Time Nick Message 03:42 paramat can confirm MTG is ready for release 03:43 paramat merging #8260 03:43 paramat "README: Remove references to Minecraft and InfiniMiner" 03:54 paramat closed #8231 "Check if minetest.conf.example needs updating" so all blockers cleared 03:55 GreenDimond :0 03:56 * GreenDimond runs around in circles "it's here it's here! er.. almost here!" 03:56 VanessaE wait....... 03:56 VanessaE don't go releasing it right now 03:56 VanessaE wait until morning 03:57 GreenDimond lol im sure they wont 03:57 paramat lol 03:57 GreenDimond They've gotta talk about it first x) 03:57 GreenDimond should do a lightning round of testing 03:57 GreenDimond I can ping everyone on Discord and tell them to full send on the latest build of MT >:) 03:58 VanessaE I'm just saying people need to be available to field questions 04:16 paramat and here's a PR that i feel needs considering for release https://github.com/minetest/minetest/pull/8303 "Change 'num_emerge_threads' default to 1". back this evening 04:17 VanessaE no 04:17 VanessaE that's not a bug fix and we're in hard freeze now 04:19 GreenDimond save it for 5.0.1 ;D 04:23 VanessaE paramat, so either you guys lift the feature freeze, or that ^ 04:26 argyle77 It seems like a bug fix 04:26 GreenDimond but not an essential one, if it is 04:30 argyle77 Actually, it seems like it would be a bug fix, but I have 4 cores and couldn't reproduce the bug with the default 0 setting. 04:34 argyle77 Therefore, I don't have any basis to argue with GreenDimond's statement. 04:38 cheapie I've never actually had any crashing with the default and 8 cores. 04:38 VanessaE me neither. 04:40 argyle77 Does something already know not to use multiple cores with sqlite by default? My cpu usage didn't seem as high with 0 as with 2 explicitly set. 04:45 argyle77 Or maybe it doesn't know I have 4 cores. num_emerge_threads = 2 and num_emerge_threads = 0 definitely behaves differently on my 4 core machine. 05:13 sofar I put an abort in unsigned int Thread::getNumberOfProcessors() 05:13 sofar and it doesn't get aborted 05:14 sofar conclusion: the code is unused entirely (num_emerge_threads = 0) 05:16 sofar 2019-03-01 21:15:37: ERROR[Main]: emerge threads: 1 05:16 sofar 8 core cpu, how does nthreads =1 when num_emerge_threads = 0? 05:17 sofar hold on one more try 05:17 GreenDimond black sorcery 05:17 sofar nah it's really num_emerge_threads = 0 -> nthreads = 1 05:19 sofar `nthreads = Thread::getNumberOfProcessors() - 2;` 05:19 sofar evaluates to `0` on my machine 05:19 argyle77 maybe a thread is launched before the number of allowed threads are checked 05:21 sofar what the 05:21 sofar if I errorstream << Thread::getNum... again it hits the abort9) 05:21 sofar so, something is optimized away? 05:28 argyle77 if (nthreads < 1) 05:28 argyle77 nthreads = 1; 05:30 argyle77 it seems like if you explicitly set it to 0 getNumberOfProcessors should execute 05:31 argyle77 but if you leave it out, it will always be 1. 05:32 sofar the issue is that g_settings->getS16NoEx("num_emerge_threads", nthreads) always evaluates to `false` 05:33 sofar I get `8` properly from Thread::getNumberOfProcessors() 05:33 argyle77 a bug that avoids another bug. 05:33 sofar yup 05:33 sofar on top of that, with sqlite, it should really be forced down to 1 (at least by default) 05:34 argyle77 it may not just be sqlite though. I got the same crash with postgresql 05:34 cheapie I'd think that, with sqlite, a setting of 0 or 1 should make it use 1 thread and >1 should give an error. 05:41 argyle77 I gotta go. Gnight all! 05:42 GreenDimond o/ 05:42 * GreenDimond waves arm into oblivion 05:46 sofar o~ 06:01 sofar #8304 06:04 sofar paramat: if you read this, my analysis shows that #8303 is correct but technically not needed because #8304 shows that the risk is never exposed unless users have a different value (which is entirely unlikely), and even then neither #8303 nor #8304 prevent the risk of a user setting the value to e.g. 2 06:04 sofar so #8303 doesn't need to be in 5.0.0 09:56 nerzhul merged #8305 16:45 Krock Can I somehow create a temporary InvRef (non-detached)? 16:45 rubenwardy unfortunately not 16:45 rubenwardy you could create a detached one then delete it 16:46 Krock well, that's not supported in 0.4.x 16:46 Krock (delete it) 16:46 Krock and deep copy isn't possible either 22:00 sofar oh err 22:00 sofar Calinou: I thought that was a minetest-mods PR 22:00 sofar haha, dumb me 22:00 Calinou speaking of minetest-mods, can you enable Travis CI on moreblocks, moreores and maptools please? 22:01 Calinou (I can't do it on my own) 22:01 Calinou also, I can't seem to edit the GitHub topics on moreblocks and maptools (but I can on moreores) 22:02 Calinou anyway, thanks for merging my PR :) it's just a deprecation fix anyway, and it passed CI 22:02 Calinou I have a few other website maintenance PRs in planning too 22:03 sofar I gave you admin instead of write on those 2 repos 22:03 Calinou (removing unnecessary/unused libraries, updating libraries, possibly migrating to Bootstrap 4) 22:03 Calinou thanks :) 22:04 sofar how does one enable travis now that integrations are outdated, anyway? 22:05 Calinou to enable it on my repository, I go to my Travis CI settings page and check the repository 22:05 * sofar waves to sfan5 with https://github.com/minetest/master-server/pull/29 22:05 Calinou but I can't do that on minetest-mods repositories 22:06 paramat sofar interesting thanks, might be best to not merge #8303 before release then? however you write "this is the right step forward for 5.0." and " #8303 doesn't need to be in 5.0.0" so i'm a little confused 22:06 sofar haha, yes, it is confusing 22:06 sofar fixing the whole stack of bugs is wanted, and I believe setting the num_emerge_threads to 1 is good 22:07 sofar as a precaution 22:07 paramat still need to study your findings properly though to understand 22:07 sofar (as default) 22:07 sofar yeah it's not a simple bug 22:07 paramat ah ok so i should merge 22:07 paramat #8303 22:07 sofar no, there is no need to merge in 5.0 22:07 sofar but after, yes 22:07 paramat ok =) 22:08 sofar I tried to explain that the actual cpu thread count code is never used due to a logic error 22:09 sofar there's a few ways to fix it, I made a PR to fix it in one way 22:09 paramat once i understand it should be clear to me. yes reading irc stuff 22:15 sofar Calinou: it did something (travis)