Time Nick Message 00:04 paramat nore or anyone, #5494 is a blocker and is now rebased and tested, please can anyone review? 00:04 ShadowBot https://github.com/minetest/minetest/issues/5494 -- Fix 'alpha' property for liquid nodes by juhdanad 00:08 nore paramat: see comment, I'm fine with the code and I see no obvious bug, but can't test 00:29 paramat thanks 00:29 paramat might merge later 00:30 paramat i made a silly mistake in #5766 will fix 00:30 ShadowBot https://github.com/minetest/minetest/issues/5766 -- Particle spawner: Do not spawn particles distant from player by paramat 05:50 nerzhul paramat you should really stop. Blocker are not for PR which doesn't fix bugs. stop that shit please. 05:51 Zeno` I meant to ask earlier what 582ee15d has to do with anything 05:52 nerzhul yes, lol 05:52 Zeno` That frees memory when the object is destroyed; i.e. fixes a leak 05:52 Zeno` umm I'll run massif 05:53 Zeno` where was 582ee15 referenced from again? 05:53 Zeno` (I want to connect to the same server as in the report) 05:53 nerzhul https://github.com/minetest/minetest/issues/5740 05:54 Zeno` cheers 05:56 Zeno` pity fixer left. But I'll test on that server anyway 05:58 Zeno` s/test/collect data 05:59 nerzhul Zeno`, can you stop paramat shit on this PR ? https://github.com/minetest/minetest/pull/5640#issuecomment-302308622 06:00 paramat nerzhul that PR is essential, read the comment 06:00 paramat dont try to provoke someone else 06:01 paramat argue withme yourself 06:01 nerzhul which bug does it fix ? i don't see any issue here 06:01 paramat "What about this? minetest/minetest_game#1707 won't work without this pull merged." The MTG feature is broken unless this PR is merged, therefore this is a blocking issue. 06:02 nerzhul minetest_game PR should not affect core engine it's two unrelated entities 06:02 nerzhul it's not a engine bug, it's a feature. 06:02 paramat another random rule you've decided 06:03 paramat engine and game are closely interdependent 06:03 nerzhul i don't decide anything, everybody said core is core, and MTG is just a game, a MT game feature should not block core release, as it's a feature not a bug. 06:03 nerzhul no 06:03 nerzhul if game is too linked to core there is a real problem 06:03 paramat MTG is broken unless this PR is merged 06:03 paramat you're making zero sense 06:03 nerzhul it's broken because of what ? you added a inexistant core feature ? 06:04 nerzhul it's not the first time i see a MT game pull for a feature which require core enhancement, but it's enhancement not a bugfix. 06:04 paramat the engine code that the MTG feature depends on has not yet been merged, the author, not me, has commented to state that this Pr is essential due to the game PR having been merged 06:04 nerzhul is this our problem if you develop feature for non added, non approved features in core ? 06:05 paramat 'you' are not separate from MTG 06:05 nerzhul sorry but core should have the feature and THEN you can merge mtg pr 06:05 paramat but yes, the MTG PR should have waited for the engine code 06:05 nerzhul it's like you buy wheel for your car but you don't have a car, be realist man 06:06 paramat that changes nothing 06:06 paramat if release goes out without this engine PR it will be broken, so it's a blocker 06:06 nerzhul this will not make the PR merged to core, you just try to enforce a PR merge before release because you broke MTG 06:06 nerzhul not my problem if MTG is broken because engine doesn't have feature you added to MTG, it's your problem 06:07 paramat yes a mistake was made, the order was wrong 06:07 paramat that makes no difference 06:07 nerzhul it's a great difference 06:07 nerzhul you made a mistake and now you want to add a blocker on a feature PR. 06:07 paramat you cannot act as if MTG is irrelevant 06:07 nerzhul it's not like you don't added feature because you waited for cor engine. 06:08 nerzhul MTG is relevant if it follows engine, not if it adds features which are not supported by core. 06:08 paramat yes a mistake was made, makes no difference 06:08 nerzhul it's like if VanessaE added a new great feature on dreambuilder and open a PR on core to add the feature and said "fix that shit it's broken in my world" 06:09 paramat that's not the default game supplied with MT engine 06:09 paramat but, you have the right to suggest the MTG PR is now removed while we discuss the addition of both PRs 06:09 nerzhul there is no default game with engine in all distros, it's only for windows and android, and i can set a working tag when releasing 06:10 nerzhul (i.e. before your merge on mtg) 06:10 paramat so, the choice is, remove the MTG PR or ad the engine PR, otherwise release is broken 06:11 paramat *add 06:11 nerzhul remove it from MTG, wait for merge and re-add it 06:11 nerzhul it's simple we are using git not CSV 06:11 nerzhul CVS* 06:11 paramat coloured itemstacks is much wanted 06:11 nerzhul yes, and i want also many pr in core, it's not a reason to add blocker on those features 06:11 paramat that would be a lot of unnecessary work, no point 06:12 nerzhul s/unnecessary/delayed/ 06:12 Zeno` a blocker label on a PR? 06:12 nerzhul yes, on a feature PR not a bug fix pr 06:12 paramat i don't add blocker for what i want to see merged, i add what is essential for release, do not misrepresent what i am doing 06:12 Zeno` seems kind of unusual 06:12 nerzhul it's essential because you break mtg, not because core is broken lol 06:12 paramat we have done it before 06:12 nerzhul who ? you ? because nobody verified ? 06:13 paramat many people have 06:13 nerzhul if somes wants to look at this PR and approve it before feature freeze, they are free and we can merge, if nobody looks at it, sorry delayed for next release 06:14 paramat the game PR would need to be removed, and the coloured iitemstack feature needs this to complete it's implementation 06:14 nerzhul ofc 06:16 Zeno` isn't being in the TODO list enough? 06:16 Zeno` I don't understand 06:16 Zeno` blockers are for bug reports 06:16 paramat anyway i feel using 'blocker' on PRs is useful because: "Using 'blocker' in PRs as we have done before makes dev easier because we can then filter PRs to see what has to be merged, instead of filtering issues and then following a link to the PR. 'blocker' also shows up in thread lists" 06:17 nerzhul easier before having milestones, i agree, but now we have milestone we can put every PR needed for release into it (like every dev workflow using milestones in fact) 06:17 paramat the project TODO does not show up as a label in thread lists 06:18 nerzhul in PR list maybe, but for release you just need to go to milestone and look at remaining issues 06:18 nerzhul https://github.com/minetest/minetest/milestone/8 06:18 paramat no, read my comment, milestone is not blocker, different levels of importance 06:18 paramat "A milestone is not the same as a blocker, a milestone is what should be considered for release but is not absolutely essential, a blocker is what must be included." 06:19 nerzhul milestone is not blocker, but feature PR is not blocker, and milestone show all the remaining issues/PR to finish to close the milestone and release. You really should read some books about dev workflows 06:19 paramat and these have been used this way for a long time 06:19 nerzhul before using milestones, yes. now we are using them 06:19 paramat i don't care about your books, what matters is how we work in Mt dev 06:20 nerzhul in MT we are using milestones since this release cycle. 06:20 paramat no, milestone and blocker have been used together as i have described for a while now 06:20 nerzhul stop that shit on feature PR, use blocker on blocking & relevant issues. 06:21 nerzhul you make shit on mtg by merging a pr before a core pr merge, it's your fault that's the point. 06:21 paramat yes 06:21 nerzhul it's not the reason to debate and add blocker on a pr because you do a mistake on mtg 06:21 paramat doesn't change the fact the engine PR is needed for release 06:21 nerzhul no 06:22 paramat so without the engine code release will not be broken? 06:22 nerzhul the MTG PR will be in the mtg tree IF this PR is merged 06:22 paramat ok 06:22 nerzhul if somes wants to look at this PR (i looked at the quality it's good, but the feature is not in my knowledge area) and approve it, they are free 06:23 paramat so what's your opinion on the 2 PRs are they needed for release? 06:23 nerzhul which PR? 06:23 paramat let's ask the author 06:23 paramat MTG 1707 and MT 5640 06:23 nerzhul this is a new feature as it seems, it can be delayed, they are interesting and we have soooo many new features on this release, waiting for next release can be done :) 06:24 paramat ok i'll ask the author 06:24 nerzhul it's not mt 0.4.15 which was a little bit tiny compared to this :) 06:25 paramat anyway, what are people's opinion about being able to use 'blocker' on PRs? for example if there is a PR which *must* be merged but has no corresponding issue? 06:25 paramat seems useful to me and no argument against 06:27 nerzhul if it's an important bugfix okay, if it's a feature no 06:28 paramat hmm, well, what if it's a feature that must be included to avoid a broken release, as this PR? 06:29 paramat i guess we'll all have to decide the usage 06:30 Zeno` ok, I think there is indeed a leak 06:30 paramat so, just to explain, i will not re-add the lable, just to avoid iritating you, but i will add an engine issue about NTG1707 and MT5640 and what to do about them, the issue will be a blocker 06:30 nerzhul mtg should wait for core to have the feature 06:30 paramat yes 06:31 paramat *label 06:31 paramat maybe an issue in MTG too since the mistake was made there 06:34 Zeno` hmm 06:34 Zeno` I'll add to the bug report I think 06:34 Zeno` http://imgur.com/a/NcrVq 06:35 nerzhul the block cache is quite consuming but did you leave the game ? 06:35 Zeno` leave? 06:36 nerzhul yeah, return back to main menu 06:36 Zeno` nope, but that's not the issue 06:36 Zeno` the issue is increasing memory usage 06:36 Zeno` which is obviously happening 06:36 nerzhul fixer said memory was not reclaimed when return back to main menu no ? 06:36 nerzhul i think yes, and it's due to the cache, maybe we should add a size limit 06:37 Zeno` yeah 06:39 paramat i guess this is not the mapblock mesh data? that tends to increase slowly up to about 1GB then levels off (according to the 'client mapblock limit' setting) 06:40 paramat although complex meshes would be more data 06:41 Zeno` I wonder what I did just after 8e+11 to make it drop memory. Hmm 06:41 Zeno` it's certainly the cached meshes 06:41 paramat but that's expected and normal surely? 06:42 Zeno` well yeah. But why does in then go back up again and then continue to increase again 06:42 paramat ah that 06:44 paramat erm, could be a mod making heavy use of noisemaps and lua voxelmanips but without the memory use optimisations i have posted in the forum? maybe that is adding a fluctuation on top? 06:44 paramat 470MB is not much for MT memory usage 06:46 Zeno` nah, but I didn't stay connected for long and didn't really move heh 06:46 Zeno` I'll do it again doing nothing at all (I'll just stand at spawn) 06:47 Zeno` it's hard because people keep killing me lol 06:47 paramat hehe. useful as will rule out buildup of mesh data 06:48 Zeno` after dinner I'll do it using a local server as well 06:50 Zeno` I haven't looked at the code yet (still getting data) but *is* the cache size limited? 06:50 Zeno` if it's not, then that's the leak 06:50 Zeno` brb 06:54 nerzhul i don't think it's limited 06:55 nerzhul also i think it should be very very limited on android or disabled 06:55 nerzhul else some Samsung Galaxy will explode :p 06:56 Zeno` I don't think it's limited either 06:57 Zeno` (which is obviously a problem... it should be a circular queue or something replacing older cache entries after the limit is reached) 06:57 nerzhul seems there is some size 06:57 nerzhul cleanupCache refers to some limits 06:58 Zeno` hmm, yeah. Called from addBlock 07:01 nerzhul refcount on cache is linked with queue pop, but i don't limits 07:01 nerzhul i should go, good luck 07:01 Zeno` why is cache_seconds divided by cache_seconds_max as well? 07:01 Zeno` hmm ok 07:01 Zeno` have fun :) 07:03 Zeno` paramat, can you check line 278 of mesh_generator_thread.cpp? 07:04 paramat ok 07:06 paramat by cache do you mean general cache or mapblock mesh data cache? 07:07 Zeno` https://cloud.githubusercontent.com/assets/5899883/26190362/60d06080-3bec-11e7-9730-4ea33817c544.png 07:07 Zeno` mapblock mesh 07:07 paramat hmmm 07:11 Zeno` ugh. I have to go, but I'll bbl. Hopefully the data can somehow help 07:11 Zeno` if requested I'll post the raw data 07:11 paramat ok 07:12 Zeno` doing now for the first graph just in case I'm gone longer than I expect 07:13 Zeno` gimme a sec, I need to make sure there is no private stuff in it :) 07:13 paramat so, while we decide i have removed 'blocker' from all PRs and if necessary opened a new corresponding issue to be the 'blocker', so 2 new issues 08:42 paramat will merge game#1732 game#1738 in a moment 08:42 ShadowBot https://github.com/minetest/minetest_game/issues/1732 -- redo new open chests to use all single textures by VanessaE 08:42 ShadowBot https://github.com/minetest/minetest_game/issues/1738 -- Remove set_inventory_formspec in default, remove delay in sfinv by sfan5 08:57 paramat merging 09:02 paramat merged, down to 13 PRs, i like it 09:25 nerzhul if it was same on core :( 09:28 VanessaE thank you, paramat 09:50 VanessaE about #5640, can we please just get it merged or get the corresponding minetest_game commit removed/reverted? 09:50 ShadowBot https://github.com/minetest/minetest/issues/5640 -- Automatic item and node colorization by juhdanad 09:50 VanessaE stop arguing over the semantics of "blocker" versus "milestone" (especially since either one can simply be discarded if warranted, as has happened in the past) 09:53 nerzhul we just need a proper workflow heh :) just on feature freeze if this PR hasn't been approved for a merge paramat should revert on MTG 09:53 VanessaE I know, but workflow discussions are somewhat irrelevant at this point :) 09:54 VanessaE especially since the merge to mt_game happened almost a month ago... 09:56 VanessaE point is, I just went through a flurry of mod updates for my servers (and no, as far as I know there are no Dreambuilder updates that depend on un-merged MT engine pulls ;) ), and I'd like to get an engine build into place that isn't gonna break. 09:56 VanessaE (I have not uploaded said changes, but I would like to, and to issue a Dreambuilder build update as well, but I can't yet) 10:00 Zeno` celeron55, can you please comment on #5740 10:00 ShadowBot https://github.com/minetest/minetest/issues/5740 -- Minetest client memory leak 10:50 celeron55 doesn't ring any bells to me 10:50 VanessaE ok.. paramat can you tell me in one short sentence what is likely to break if I fire-up a fully-loaded Minetest instance with engine and _game at HEAD, but without #5640? because I just did, and couldn't make it crash. 10:50 ShadowBot https://github.com/minetest/minetest/issues/5640 -- Automatic item and node colorization by juhdanad 10:51 VanessaE (everything that uses node colorization in my environment depends on Unified Dyes, so I think it "bypasses" the missing feature as a side effect) 10:53 paramat unfortunately i cannot because i don't understand this hardware colouring work 10:53 paramat i asked juhdanad in the thread though 11:13 VanessaE ... 11:13 VanessaE mmmh 11:43 Fixer interesting, newest server, patched client with updated particle spawner thing, particle count was 0 most of the time, I walked around some big parts of server including spawn, went back home, afked, boom, noticed 350-380 particles all the time o.O where are they comming from, also, fps drop bug, hm 12:11 nerzhul VanessaE, thanks for the return back 12:12 VanessaE ? 12:12 paramat thanks for testing 12:13 VanessaE sure 12:13 paramat sorry that was for Fixer :] 12:14 VanessaE oh ok :) 12:14 VanessaE I was like...what did I test? :) 12:14 nerzhul VanessaE, about 5640 pr 12:15 paramat a server may not be a good place to test, as all sorts of weird stuff might be going on 12:15 VanessaE oh ok. well note that I just did a cursory test of course 12:15 VanessaE I didn't exactly try to fuzz it or anything 12:17 paramat sorry that was for Fixer too, i should address these properly 12:17 VanessaE that's okay, my comment was aimed at nerzhul :) 12:21 VanessaE nerzhul: but again, I did not load 5640, I ran my brief test without it 12:21 VanessaE nerzhul: so don't rely on me here :) 12:25 Zeno` well this confuses things :/ 12:25 Zeno` 2 hours, not moving at all: http://imgur.com/a/MqFUa 12:26 Zeno` maybe mesh_generator_thread.cpp line 323. I have no idea now 12:27 Zeno` the peak at the start is from when Fixer teleported me to a safe place (where I couldn't die) on the server 12:27 Zeno` from that point I didn't touch it at all 12:28 nerzhul Zeno`, seems relatively stable 12:28 Zeno` stable? 12:29 Zeno` it's stably increasing how much RAM it's using if that's what you mean 12:29 Zeno` I'm not going to run massif for 8-15 hours but if that graph trend continues then there is a leak :P 12:32 Zeno` running under massif is much much slower than normal so probably under normal circumstances the slope of that trend is higher 12:41 Zeno` is a RAM usage increase of somewhere between 10 and 20% hour "normal"? 12:42 Zeno` when the player is just standing there doing nothing? 12:42 kilbith I observed that 12:42 kilbith I end up on the swap partition after leaving MT for one hour 12:43 kilbith leaving it up, I mean 12:43 kilbith err, running* 12:44 Zeno` but surely that should not happen 12:47 Zeno` I've spent 6 hours (maybe more) already collecting data so I don't want to do more tonight. Tomorrow I think I revert that cache and see if it still happens 12:48 kilbith Minecraft C++ Edition doesn't take more than 1 GB of memory 12:48 paramat interesting, can't think of why a client would increase memory like that 12:48 kilbith and it's hell more optimized than MT 12:49 kilbith so we can do it 12:49 Zeno` paramat, the graphs show what's using the memory 12:49 Zeno` it's the mesh cache 12:49 Zeno` why? I dunno yet 12:50 kilbith fortunately we have a serial-profiler like you 12:52 paramat yeah weird 12:53 paramat erm which colour is the mesh cache? 13:01 Zeno` note that time() does not return an int. int t_now = time(0); 13:02 Zeno` (and pedantically it should not have the argument 0 as well, but that's 99% surely not the problem) 13:02 Zeno` I do wonder if it being an int is the problem though. hmm 13:03 Zeno` paramat, I dunno. Look at the legend :D 13:04 Zeno` (line 281 of mesh_generator_thread.cpp) 13:06 Zeno` maybe int is ok though. I dunno. (the return value of time() is in seconds so technically it should be) 13:19 nerzhul kilbith, minecraft uses less memory also because it's not modular and doesn't use many strings like minetest it help greatly for performance 13:19 nerzhul performance => memory usage 13:19 nerzhul and yes, time_t t = time(NULL) is the real call 14:31 kilbith https://lut.im/fDuE0xumFX/cujzDdfT6ygixaGh.png 14:31 kilbith new regression likely caused by: https://github.com/minetest/minetest/commit/9b5effff 14:31 kilbith it occurs when you create a new worlds, enable all mods and save 14:31 kilbith -s 14:48 GreenDimond The HOMETOWN memleak seems to be getting worse. 14:50 GreenDimond Command lag of about 5 seconds. 14:51 Fixer GreenDimond: Edad runs it under gdb, thats why so laggy 14:51 GreenDimond It normaly isnt this laggy 14:51 Fixer GreenDimond: under gdb it is 14:51 GreenDimond Since when is it run under gdb? 14:52 Fixer today it seems 14:52 Fixer he wants to catch a backtrace of crash server is having 14:52 GreenDimond Ah 14:53 GreenDimond I crashed it earlier to restart it 14:53 GreenDimond running /teleport playername 0,30000000,0 crashes it 14:53 GreenDimond so I ran /teleport GreenDimond 0,30000000,0 and boom server restart 14:54 Fixer GreenDimond: why do you do that? 14:54 GreenDimond Normaly I dont 14:54 GreenDimond But the lag was so bad 14:54 Fixer this is known bug I beglieve 14:54 GreenDimond I discovered the bug yesterday 14:54 Fixer nevermind the lag 14:54 Calinou when you think something is a bug, you buglieve it, Fixer 14:54 Calinou ;) 14:54 Fixer he will switch to release build later 14:55 GreenDimond just doing /teleport 0,3000000,0 warns you and doesnt let you go there 14:55 GreenDimond but tping players crashes 14:55 GreenDimond build release will break a lot :( 14:57 GreenDimond Have fun Fixer xD 15:08 Fixer is it possible to somehow output what exactly particles are generated? i have 200-400 particles right now constantly on Hometown and I'm at my base having no idea from where they coming 15:10 red-001 ? 15:11 Fixer list of particles, their coordinates, names... 15:25 nerzhul Fixer, i told your to not run under GDB but enable your kernel coredumps 15:25 Fixer nerzhul: windows... 15:26 Fixer it has dumps too, but i don't want having any deals with huge msvc/windbg etc 15:26 nerzhul on windows there are minidump non ? 15:26 nerzhul no* 15:26 Fixer it has 15:26 Fixer nerzhul: but I compile via mingw 15:26 Fixer it is gdb friendly only iirc 15:27 nerzhul oh i don't said it's not friendly :p 15:33 Fixer this may help me on windows: https://github.com/jrfonseca/drmingw (but i don't really want to since my known crashes are backtraces) 15:58 kilbith strangely the MinimapUpdate thread is up even when you disable that feature 15:59 nerzhul oh 15:59 nerzhul can you open an issue 15:59 nerzhul ? 17:21 Krock kilbith, seen the new problem when enabling all mods, will be fixed in #5770. 17:21 ShadowBot https://github.com/minetest/minetest/issues/5770 -- Mainmenu: Fix problems with recently added "game" type by SmallJoker 17:23 nerzhul Krock, as this fix is used in many places can you add a function for it ? 17:23 nerzhul i mean the repeated string.find(mod.typ, "game") a function name with a good name make sense 17:24 Krock hmm yeah. 17:25 Krock perhaps adding another property to the modlist would help too 17:33 Krock heh. A typo in a typo fix :3 17:35 kilbith nerzhul, it's irrelevant 17:35 kilbith my remark was about avoiding `find` because it iterates over the whole string under the hood 17:37 Krock workaround applied 17:38 nerzhul kilbith, i don't talk about lua function, just the repeated condition which is relevant for a function 17:38 nerzhul condition check* 17:39 kilbith not really 17:39 nerzhul kilbith, Krock #5771 17:39 ShadowBot https://github.com/minetest/minetest/issues/5771 -- When minimap is disabled in configuration, really disable it by nerzhul 17:40 nerzhul i choose to use the minimap pointer as a test if minimap is disabled instead of calling config because minimap disable mean no object instanciated 17:44 Krock as long they have the same meaning it's no problem 17:45 Krock what if the minimap is enabled but not shown? 17:46 Krock it will continue adding blocks, not? 17:46 nerzhul i don't look at this part, but i think you are right because i saw strange things when doing this disable, i think you look at the PR and see a thing about this :) 17:51 Krock nvm, the blocks wouldn't be added to the minimap if they're not passed in addBlock() 17:51 Krock since it only is called on mesh update 17:51 nerzhul yeah i can improve it to really increase FPS when disabled but not the purpose of this pr 17:59 Krock only found one thing, otherwise good 18:01 nerzhul okay just documentation :) 18:08 nerzhul Krock, is the doc fix good for you ? 18:10 Krock yes. However, I'd like if you don't merge this PR instantly and give the other devs the possibility to look at it 18:12 nerzhul yeah i will merge it tomorrow , no problem 18:14 nerzhul Krock, just is_game fit, content is implicit by the "object" type 18:15 nerzhul no ? 18:16 Krock well, "is_game" alone doesn't look enough informative to me 18:16 nerzhul ok 18:17 Krock but technically you're right. Just wanted to make sure it's clear what the param means 18:54 nerzhul pushing a trivial fix on ScriptApiBase::stackDump pointed by GCC 7.1 (conversion overflow) 21:01 Hijiri any chance for last minute reviews on #5612? 21:01 ShadowBot https://github.com/minetest/minetest/issues/5612 -- Allow overriding tool capabilities through itemstack metadata by raymoo 21:01 Hijiri #5612 does ShadowBot not recognize it if it has a ? 21:01 ShadowBot https://github.com/minetest/minetest/issues/5612 -- Allow overriding tool capabilities through itemstack metadata by raymoo 21:01 Hijiri oh nevermind 21:01 Hijiri sorry