Time Nick Message 00:16 Fixer i got 1000 mapblocks in range, 450 drawn, 1.62+E6 vertices drawn, v_range 160, 24fps, is this expected for spawns? I don't remember such high numbers before 00:17 Fixer what is worrying is 1000 mapblocks 00:17 Fixer wait 00:18 Fixer thats 10x10x10 mapblocks, which is corresponds to my v_range 160 00:18 Fixer guess it is fine 00:20 Fixer nevermind, everything is fine 00:20 Fixer those water cities is a biiiig mistake, shittons of unneded stuff rendered 00:21 Fixer in my tests by replacing that water with solid ground you can double your fps 02:53 paramat will merge game#1054 game#1392 soon 02:53 ShadowBot https://github.com/minetest/minetest_game/issues/1054 -- Keys - allow easy sharing of access without commands by sofar 02:53 ShadowBot https://github.com/minetest/minetest_game/issues/1392 -- Improve arms and orientation of character model by Rogier-5 02:55 paramat in fact, in a moment 03:05 paramat now merging 03:13 paramat complete 08:49 hmmmm jeez okay 08:49 hmmmm i got highlighted a bunch, what's this all about 08:57 hmmmm what 08:57 hmmmm no, a reference can't be null 08:57 hmmmm that's part of the reason for having references 08:57 hmmmm each thread has its own stack, so it can't be a "race condition" either 08:57 hmmmm no other thread can access this data 08:58 hmmmm wtf are you smoking zeno` 08:59 hmmmm this is clearly a case of either a). stack corruption or b). your debugging info being bogus 08:59 hmmmm you need to drop down to -O0 to get any meaningful backtrace 09:27 Zeno` hmmmm, well if that's the case WTF is the point of "Release with Debug" info builds? 09:28 Zeno` the term "null reference" was probably a mistake. It's a synthetic pointer and a reference created by deferencing a pointer, so it certainly can be NULL 09:38 Zeno` you most certainly *can* debug a release build as well. You just can't do it without the debug info being present because things will point to the incorrect lines 09:57 Hijiri dereferencing null pointers is undefined behavior 09:58 Zeno` well, yes 09:58 Zeno` and usually a crash 09:58 Hijiri well, the reference still isn't null, it's just buggy and undefined 09:58 Zeno` but I'm looking at trying to get a backtrace with -O0 now, even though it's not necessary 09:59 Zeno` what is a "synthetic pointer" btw? 09:59 Hijiri let me check stack overflow 09:59 Zeno` that is what valgrind is calling it... it normally calls things either a pointer or a reference 09:59 Zeno` I've never seen the term "synthetic pointer" 10:01 Zeno` stack overflow (first result) is a clear as... mud 10:01 Zeno` hehe 10:02 Hijiri I think it says that the value isn't actually a pointer, it's just called one by the debug info 10:02 Hijiri and it's actually optimized out 10:02 Zeno` maybe it's suggesting that the object was optimi... 10:02 Zeno` yeah 10:02 Zeno` hmm 10:02 Zeno` dat's odd 10:02 Zeno` the reference is *needed* 10:02 Zeno` this might be quite difficult to find :/ 10:03 Hijiri if it's a synthetic pointer I'm guessing that means it cannot be null 10:03 Zeno` it was my fault for using that term; a mistake 10:03 Zeno` at the time I had not seen the "sythetic pointer" mentioned 10:03 Hijiri I haven't seen it either 10:04 Hijiri so I at least learned something from you mentioning it 10:04 Zeno` I won't change the comment though... I don't like changing comments that are more than about 10 minutes old and never if people have responded 10:04 Zeno` the reason I said null pointer is that what gdb called it when I tried to examine its value 10:04 Hijiri If nobody has responded I might just do a strikethrough and put corrected stuff underneath 10:04 Zeno` but then I read the full trace 10:06 Zeno` if it's caused by a compiler optimisation, then how to debug? 10:07 Hijiri why not compile with -O0 like people suggested? 10:07 Zeno` it's like the dinosaur or the egg problem 10:07 Hijiri oh 10:07 Hijiri caused 10:07 Zeno` because then the object will not be optimised away 10:07 Hijiri So you tried it without optimizations and it didn't crash? 10:07 Zeno` am trying now 10:08 Zeno` and so far (sitting in the area where the crash happens) after 20 minutes it hasn't occurred 10:08 Zeno` with optimisation it will happen within 5 minutes 10:08 Hijiri means either the bug is caused by undefined behavior, or (less likely) a bug in the optimizer 10:09 Hijiri If I was debugging I would just leave it overnight sitting there just in case 10:10 Zeno` yeah I plan to 10:10 Zeno` I wonder if it's because the function it calls is an inlined class member 10:11 Zeno` doesn't make sense, but this is new territory for me. Until 1 hour ago I'd never even heard of a synthetic pointer 10:11 Hijiri I doubt that's the issue, calling inlined methods is not undefined behavior 10:12 Zeno` yeah I know 10:12 Hijiri and it's a common enough use case that a bug in the optimization there would probably be found quickly 10:12 Zeno` my maybe *that* class is being optimised away for some reason. None of this makes sense to me 10:12 Zeno` I admit that I am in new territory and don't really know how to proceed 10:13 Zeno` unless I can get a backtrace with -O0 but these "synthetic pointer" descriptions are scaring me 10:14 Hijiri LightQueue? 10:14 Hijiri it looks like a struct, not a class 10:14 Hijiri unless structs are classes in C++ 10:14 Zeno` inline void push(u8 light, const relative_v3 &rel_pos, ..... 10:15 Hijiri looks like part of LightQueue 10:15 Zeno` wait... the queue is a struct? 10:15 Zeno` it doesn't really matter... there is no difference between class and struct except for default permissions 10:15 Hijiri yes, I just found out on SO 10:16 Zeno` celeron55 would actually prefer us to never use class and always use struct (or he used to) 10:16 Zeno` there's no difference (apart from those default permissions) 10:19 Hijiri I should probably sleep, good night 10:22 Zeno` night 10:22 Zeno` I got it! 10:22 Zeno` I got it! 10:22 Zeno` yay 10:23 * Zeno` does happy dance 10:34 Zeno` hmm 10:34 Zeno` it's still weird 10:34 Zeno` lol 13:53 Zeno` `op 13:53 Zeno` ;op 13:54 Zeno` ;mop my floor 16:34 Zeno` hmmmm, new backtrace is up 16:54 hmmmm this is a slightly different crash than the last one 16:54 hmmmm either way, malloc() is crashing. malloc() can crash if the heap is corrupted in a certain way. 16:54 hmmmm run valgrind 17:21 Zeno` hmmmm, I'm tired. It will probably take a while :) 17:21 Zeno` why don't you? :P 17:22 Zeno` I guess I can do it tomorrow *sigh* 17:25 hmmmm i'm not doing minetest atm 17:29 Zeno` Fixer, you know the 0,0,0 bug 17:29 Fixer Zeno`: and he knows you 17:30 Zeno` on Redcat creative if you don't climb the ladder and see the sign will you still become logged in eventually? 17:30 Zeno` if not, then I cannot possibly run valgrind on this bug because it would take me hours just to get the to of the ladder 17:31 Zeno` I'm not going to spend hours climbing a ladder; we'd be better off reverting all the commits since the last release 17:32 hmmmm how often does this crash really happen? 17:32 Zeno` hours climbing a ladder (where I have to be present) and then hours more waiting for the login to happen and then hours more 17:32 Zeno` hmmmm, with release build 100% of the time 17:32 Zeno` within 2 or 3 minutes of being at the location 17:33 Fixer Zeno`: you wait hours to get in RedCat? holy shit, it is up to 15 minutes for me 17:33 Fixer before real login 17:33 Fixer you can walk on server, watch stuff 17:33 Zeno` Fixer, I mean it would be hours (and hours) if I was running the client under valgrind 17:33 Fixer ah 17:34 Zeno` it would literally take an hour (or more) to climb that ladder from 0,0,0 17:34 hmmmm have you seen this happen with a debug build? 17:34 Zeno` and then there's the waiting for the login process etc etc 17:34 Zeno` hmmmm, yes, the last backtrace is with -O0 17:34 hmmmm er wait, nevermind 17:34 hmmmm i forgot that -g on clang/gcc only adds debug symbols 17:35 hmmmm is there some kind of option that pre-fills memory with marker values on initialization? 17:35 red-001 could the duplicate label be removed from #4474 as the other PR was closed? 17:35 ShadowBot https://github.com/minetest/minetest/issues/4474 -- Avoid sending a chatmessage that a player joined if in singleplayer mode by red-001 17:38 Zeno` since it happens within seconds to minutes if compiling with "release with debug info" perhaps I can bisect it 17:39 Zeno` but at this point, unless someone buys me a supercomputer, valgrind is outa the question 17:39 Zeno` well, I don't even need to the debug info to bisect 17:39 Zeno` maybe that's a start 17:40 Zeno` to have the* 17:41 Zeno` it might work 17:41 Zeno` :3 17:42 Zeno` BUT if I find the offending commit I'll probably suggest reverting it completely. This will take a whole day at least and after I find it I'm probably not going to want look further 17:43 Zeno` I don't even know if I have the energy to commit to a whole day of bisecting :( 17:44 Zeno` *sigh* 17:44 Zeno` such is life I guess 17:49 Fixer Zeno`: you are bisecting crash or 0 0 0 ? 17:49 Zeno` the crash 17:49 Zeno` well, if I do bisect 17:50 Zeno` I'm not even sure how to start diagnosing the 0,0,0 yet 17:50 Fixer Zeno`: easiest thing is to do bisect, login to RedCat and minimize the thing 17:50 Zeno` Fixer, of course it's easy. But it takes time. And probably a lot of time 17:51 Fixer just minimise the damn thing, make sure you are viewing that fountain thing 17:51 Zeno` and I have to do that... what... 50 times? 17:52 Zeno` I cannot reproduce it locally 17:52 Zeno` what fountain thing? 17:53 Zeno` I cannot see a fountain at the bottom of the ladder 17:55 Zeno` I can't bisect the 0,0,0 issue 17:55 Zeno` because it doesn't seem to happen locally 17:59 Fixer Zeno`: wait, where it crashes? at that coordinates or at 0,0,0? 18:01 Zeno` no 18:01 Fixer Zeno`: try running 0.4.14 stable, if it not crash, bisect should take 5-6 times at most, don't build debug, build the way it crashes 18:02 Zeno` I have to wait before I get away from that location 18:02 Zeno` yes, but that takes up to 10 minutes to get to where the crash occurs! 18:02 Fixer ah, ok 18:02 Fixer is that place with water fountain? 18:02 Zeno` there is a waterfall 18:02 Fixer right 18:03 Zeno` as for the 0,0,0 bug.... hmmmm, can you apologise to VanessaE? 18:03 hmmmm ...? 18:03 Zeno` it happens on her servers and she can be relied on to provide logs 18:04 Zeno` she says she left because of something you said to her 18:04 hmmmm look i really haven't been paying attention to any of this lately 18:04 Fixer Zeno`: enable local_map_saving - save waterfall and ask Shara to share her modfolder, load in singleplayer ??? success? or just don't bother and minimise the damn thing 18:04 Zeno` hmmmm, what did you say btw? 18:04 hmmmm i haven't said anything whatsoever... 18:05 hmmmm i literally have not the slightest clue what she's talking about 18:05 Zeno` hmmmm, she said she quit #minetest and #minetest-dev because of some attack you made 18:05 hmmmm maybe i was getting kind of ticked off with the "you owe it to me to work on minetest" attitude, but other than that, no clue 18:06 hmmmm christ m8 talk about entitled 18:06 Zeno` well, can you just say sorry and ask her back here? 18:06 hmmmm i'm not getting paid to work on mt 18:06 hmmmm i really don't have any idea what i'd be apologizing for though. i don't recall any specific situation. 18:06 Zeno` well neither am I, which is why I'm reluctant to spend 48 hours on a single bug 18:06 hmmmm that's generally the beef i had with vanessae though 18:06 hmmmm eh you don't have to 18:06 hmmmm none of us have to 18:06 Zeno` of course 18:07 Zeno` but let's get VE back at least 18:07 Zeno` she can provide info about the 0,0,0 thing maybe 18:07 Zeno` she has a lot of servers 18:07 hmmmm the minute you feel like you "have" to work on MT is the minute it becomes a jobb 18:07 Zeno` and it happens on all of them 18:08 Zeno` we *need* this feedback 18:08 Zeno` and if the server operators are not here we don't get it 18:08 Fixer !seen VanessaE 18:08 ShadowBot Fixer: I saw VanessaE in #minetest-dev 15 weeks, 3 days, 23 hours, 16 minutes, and 1 second ago saying "yeah, it's been around for a while. My Creative server also has this problem, but on that one, I can't be sure if the seed has/hasn't changed at some point. with the one I showed you, I'm 100% certain." 18:08 hmmmm wow 15 weeks 18:09 Zeno` she's online. Just not in any minetest related channels 18:10 Zeno` hmmmm, I have no idea what you said to offend her and I really don't want to know 18:10 hmmmm i found it 18:11 Zeno` but I'd like her back in the channel 18:11 Fixer http://irc.minetest.net/minetest-dev/2016-08-08 here is vanessae discussion with hmmmmmmmmmm 18:11 hmmmm http://irc.minetest.net/minetest-dev/2016-08-08#i_4672691 18:11 Fixer lol 18:12 Zeno` seems pretty trivial to me 18:12 Zeno` hmmmm, can you ask her back? 18:12 hmmmm sure 18:12 Zeno` thanks 18:12 hmmmm hrmm now that i'm looking at it 18:12 hmmmm i don't think vanessae quit because of me in particular 18:12 hmmmm but a general frustration over bugs not getting fixed in a timely manner 18:12 Zeno` probably not 18:13 hmmmm look this is not a commercial software product 18:13 hmmmm it's not even a popular open source game anymore 18:13 hmmmm the only people who work on MT are turbo-dorks 18:13 Zeno` but yeah. I think she might come back if you just say "hey I made a mistake and we want you back" 18:13 Fixer it was popular? 18:13 Fixer i play it since 2011 and getting impression that it has bigger popularity now 18:15 Fixer it has more blocks, more features, more fps, more mods, more players 18:15 hmmmm really? 18:15 hmmmm seems the opposite to me 18:15 hmmmm sorta went the opposite way when minecraft fell out of vogue 18:15 Fixer 4 years ago there were like 3-4 servers with players at most 18:16 Fixer tsundere.fi and c55.me was having like 10-20 players 18:16 Fixer were* 18:16 Fixer now it is 565 players 18:16 Fixer 177 servers 18:16 Fixer of different kind 18:17 Fixer but some gameplay bugs are persisting for years 18:17 Zeno` yay! 18:17 * Zeno` hugs VanessaE 18:17 VanessaE hi 18:18 Fixer that are not that hard like water reflow, or smooth lighting or occlusion or whatever 18:18 hmmmm lol 18:18 hmmmm those are hard bugs 18:18 Fixer wait 18:18 Fixer i didn't finish 18:18 hmmmm no offense but have you tried fixing them yourself? 18:18 Fixer people annoyed not but those 18:19 Fixer people annoyed by "taking your bones after death" or "avoiding fall damage when falling 1000 blocks down or some roof" 18:19 Fixer or going through solid block when sneak laddering 18:19 Fixer or not able to jump on edge of node 18:20 Fixer or stucking in stairs 18:20 Fixer or falling through stairs or slabs 18:20 Zeno` thank you 18:20 Zeno` I can sleep now. WB VanessaE 18:20 Fixer or fucking merging some carts in 18:21 Fixer finally 18:21 Fixer Hi VanessaE 18:22 Fixer i understand it is hard to fix water reflow, or make smooth lighting for everything, or better occlusion culling, but gameplay needs fixes too 18:23 hmmmm i thought paramat et al were working on that quite a bit 18:23 hmmmm i see like 5 mt game prs get merged for every one engine pr 18:23 Fixer for some reasons those kinds of bugs are left out for years 18:23 Fixer anyone even play this game besides me? 18:23 Fixer like for real 18:23 hmmmm that's why i said it's not more popular than it used to be 18:24 hmmmm the game was much simpler back then too 18:24 hmmmm people didn't have such high expectations 18:24 Fixer yes, but it had the same bugs tbh 18:24 hmmmm it was even buggier 18:24 Fixer i remeber water 4 years ago 18:24 Fixer that crappy flow 18:24 hmmmm like remember the mapgen v6 shadow bug that existed for years until i came along and fixed it 18:24 Fixer maybe 18:24 Fixer i seen so much 18:25 Fixer or crash at world edge 18:25 hmmmm so 18:25 Fixer there still shadows after vanessas moretrees bug 18:25 hmmmm for every new PR it seems like 2 new bugs are added 18:25 hmmmm i know we can break this down on a case by case basis but i'm just generalizing overall trends 18:26 hmmmm i've tried to push MT in the direction of higher quality, a bit more time spent validating the code being merged before merging it 18:26 hmmmm but i get pushback 18:26 hmmmm tons of pushback that MT is "just a game" and "you can't hold up progress forever" and so on 18:26 hmmmm not necessarily more code review, but more validation should be required 18:26 Fixer what new gameplay bugs you can name? 18:26 hmmmm uh 18:26 hmmmm how bout the lighting crash 18:27 Fixer thats just crash 18:27 hmmmm that is a bug 18:27 hmmmm ... 18:27 Fixer yes 18:27 hmmmm a crash is a bug 18:27 hmmmm you understand this right? 18:27 Fixer ofc 18:27 hmmmm and it's the worst kind of bug imaginable for us, heap corruption 18:27 hmmmm because the only real way to diagnose it is by running under valgrind 18:27 Fixer i mean gameplay one, without crashes 18:27 juhdanad Hi! I have tried to reproduce the lighting bug. 18:28 hmmmm memory access emulation is suicide for performance, and this is a performance intensive 3d application 18:28 * DI3HARD139 still uses mapfix for v6 and v7 maps 18:28 juhdanad It happens also when you just stand next to the waterfall. When a 'Waterx' node is placed by the server. 18:29 DI3HARD139 yep 18:29 Fixer btw, minetest works stable enough to me, crashes are veeeery rare to me, so actual inside of game gameplay bugs bother me (at least me) 18:29 hmmmm juhdanad: hi, your name sounds familiar. were you the original author of the new lighting code PR? 18:29 Fixer it crashes like few times a year at most 18:29 juhdanad Yes, I am. 18:29 hmmmm ahh 18:29 Fixer what i mean under gameplay bugs = bugs related to survival games 18:30 hmmmm Fixer: not our department 18:30 Fixer why not? 18:30 Fixer it is engine problems 18:30 hmmmm why not talk to PilzAdam, paramat, etc. they work on the minetest game 18:30 Fixer noo 18:30 Fixer bug is in engine 18:30 Fixer iirc 18:30 hmmmm what is the bug you talk about 18:30 Fixer for example this: https://github.com/minetest/minetest/pull/3810 18:31 hmmmm oh this shit 18:31 hmmmm ugh 18:31 hmmmm i just don't understand the actual issue there and not many people have tried to fully understand it 18:31 hmmmm if you'd like to give it a shot you're welcome to :) 18:32 Fixer btw i tested that PR and it kinda fixed the problem 18:32 Fixer or remember those shadows after vanessa trees bugs? 18:32 Fixer still not fixed 18:33 VanessaE s/vanessa trees/L-system trees on mgv7/ 18:33 Fixer hmmmm: or blowing up TNT in chains causing huuuuuge black areas 18:34 Fixer more simpler bugs: ability to take your bone in survival after death due to lag 18:34 Fixer i do it all the time 18:34 VanessaE the only sane way to fix lighting issues like that would probably be the 'hardware' idea RBA once had, but that's non-trivial to implement and has backward-compat issues. 18:34 Fixer falling bug: you can fall down like 50 nodes and take 0 damage because I hold shift and I'm kinda MT God now 18:35 Fixer at least 7 out of 10 tries 18:35 Fixer those are ridiculous bugs 18:35 Fixer nobody cares about falling it seems 18:36 Fixer or I can jump from 1000 nodes up to the ground, look up, and enjoy the ride, maybe hold shift just for good measure, and I will not even die after that (because damage resets) 18:37 Fixer or 18:37 Fixer OR 18:37 Fixer invulnerability (sorry for my english) when you are in a boat/cart etc 18:37 Fixer thats not survival 18:37 hmmmm so 18:38 hmmmm we have a bunch of devs that work on very different parts of MT 18:38 Fixer thats very crappy survival %) 18:38 hmmmm MT is actually really big 18:38 hmmmm and over complicated 18:38 Fixer yes 18:38 hmmmm you have to find somebody who is both willing and able to fix those very specific problems 18:38 Fixer but for some reason this crap never gets fixed, survival kind of bugs 18:39 Fixer that are not related to mtg 19:13 sfan5 :/ 19:13 sfan5 i've tried reproducing zeno's bug on redcat create at the spot he described 19:13 sfan5 nothing 19:14 Fixer try standing a bit more 19:14 Fixer did you have that stuck 0 0 0 at spawn btw? 19:14 juhdanad You should stand in the waterfall. 19:14 sfan5 with the debug build yes 19:14 sfan5 well no 19:15 sfan5 their spawn was at 0,0,0 iirc 19:17 Fixer if you spawn at 0 0 0, thats the bug 19:17 sfan5 no i'm trying to reproduce a crash 19:18 juhdanad I'm trying debugging with valgrind but my FPS is ~0.5. Do you have any suggestions? 19:18 sfan5 well i have 2 fps 19:19 juhdanad sfan5: if this helps, my client crashes when the waterfall changes. 19:20 juhdanad There's a segmentation fault just after the light update. 19:20 sfan5 mine doesn't 19:20 sfan5 os / compiler? 19:22 juhdanad Ubuntu 16.04 LTS / gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 20160609 19:23 sfan5 Fixer: well this is weird, i'm standing at 0,0,0 19:23 sfan5 and it shows the world at 0,0,0 19:23 sfan5 however areas says i'm at the waterfall 19:23 Fixer at RedCat? 19:23 Fixer try typing in chat 19:23 juhdanad Yes, happens to me too. 19:23 sfan5 i guess running with valgrind makes it buggy 19:23 Fixer open inventory? 19:24 Fixer wait, i'm going in 19:24 sfan5 it didnt even send me the welcome msg yet 19:24 Fixer sfan5: thats the 0 0 0 bug 19:24 sfan5 looks like it lost some packets 19:25 Fixer i'm connecting 19:25 juhdanad Then might the 0 0 0 bug be related to slow minetest clients? 19:25 Fixer bo 19:25 Fixer no* 19:25 Fixer people with good internet also have it 19:25 Fixer iirc 19:26 Fixer connected 19:26 Fixer i'm at 0 0 0 19:26 Fixer i see no sfan around 19:26 sfan5 well this bug has never happened to me 19:26 sfan5 (i just logged off also) 19:26 sfan5 except now 19:26 juhdanad Am I az 0 0 0? 19:26 juhdanad *at 19:26 Fixer turn F5 19:27 juhdanad I mean did you see me there? 19:27 Fixer no 19:27 Fixer i see nobody 19:27 Fixer no chat 19:27 Fixer no inventory 19:27 Fixer no commands 19:27 Fixer nothing 19:27 sfan5 ^ 19:27 Fixer i'm in limbo at 0 0 0 19:27 Fixer i can walk on map just fine 19:27 VanessaE just fyi, that is me on RedCat also, just to see what happens. 19:27 Fixer but nothing else 19:27 Fixer i don't see you VanessaE 19:27 juhdanad Then I should be next to the waterfall, despite I see myself at 0 0 0. 19:27 Fixer i'm helplessly walking near 0 0 0 19:28 VanessaE I'm at what appears to be the "real" spawn, but only part of the map is loaded. 19:28 VanessaE (yet I see people moving around) 19:28 Fixer i will grab my tea 19:28 VanessaE there it goes. had to walk close to the walls of the room. 19:28 Fixer i'm standing at 0 0 0, when bug ends - you will notice I will show up on waterfall automatically 19:28 Shara I've had requests to restart RedCat, though won't while you are testing. 19:28 sfan5 2016-11-25 20:28:47: VERBOSE[ConnectionReceive]: con(18/3470)WARNING: ACKed packet not in outgoing queue 19:29 sfan5 2016-11-25 20:28:52: VERBOSE[ConnectionSend]: con(18/3470)RE-SENDING timed-out RELIABLE to 51.254.216.93(t/o=0.236): from_peer_id=3470, channel=0, seqnum=65523 19:29 sfan5 using valgrind fuck with the connection code a lot 19:29 Shara 0,0,0 basically vanishes for a couple of hours after a restart. 19:29 Shara Testing if I get it myself now 19:30 Shara I have 0,0,0 for less than a second. Even that is very rare for me 19:31 sfan5 2016-11-25 20:30:58: INFO[Main]: Client: Detached inventory update: "creative_Bertino547" 19:31 sfan5 there are shared creative inv for every player 19:31 sfan5 sent to every player 19:31 Fixer lol, still at 0 0 0 19:32 Fixer done 19:32 sfan5 2016-11-25 20:32:03: INFO[Main]: ClientEnvironment::processActiveObjectMessage(): got message for id=40798, which doesn't exist. 19:32 Fixer i'm at waterfall 19:32 Shara Server up for approx 22 hours, plus busy, so it will be at it's worst now 19:32 Shara Yet I still get a much longer delay on Xanadu. 19:33 VanessaE other than the map being slow to load, I don't seem to be having trouble with the server. 19:33 Fixer so about 6 minutes wait time today 19:33 Fixer VanessaE: i have the same problem on your Creative survival, or how it is called 19:33 Fixer one without dreambuilder 19:33 VanessaE that's "Building". 19:34 Fixer VanessaE: with 0 0 0 bug spawn on a desert island, right? 19:34 VanessaE ok, grey screen at login that time. took several seconds to put me at the spawn. 19:35 VanessaE no, just all-grey. didn't catch what coords I was at though. 19:35 Fixer VanessaE: it is not this, you login to spawn, walk at spawn or even get out, but can do nothing, wait X-XX minutes before you are teleported to your place, get welcome formspec, inventory and stuff 19:36 VanessaE Fixer: it's being slow, but not like you're experiencing 19:36 VanessaE the rest of the map around me just now loaded, though 19:36 VanessaE (without me so much as panning/looking around) 19:36 Shara It seems to effect specific people badly... and others basically not at all. 19:36 Fixer grey screen is not that 19:36 VanessaE nono 19:37 VanessaE not the grey screen. 19:37 VanessaE that came first, this was after the world and that welcome formspec appeared. 19:37 Shara Grey screen is just because server is struggling right now. Seems it's been hit hard today. 19:37 Fixer i'm at crashy waterfall right now 19:37 VanessaE let me try again. 19:38 VanessaE ok, I spawned at 0,0,0 for a few seconds, with grey screen, before being teleported to the library. 19:39 Shara What I'd love to know is why only certain people get hit really badly. (Like Fixer) 19:39 Fixer VanessaE: you have just few seconds, i have 6 minutes, or even 15 19:39 juhdanad Strange: after I quit from the valgrind version and rejoined with base minetest, my position was reset to 0 0 0. 19:40 Fixer just wait a bit on 0 0 0 19:40 VanessaE so where's this glitchy waterfall at? 19:40 juhdanad So my position is lost if I quit while I'm stuck. 19:40 Fixer 1483 x 2903 19:40 Fixer or tp to me 19:40 Shara I would also ask people compare to Xanadu if they can for 0,0,0 there, since for me Xanadu is far worse 19:40 VanessaE I'm on a no-interact account at the moment, teleport me there please. 19:41 Fixer I have no problems on Xanadu 19:41 Fixer never had 0 0 0 at it 19:41 Shara 10 seconds + every single time on Xanadu for me, unless it's 'right' after a restart 19:41 Fixer juhdanad: i see you now 19:42 Fixer Shara: please teleport VanessaE to that location 19:42 juhdanad Yes, and my client just crashed. 19:43 Fixer lol 19:43 Fixer linux? 19:43 Shara Sorry, drowning in messages here. 19:44 Shara Please give me exact coords 19:44 juhdanad But when I comment my light updating function out, it doesn't crash. 19:44 Fixer interesting 19:44 Fixer Shara: 1483 3.5 2903 19:45 garywhite What's going on? 19:45 Shara Done 19:45 garywhite Fixer: Are you talking to coords on RedCat? 19:45 Fixer yes 19:45 VanessaE EEP! 19:45 VanessaE crashola. 19:45 Shara garywhite: they are trying to figure out 0,0,0 19:45 Fixer lol 19:45 Shara Okay, now I need to see what's at these coordinates 19:46 VanessaE within a few seconds of arriving at the waterfall, the map loaded up just fine and then, boom. 19:46 Fixer yet I who has all sorts of bugs loving me so much are not affected at all 19:46 sfan5 proposed issue solution: switch to arch linux 19:46 sfan5 doesn't happen for me 19:46 Shara Didn't happen to me 19:47 Shara But that is one crazy waterfall 19:47 VanessaE lemme pull this up in gdb 19:47 Fixer Shara: just don't touch that waterfall 19:47 Fixer ok 19:48 Shara As per RC chat: 19:48 Shara (7:47:32 PM) Shara: I'm not touching it 19:48 Shara (7:47:40 PM) Shara: It's already protected by the player who made it 19:48 VanessaE ok, bounce me over there again, please. 19:48 Shara Did 19:48 VanessaE crashed. 19:48 VanessaE bt coming. 19:49 Shara AFK a little while. 19:49 Fixer hah 19:49 Shara Fixer, I'll give you both teleport and bring so you can send any other testers there :) 19:49 Fixer ok 19:49 VanessaE http://pastebin.ubuntu.com/23533890/ 19:51 VanessaE (that's at commit 0d1c9598 from Nov 22, btw) 19:52 Fixer posted this in that issue topic 19:52 VanessaE saw 19:52 VanessaE however, zeno already posted a bt, so maybe not so useful 19:53 garywhite What server is everyone referring to? 19:53 Fixer Red Cat Creative 19:53 VanessaE RedCat Creative 19:53 garywhite oh ok, I'll be coming on in a minute 19:53 VanessaE something unusual about a waterfall is causing client crashes 19:53 Fixer VanessaE: whats the OS? 19:54 VanessaE Fixer: Debian 8.6 19:54 sfan5 https://gist.github.com/anonymous/8816a606504a9c1cfe6304125bd891af 19:55 sfan5 the 0,0,0 bug is definitely not the clients fault 19:55 sfan5 the server just doesn't send the rest of the important packets for a full join 19:55 sfan5 or at least they don't reach the processing code 19:55 Fixer it sends map block data for sure 19:55 Fixer for me 19:55 sfan5 (pkt << are received packets, pkt >> are sent packets) 19:56 Fixer i can walk everywhere just fine, blocks load, but nothing else 19:56 garywhite This is the weirdest waterfall I've seen 19:56 VanessaE what's creating those water updates? 19:56 VanessaE some oscillating pistons or something? 19:57 garywhite Vanessa are you seeing how weird this looks? 19:57 VanessaE garywhite: I was for a moment - "holes" in the water, descending 19:57 Fixer strange plants? 19:57 VanessaE like the flow is being periodically interrupted 19:57 sfan5 hm a toclient_time_of_day should be reaching the client every 5 second 19:57 sfan5 s 19:58 garywhite I've never seen anything like it 19:58 VanessaE I saw red plants here and there around the server, thought it was just some mopd 19:58 sfan5 i guess i'll have to look at it with wireshark 19:58 VanessaE mod* 19:58 garywhite Those are crystals from caverealms 19:58 VanessaE wat 19:58 garywhite Responding to Fixer 19:58 Fixer ah 19:58 VanessaE oh 19:58 Fixer i see their are triggering the water somehow 19:58 sfan5 also concerning the 0,0,0 bug: looks like Server::StageTwoClientInit() isn't getting reached 19:58 garywhite The red plants may be related to the space mod, and Fixer it's actually seacoral 19:59 VanessaE oh yes I see the blue crystals 19:59 juhdanad The server sends node placements to the client (Waterx nodes), in about every seconds. 19:59 juhdanad But only if you are near. I don't know what it is. 20:00 Fixer here this 0 0 0 bug report if needed: https://github.com/minetest/minetest/issues/4403 20:05 VanessaE strange, I've been floating here near the fall for some time now (5+ mins) and not getting the crash 20:05 Fixer VanessaE: go down to the base of it 20:05 sfan5 why the fuck does the netprotocol send pings multiple times per second 20:05 VanessaE ok, I'm there. 20:06 VanessaE sfan5: hasn't it always? 20:06 sfan5 dunno 20:06 sfan5 seems a little excessive to me 20:07 Fixer yet when server is out you can walk some time with RTT=-1 in your statusbar 20:07 garywhite Shara: Is it OK if I record a quick video of the waterfall to send to #minetest to see if anyone there can figure it out? I'll be sure to turn off my chat 20:07 sfan5 okay it's definitely the servers fault 20:07 sfan5 no TIME_OF_DAY packet in wireshark 20:07 Fixer did I tell anybody RTT is broken? it seems stuck at some value after login, not unlike in 0.4.4 where it measured it once a second or so 20:09 Shara garywhite: I don't see how that's helpful. Maybe leave leave the devs to finish their checks before adding more people to the mix. More worried about 0,0,0 bug myself than a weird waterfall 20:09 Shara -leave* 20:09 garywhite ok 20:15 sfan5 somethings broken 20:15 sfan5 broken quite a lot 20:15 sfan5 2016-11-25 20:47:05: ACTION[Main]: pkt >> TOSERVER_CLIENT_READY 20:15 sfan5 2016-11-25 20:47:12: ACTION[Main]: pkt << TOCLIENT_DETACHED_INVENTORY 20:15 sfan5 https://github.com/minetest/minetest/blob/0d1c9598a0d2a4f21dc57de32efca2dc52b6b146/src/server.cpp#L1097 20:15 sfan5 according to the code CLIENT_READY should cause a few other packets to be sent 20:15 sfan5 before detached inventories are sent 20:15 sfan5 but those packets appear nowhere 20:16 VanessaE sfan5: you DO realize you guys are gonna have to get sapier in on this.. :) 20:16 sfan5 someone should just rewrite the network protocol stuff tbh 20:17 Fixer VanessaE: you crashed? 20:17 VanessaE huh. 20:17 VanessaE yeah, I guess I did 20:17 Shara Is it worth me doing a restart to see if that does or does not fix the weird waterfall? 20:17 VanessaE I buried the client under another window to do something else, guess it crashed while I wasn't looking 20:17 sfan5 even without the bug (or maybe it's just not noticeable?) the server has sent hundreds of detached invs before TOCLIENT_PRIVILEGES happens 20:17 sfan5 Shara: no 20:18 juhdanad It won't, I think. It is client-side. 20:18 sfan5 how the fuck does the protocol order packets 20:18 sfan5 is it just rand() or what 20:18 Wuzzy Hi all 20:18 Shara Okay. Will leave it run then. If possible please let me know when done testing though. It needs a restart quite badly. 20:19 Wuzzy Does this error look like an engine error to you?: 20:19 Wuzzy https://forum.minetest.net/viewtopic.php?f=11&t=9691&view=unread#p240623 20:19 sfan5 wait what 20:19 sfan5 why does the server even send detached invs 20:19 VanessaE sfan5: well there was a big deal a while back to support really OLD clients 20:19 VanessaE maybe that explains the confused order 20:19 Wuzzy ouch, it seems I must have hit a nerve here :D 20:19 sfan5 oh nvm found it 20:20 garywhite Shara why don't you go ahead while none of the testers are on 20:20 Fixer Shara: no, do not restart 20:20 Shara garywhite: sfan5 just specifically asked me not to restart 20:20 sfan5 why does it send detached invs before you join the game even 20:20 sfan5 ???? 20:20 Fixer Shara: it persists after restarts, it is useful for testing 20:20 sfan5 Fixer: no 20:20 garywhite Hmm...Shara when did you notice the waterfall? 20:21 VanessaE sfan5: as I recall, with really old versions, there was no "you're officially in the game now" packet (whatever that amounts to), 20:21 VanessaE so it just blindly sends "everything" 20:21 Fixer juhdanad also testing it 20:21 Shara I did not. Zeno did. 20:21 VanessaE (but I thought that was limited to a certain range of client+server combos) 20:21 garywhite When did Zeno point it out? 20:21 Shara I have never seen liquid behave that way on RC before. 20:22 sfan5 2016-11-25 20:42:49: ACTION[Main]: pkt >> TOSERVER_CLIENT_READY 20:22 sfan5 2016-11-25 20:42:58: ACTION[Main]: pkt << TOCLIENT_PRIVILEGES 20:22 sfan5 10 sec difference 20:22 sfan5 despite the cmd handler for CLIENT_READY *directly* sending the privs: https://github.com/minetest/minetest/blob/master/src/network/serverpackethandler.cpp#L716 >> https://github.com/minetest/minetest/blob/master/src/server.cpp#L1100 20:22 sfan5 first come first served might not be the best thing for packet ordering 20:23 sfan5 hm nah 20:23 sfan5 the other problems are that there are hundreds of creative inventories sent to users 20:23 VanessaE hundreds? 20:23 sfan5 it send me a detached inv "creative_" for seemingly every player 20:23 sfan5 if these get created only when the players are online 20:24 sfan5 that would explain why it doesnt happen after a fresh restart 20:24 juhdanad Fixer, could you teleport my instances (juhdanad2) to the waterfall please? 20:24 garywhite I can do it, just lemme log in from IRC 20:24 garywhite oh, nvm then 20:25 sfan5 yup that is definitely the problem 20:25 Fixer you have 10 seconds wait time? you are lucky 20:26 sfan5 VanessaE: https://gist.github.com/sfan5/85c4c78a635684475d5cf9f440b69356 20:27 VanessaE holy shit 20:27 sfan5 let me quote torvalds on this: whoever wrote that mod should be shot 20:27 VanessaE lol 20:27 Shara It's a mod that's to blame? 20:27 sfan5 probably the whole TOCLIENT_DETACHED_INVENTORY pkts are clogging up the pipe 20:27 Fixer LOL 20:27 VanessaE unified inventory is in use on that server... 20:27 VanessaE surely it's not doing that? 20:27 Shara VanessaE: But Xanadu gets this and does now use unified 20:27 sfan5 and causing insane waiting time (minutes) for logging in with slow clients 20:28 sfan5 lemme try xanadu 20:28 Shara Tenplus1 reported my other server also trapped him at 0,0,0, and that doesn't use unified 20:28 Fixer i will try Xanadu right now with second client 20:28 Shara Will also try Xanadu 20:28 sfan5 2016-11-25 21:28:14: ERROR[Main]: Sorry, no new players being admitted at this time! 20:28 sfan5 or not... 20:28 sfan5 Shara: which "other server"? 20:28 Shara No, didn't get it on Xanadu this time, which is odd for me 20:29 Fixer sfan5: i can give PM you my pass to this server 20:29 Shara sfan5: One that isn't one public list and has maybe ten players total... 20:29 sfan5 Fixer: if you trust me with that ,sure 20:29 garywhite1 Switched to my Linux VM 20:29 Shara on* 20:29 garywhite1 Shara are you referring to DL? 20:29 Shara Yes 20:30 garywhite1 oh ok 20:30 Fixer sfan5: i trust you, i've just logined without problem, i will pm you now 20:30 sfan5 Fixer: also can you try to observe the 0,0,0 bug on "floatlands test server" 20:31 Fixer will try now 20:31 Fixer sfan5: i will change it once you don't need it for testing, login as you like 20:33 Fixer sfan5: floatlands test server - very fast login, works nice, spawned at -125 1289 115 20:33 sfan5 yup 20:33 sfan5 it should have exactly 0 detached invs 20:33 sfan5 (or very few not sure what mods use some) 20:33 sfan5 Fixer / VanessaE: similar situation with xanadu though not as extreme https://gist.github.com/sfan5/b99c16f15c9645fa0ada73a975349990 20:34 Fixer sfan5: try vanessas server, i will tell which 20:34 Shara Xanadu has currently been up only half as long as RC 20:34 Shara Though that would usually be enough to give me trouble. 20:34 Fixer sfan5: try this one daconcepts.com:30008 20:34 sfan5 sec 20:34 sfan5 summarised problems: 20:34 sfan5 * modders like to break stuff 20:35 sfan5 * network proto needs batched acks 20:35 sfan5 * mt needs batched detached inv updates 20:36 sfan5 2016-11-25 21:36:11: ERROR[CurlFetch]: An unhandled exception occurred: Das Argument ist ungültig 20:36 sfan5 huh 20:38 sfan5 Fixer: similar thing https://gist.github.com/sfan5/944746fabea3151029b7cf298b0f0dc7 20:38 Fixer just connected to it, 0 0 0 bug 20:38 sfan5 you must be having a really slow 1) computer 2) network or 3) broken netproto code 20:39 Fixer depends on server 20:39 sfan5 those three options also apply to the server 20:40 Fixer i had this problem even without much traffic on my side 20:40 AcidNinjaFWHR it looks like a bottleneck 20:40 Fixer i have 2 mbit connection 20:40 Fixer reliable one 20:41 AcidNinjaFWHR can be a problem on the server, it's not necessarily related to your network speed. 20:42 VanessaE 65 Mbps connection here, and my servers are on a 250M+ pipe 20:42 Shara If it's purely the server, why do only certain people get hit so hard by it? 20:42 VanessaE so I think "slow/network" can be ruled out? 20:42 Fixer if i download/upload something - it may affect 0 0 0 bug for sure 20:43 VanessaE Fixer: btw, I could see your chat attempts earlier. 20:43 VanessaE 5 mins ago when you tried to sign onto Building and got stuck, I mean. your chat was coming through to IRC just fine 20:43 Fixer anyway, i have free bandwidth most of the time, it is not chocked that much 20:43 sfan5 Shara: it can be the server OR the client 20:46 Fixer sfan5: when I stand at 0 0 0 waiting for bug to go away, i can type in chat in people will actually read it, but i see nothing 20:46 sfan5 sounds correct 20:46 sfan5 server -> client direction is clogged 20:46 sfan5 your actions will go through 20:46 sfan5 but you won't see the result 20:50 sfan5 only mods used on all three buggy servers: castle, homedecor, inbox, itemframes, signs_lib, worldedit, worldedit_commands, worldedit_infinity 20:51 sfan5 (assuming nobody has renamed any mods) 20:54 Shara I suspect different versions of castle are used in each case. RC is actually running mycastle, which by now might be quite different. 20:56 sfan5 https://github.com/minetest/minetest_game/blob/master/mods/creative/init.lua#L17 20:56 sfan5 ohhhhh boy 20:58 Fixer i always remember that ragecomic with "comment my code, nuh, 1 week later" 21:00 sfan5 https://github.com/minetest/minetest_game/commit/b1e2ece63842ff5228798feb9259398c75524f8b 21:12 sfan5 however there's also other mods doing this 21:12 sfan5 e.g. "armor" 21:13 sfan5 there's also invsaw_, bags_, refill 21:13 AcidNinjaFWHR only inbox & worldedit are in common with my server, and I got this 0,0,0 bug too sometimes. 21:13 sfan5 creative mode enabled? 21:13 AcidNinjaFWHR I honestly don't think the problem comes from a mod. 21:14 AcidNinjaFWHR no, survival. 21:14 sfan5 huh 21:14 sfan5 ip? 21:14 AcidNinjaFWHR paly2.ddns.net 30000 21:14 AcidNinjaFWHR but i haven't seen the bug since weeks 21:14 sfan5 when did you last restart? 21:15 AcidNinjaFWHR oh you may wanna join 21:15 sfan5 i just did 21:15 AcidNinjaFWHR there's the bug 21:15 AcidNinjaFWHR ok 21:15 AcidNinjaFWHR we restart every morning 3.30 am (we do backup the world) 21:15 sfan5 then the bug will not happen often 21:16 sfan5 or rather be more likely to appear just before you restart 21:16 sfan5 2016-11-25 22:15:13: INFO[Main]: Client: Detached inventory update: "789_armor" 21:16 sfan5 2016-11-25 22:15:13: INFO[Main]: Client: Detached inventory update: "789_bags" 21:16 sfan5 2016-11-25 22:15:13: INFO[Main]: Client: Detached inventory update: "789refill" 21:16 sfan5 mods to blame: armor, bags and ??? 21:16 sfan5 could you do a grep for create_detached_inventory in your mod folder? 21:16 Hijiri if player inventories could have callbacks then people probably wouldn't use the "modname_playername" detached inventory "idiom" 21:16 AcidNinjaFWHR k 21:17 sfan5 Hijiri: nah the problem is you need to serve an unique inv to each player 21:17 sfan5 e.g. with the search feature in creative 21:17 Hijiri yeah, and you could use player inventories for that if they supported callbacks 21:17 sfan5 oh you can add more inv lists, right.. 21:18 sfan5 i guess for now an acceptable workaround would be to declare a player for each detached inv 21:18 sfan5 so it only gets sent to them 21:18 Hijiri or maybe (for the specific case of creative search) a way to specify an inventory list formspec item thatis non-contiguous 21:19 sfan5 this could also explain RAM usage rising the longer you run 21:19 Hijiri supply a list of extents or something 21:19 AcidNinjaFWHR ods/3d_armor/3d_armor/armor.lua:372: local armor_inv = minetest.create_detached_inventory(name.."_armor", { 21:19 AcidNinjaFWHR only got this line sfan5 21:19 sfan5 huh? 21:20 sfan5 where do "bags" and "refill" come from? 21:20 sfan5 grep for "refill" maybe 21:20 AcidNinjaFWHR nothing 21:21 AcidNinjaFWHR oh my bad was in the wrong folder 21:21 sfan5 ... 21:21 AcidNinjaFWHR so yeah, there is also /mods/unified_inventory/bags.lua:138 21:22 sfan5 what about "refill" tho 21:23 AcidNinjaFWHR sfan5, the search for create_detached_inventory: https://paste.sh/PJ5hIv8F#v8ik8mW8pHb5CokQkJRmC_rE 21:23 sfan5 unified_inventory too 21:23 Fixer sfan5: bags refill maybe unified inventory and minetest_bags? 21:24 sfan5 see paste above 21:24 AcidNinjaFWHR refill also comes from unified_inventory 21:25 Fixer i have 800 mb of mods, i can do search on them 21:30 nrzkt sfan5, there is a real problem with detached inventory, they are broadcasted to everybody, which is just... bad 21:30 sfan5 why doesnt the documentation warn about this 21:33 nrzkt because mt does this with this feature, detached inventory should only be sent to related client, but this is not the case 21:33 sfan5 what was even the intended usage for detached invs? 21:34 nrzkt add additionnal features to players themselves 21:34 nrzkt generally 21:34 nrzkt example: bags 21:34 sfan5 those shouldn't be detached tho 21:34 sfan5 they should be attached to the player 21:34 sfan5 just an additonal list 21:34 Fixer sfan5: my list of mods with create_detached_inventory - http://pastebin.com/raw/Rf7iSigG 21:34 nrzkt this is the cased, detached inventory are attached to players generally, but broadcasted to everybody 21:35 sfan5 Fixer: it would be useful to see how they use them, e.g. minetest.create_detached_inventory("trash") is not a problem 21:35 sfan5 also it seems literally every subgame has this problem with creative 21:36 Fixer hmm 21:36 Fixer i can grep it in linux 21:36 sfan5 please do 21:37 Fixer will do now 21:40 AcidNinjaFWHR grep -rn 'yourtexthere' folder or . if you're inside it already 21:41 Fixer ok, it works 21:44 Fixer sfan5: like this? http://pastebin.com/raw/GXK2Any4 21:44 sfan5 yes 21:46 Fixer searched 275 mod and subgame folders 21:46 Fixer not everything but thats large portion of it 21:46 Fixer thats my mod collection 21:47 Fixer updated automatically by script 21:48 Fixer is not invsaw related to rubenwardy? 21:53 sfan5 no idea 21:54 sfan5 sounds more like moreblocks 21:56 sofar man, I've been watching carts go uphill all morning 22:32 Fixer lol 22:32 Fixer xanadu admin really pissed 22:35 Fixer every time i mention some bugs on that server i got kicked (they were talking about 0 0 0 one) 22:39 AcidNinjaFWHR :/ 22:39 AcidNinjaFWHR That's not cool. 22:42 paramat well you do sometimes rant and exaggerate about bugs, you did today on this channel 22:48 Fixer nah, it started earlier, what i did is just "it is under investigation", that guy is pissed about devs 22:49 sfan5 angry at devs? 22:49 Fixer yeah, why not? 22:50 Fixer %) 22:50 Fixer sfan5: remember those server stalling problem? thats when his nerves started to... became unstable 22:50 Fixer problems* 22:51 Fixer with that constant crashes 22:57 paramat ah