -- Retrieve extra speed-directed, narrow view cone to load blocks the player will encounter soon. by lhofhansl 18:07 Fixer nerzhul: are you planning to move auth.txt to database? 18:08 Fixer so people don;t need to use this https://forum.minetest.net/viewtopic.php?f=9&t=18604 18:09 red-001 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18:09 Fixer nerzhul: performance improvement seems huge, people report much less server lag 18:10 rubenwardy the player database kills performance for me, so I'm using the dummy backend. It's probably just flushing too much 18:31 nerzhul rubenwardy, i think it's due to sqlite 18:31 nerzhul we should be less aggressive and use timed save for players 18:31 rubenwardy it shouldn't be that slow 18:31 nerzhul no it shouldn't be , firefox use far more sqlite than use 18:31 nerzhul use 18:31 nerzhul us* 18:31 nerzhul sapier, for such changes we vote yes, many things changed 18:32 nerzhul Fixer, i'm working on network rework, i planed this but it's not in my radar ATM, if someone wants to contribute to create this new player database it shouldn't be difficult as the PlayerDatabase introduced new db model in core permitting to plug as many functional backends as we wanted 18:32 sapier nerzhul: well at least things are done I would've voted against but considering the count this wouldn't have changed things 18:33 nerzhul exact :p 18:33 sapier I usually don't like doing changes for the only reason of changing 18:33 nerzhul and we don't care about non supported systems 18:33 nerzhul android, windows, clang, gcc support it , it's our platforms 18:33 nerzhul and all c++11 projects are moving to use it it's by essence standard, just not a norm :) 18:34 sapier I've heared things like that a lot of times ... usually things ended up in doing a lot of additional work because of something "unsupported" hat to be supported too 18:34 sapier but of course the pragma change is most likely not very likely to be such a cause 18:35 sapier less then a two years ago swift was the in-language itself .... everyone was to be programming swift in a couple of years ... well .. do you know what swift is? 18:36 nerzhul we don't recoded minetest in another language it's not the same change heh, we just follow industry quality grade standards 18:36 sapier same for javascript about 5 years ago ... each new version of it gets more and more c++ish 18:36 nerzhul but it's not c+ it's... js :p 18:36 sapier industry qulity grade? *smile* 18:36 rubenwardy JS hardly gets more C++ 18:37 rubenwardy it gets more functional 18:37 sapier believe me nerzhul in industry you'd have been removed for a change like that ;-) 18:37 rubenwardy no templates 18:37 rubenwardy also, swift is very popular in iOS / Mac OS 18:37 rubenwardy it's never really been popular elsewhere, and doesn't really support anything else 18:38 rubenwardy and also breaks a lot 18:39 sapier ok I think i need a new ide ... everytime I run eclipse I get constant freezes 18:39 lhofhansl I use Emacs with ctags :) 18:40 sapier well ... I was thinking about something less ... rusty 18:40 Krock go for vim 18:40 sapier less!!! 18:40 sapier :-) 18:40 Krock what less? atom? 18:40 lhofhansl I think I am the only one who uses Vim, Emacs, and Eclipse... Depending on what I am doing. 18:40 sapier swift is dieing ;-) 18:41 Krock you could also try "wine notepad.exe" /s 18:41 lhofhansl :) 18:41 sapier I stopped using vim at all when debian broke it's default config 18:42 lhofhansl Hey. Any comment on #6483? It ended up pretty simple, other than the tracking of attached objects. 18:42 ShadowBot https://github.com/minetest/minetest/issues/6483 -- Retrieve extra speed-directed, narrow view cone to load blocks the player will encounter soon. by lhofhansl 18:42 sapier nerzhul: in industry you may not even remove 20 year old bugs ;-) so don't try to use industry standards in minetest ;) 18:42 Krock how accurate is the player speed on server-side? 18:42 sapier btw there are good reasons not to remove them there but industry is a completely different usecase then minetest 18:43 lhofhansl If the player is not attached it's whatever the client reports (which is correct). If the player is attached the speed is retrieved from the attached objects directly on the server, which is also correct. 18:43 bitofabyte_ sapier: I've been using Clion for the past few hours and it seems pretty nice so far 18:43 sapier clion? 18:44 lhofhansl Only question is: Are there objects other that Lua Objects that a player can attach to? (and even then it would only that this optimization is not used) 18:44 bitofabyte_ it's non free, from JetBrains 18:44 Krock ^ nerzhul loves it too 18:44 lhofhansl other *than* 18:44 sapier well as I'm using linux non free software (especially without open source code) ain't an option 18:47 Krock lhofhansl, currently there are no other types in use https://github.com/minetest/minetest/blob/master/src/activeobject.h#L35 18:50 lhofhansl Then this should be good to go. :) Helps a lot and only retrieves a small amount of extra data from the server (and that only if the speed direction lies outside of the player's view cone). 18:52 Fixer sapier: it is just people should be more realistic and stop beeing hip and such and remove that excessive hype, hype comes and goes and you end up with this MESSSSS 18:53 * Krock reads MEMESSSSS 18:53 Fixer beeing 18:53 sapier Fixer: yes realistic would be good ... especially young students ... I don't think I'm old now but I already see the second cycle of learning :-) 18:54 sapier first programmers think what the old ones did was way to complex and not required at all 18:54 sapier design their new stuff removing all the complex useless stuff (javascript) 18:54 Fixer i'm more worried by people who tell me I'm dinosaur while removing basic features from software or doing it in very anal way 18:55 sapier and then they realize those old guys haven't been wrong they just did already know what will happen 18:55 Fixer sapier: what you are saying is not new 18:55 sapier e.g. type support which is added back to javascript atm 18:55 Fixer sapier: this very same argument was in 80s iirc 18:55 sapier true :-) that's what I say it's not new ;-) 18:55 Fixer it is just revolved 18:56 sapier every new generation seems to only add additional overhead ;-) 18:56 Fixer improved hardware 18:56 Fixer on the other hand there is not much improvement area left, apart from many cores 18:56 Fixer and I want software to be fast, not fancy 18:57 nerzhul sapie ruses clion yes 18:57 Fixer i'm perfectly fine with Win9X/NT interface, just don't ruin program functionality 18:57 sapier well I don't see improvements .... everything used up by replacing bare metal by multitasking os then replacing os by single threadded webos now adding multithreading back to browsers ... everything running right on top of an operating system which is actually capable of doing same stuff ;( 18:58 sapier of course step from bare metal to os was usefull ... especially as not every architecture did support the isolation stuff 18:58 sapier but why the hell does a browser have to replace an operating system ;-) 19:01 Fixer sapier: some will say that browser is kinda emacs of XXI century, and it is very portable shitty code 19:02 sapier I see the portability when I try to run some web pages on linux 19:02 Fixer browser is OS in OS 19:03 sapier btw that OS in os does get more and more hw access this it's isolation is geting more and more weak ... well guess that's almost irrelevant as any valuable data ain't stored on the pc but in cloud anyway 19:03 paramat hi lhofhansl good to see you here. if you PM c55 about becoming a core dev you can then +1 your PR 19:13 paramat ~tell ThomasMonroe please could you keep your IRC quit message very short? 19:13 ShadowBot paramat: O.K. 19:14 Fixer sapier: kinda dislike that cloud hype, it is just damn hosting more or less, you should use local backup (and other backups) as well 19:15 sapier fixer hosting is only one part the other one is fencing ... if you're once within the fence of someones walled garden it's hard to leave 19:15 Fixer just don't join it and you will be fine 19:15 sapier Fixer: that's getting more and more hard 19:16 Fixer your wallet is your voice 19:16 red-001 paramat, can't you set your client to ignore them? 19:17 Fixer paramat: what IRC client are you using? 19:17 sapier well fixer for some things your wallet forces you to join the walled garden because you just can't pay to be out there 19:18 Fixer yeah 19:20 paramat yes i know i can, but quit messages should not be long anyway, they're not forum signatures 19:21 paramat it should be possible to have them enabled without multiple lines of spam 19:22 Fixer sapier: there are still workarounds around fences, thanks to hackers 19:22 sapier well it's always a matter of how much time and money you spend 19:32 red-001 paramat, do mapgens use g_settings? 19:32 red-001 asking because all of them claim to, but don't seem to 19:33 sofar nerzhul: https://github.com/sofar/minetest/commit/c639a9ddbcf456edfa8d1eb25ebad8eeca34c321 19:33 Krock g_settings is a fallback if none was found in the world config 19:33 sofar nerzhul: essentially, creates a new lua api: minetest.send_packet(data) 19:34 sofar and adds it to CSM - lua code example included 19:35 Krock sofar, trivial mistake. this function does not have any return value https://github.com/sofar/minetest/commit/c639a9d#diff-7b00a65f007af40388e565e5c34d8461R363 19:35 red-001 incoming people asking for it to be removed because sending chat or something 19:36 sofar Krock: this is really *not* suitable for mainline inclusion, if you ask me 19:36 sofar I mean, there's zero validity checking on `s` 19:36 sofar if you pass it an empty string, MT client crashes 19:37 sofar if you pass it a *data with incorrect command nibble, it crashes 19:37 sofar etc. 19:37 sofar more over, it will DDoS the server most likely 19:37 sofar (as it had in my tests) 19:38 red-001 btw I'm pretty sure NO_MAP_LOCK_REQUIRED is a no-op for csm 19:38 sofar afk lunch, and then we can discuss how I wrote insecure code that crashes all servers 19:40 Fixer crash all the things* 19:47 lhofhansl paramat: :) (sorry was off in meetings for a bit) 19:51 paramat no problem 19:52 lhofhansl c55 does not appear to be around 19:54 red-001 !seen celron55 19:54 ShadowBot red-001: I haven't seen celron55 in #minetest-dev. 19:54 red-001 !seen celeron55 19:54 ShadowBot red-001: I saw celeron55 in #minetest-dev 3 days, 8 hours, 25 minutes, and 25 seconds ago saying "it's a good thing lua is really as small as it can be; there's not a lot of room for security issues" 19:55 rubenwardy red-001, NO_MAP_LOCK_REQUIRED is always a no op 19:55 paramat he's probably just busy, might reply to you later 19:56 red-001 rubenwardy, then whats it's point? 19:57 rubenwardy no idea 19:57 rubenwardy https://github.com/minetest/minetest/blob/921151d97a2fb2238ab514324fb95e2732248b96/src/script/lua_api/l_internal.h#L36 19:57 rubenwardy both empty defs 19:57 rubenwardy future proofing maybe? 19:57 rubenwardy or lock removed? 20:01 red-001 looks like it was always a no-op on windows 20:01 red-001 and on linux it was part of the profiler 20:02 nerzhul sofar, a such thing will never happen 20:03 nerzhul we can do anything but it's the high security hole you can add to MT 20:03 nerzhul and yes the current network stack is not crash safe :p 20:04 nerzhul rubenwardy, if i remember kwoelr said it's used as an annotation mor ethan code 20:04 red-001 I would assume that the function is meant to demostrate that 20:04 nerzhul red-001, ofc 20:05 red-001 spliting the server and client config files is a lot easier and a lot more time consuming then I expected 20:29 sofar nerzhul: I want to use the fuzzer to expose formspec security holes though 20:29 sofar nerzhul: having audited insidethebox formspec handling I noticed there is an insane amount of exploitable surface 20:30 sofar any formspec you offer to players has some easy vector to gain potential access or destroy something you shouldn't be able to 20:31 rubenwardy you should definitely validate anything submitted 20:31 rubenwardy don't trust user input 20:31 rubenwardy or client send data rather 20:31 sofar my input validation code is excessive 20:31 sofar for itb 20:32 sofar I'm sure it's possible to do stricter validation in the core 20:32 rubenwardy if you put form sending and result receiving you can 20:32 rubenwardy like in sfinv 20:33 rubenwardy like, minetest.register_window with a formspec builder, a result receiver, and a option to validate privs 20:33 sofar that's only half the problem 20:34 rubenwardy you could offer officially supported server side meta (which I call contexts because why not) 20:35 rubenwardy which could also store the expected parameters to be received on submit, and be used to validate 20:35 rubenwardy but at this point why not just remake formspecs 20:35 sofar well any updated solution would have to do this validation as well 20:35 sofar e.g. send a formspec with elements X and Z, and on receive reject any data that isn't X or Z 20:38 red-001 sofar wouldn't it be easier then to test this by adding a function to send formspec data? Then you wouldn't have to make sure the packet is valid 20:39 red-001 just send random data using that 20:39 sofar yes, but then you'd have to make more functions for the other things that can be fuzzed too 20:39 sofar :) 20:45 nerzhul sofar, then go PR to patch that 20:45 nerzhul it seems you found the holes 20:47 red-001 formspec handling is a something thats handled by mods right now 20:48 red-001 not sure if you could even add any useful safety checks without re-designing quite a lot of stuff 20:49 nerzhul yeah, maybe it should be done at a point 20:50 sofar without some sort of server-knows-clients-state it can't be done anyway 20:51 sofar my bet is that you can just right now send formspec results to any server and it gets processed 21:16 nerzhul sofar, yeah there is no token to tell it's a LGTM request 21:16 nerzhul it's exact 21:31 sapier https://imgur.com/a/er2wU first step ;-) 23:43 Megaf paramat, https://github.com/minetest/minetest/issues/6476#issuecomment-336310996