Minetest logo

IRC log for #minetest-dev, 2015-11-04

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:21 est31 joined #minetest-dev
00:26 kahrl_ joined #minetest-dev
00:27 paramat joined #minetest-dev
00:28 paramat will merge game#718 in 20min
01:01 asl joined #minetest-dev
01:12 paramat now merging game 718
01:18 paramat done
01:27 est31 #3336
01:27 est31 https://github.com/minetest/minetest/pull/3336
02:46 paramat left #minetest-dev
03:46 Sokomine joined #minetest-dev
04:08 Robby_ joined #minetest-dev
04:09 ShadowBot` joined #minetest-dev
04:10 Player-2 joined #minetest-dev
04:25 exoplanet joined #minetest-dev
04:33 sloantothebone joined #minetest-dev
04:56 paramat joined #minetest-dev
04:59 paramat ugh docs #3337 can anyone approve so i can merge before conflicts arise?
05:00 paramat https://github.com/minetest/minetest/pull/3337 Mgfractal: Add documentation to conf.example and settingtypes
05:03 hmmmm approved
05:04 paramat woo thanks
05:04 hmmmm [04:34 PM] <paramat> hmmmm while i remember, we need a way to copy an .mts schematic into the luavoxelmanipulator (without the lighting update). browsing the forums i'm seeing a lot of modders using 'place schematic' during mapgen, sometimes multiple times, which is slow and causes lighting bugs
05:04 hmmmm agh fine, I'll do that
05:04 paramat fun!
05:05 hmmmm actually you know what
05:05 hmmmm I'm going to make you do it
05:05 hmmmm i'll give you a hint:  Schematic::blitToVManip() takes an MMVManip *
05:05 JohnnyComeL8ly Guys... https://forum.minetest.net/viewtopic.ph​p?f=6&amp;t=13534&amp;p=197200#p197200
05:05 paramat heh i suspected you might ask me
05:05 hmmmm this is so easy
05:05 paramat ok i'll try
05:10 paramat flickering bands of incorrect sky colour?
05:14 JohnnyComeL8ly Yeah, pretty much.
05:15 JohnnyComeL8ly I can't think of what a good estimate for the hertz of the flicker is...
05:15 JohnnyComeL8ly paramat:
05:16 paramat see if you can find someone else who can reproduce the bug
05:17 paramat and ideally open an issue at github
05:18 JohnnyComeL8ly Paramat: I assume you don't have this issue, right?
05:19 paramat correct
05:20 paramat now merging 3337
05:26 paramat done
05:27 est31 joined #minetest-dev
05:32 est31 joined #minetest-dev
05:34 est31 aww super great what did I break
05:35 est31 Johnnny that issue was my regression 100%
05:37 est31 argh
05:38 est31 I should perhaps test the code I write
05:47 est31 https://github.com/est31/minetest/commit/​e4d6c4c131b9d800bd604ea4ef9b7511bd83ebe7
05:47 est31 Tested it and it works
05:50 est31 Yea thats really the only place where my commit changed something beyond rename
05:50 est31 and thats where I fucked up
05:52 est31 pushing in 20 minutes.
05:52 paramat cool a fix :)
05:56 est31 new version https://github.com/est31/minetest/commit/​201988ea0d2fbe2bfde8508fb8bad3645e1e3c03
05:57 est31 (added a comment)
05:57 est31 you never know, the next time there is a cleanup this might become not thread safe
05:57 est31 and I wouldnt even blame anybody
05:58 est31 there is no indication this isn't under the lock
05:58 est31 you'd have to look at the type
05:59 hmmmm maybe this issue would have come up if we looked at Thread Sanitizer
06:00 hmmmm est:  just please make sure it doesn't happen again
06:00 hmmmm est:  I have no idea why you would push something without at least running minetest and making sure it doesn't crash instantly
06:01 hmmmm whenever I make a change, I make sure it compiles (on my configuration at least), does what it should without crashing, and then I run around the map for 5 minutes
06:02 hmmmm paramat, did you start working on the schematic-to-vmanip feature?
06:03 paramat no it's late (early) here, watching MC videos/anime
06:04 hmmmm did you want to try it?
06:07 paramat yeah i don't mind attempting it, but busy so it won't be very soon
06:07 hmmmm 'cause if you don't mind, I'll do it right now
06:10 paramat ok thats cool
06:13 JohnnyComeL8ly joined #minetest-dev
06:16 paramat left #minetest-dev
06:21 jin_xi joined #minetest-dev
06:29 hmmmm you know what
06:29 hmmmm i think i found the source of the mapgen memory leak
06:29 hmmmm *the* memory leak
06:40 Hunterz joined #minetest-dev
06:55 gregorycu joined #minetest-dev
07:18 book`_ joined #minetest-dev
07:23 celeron55 joined #minetest-dev
07:42 Luke joined #minetest-dev
07:43 Calinou joined #minetest-dev
08:58 nrzkt joined #minetest-dev
09:00 hmmmm paramat, when you're back:  https://github.com/minetest/minetest/pull/3339
09:04 Hunterz joined #minetest-dev
11:47 Taoki[mobile] joined #minetest-dev
11:52 sloantothebone joined #minetest-dev
11:57 Hunterz joined #minetest-dev
12:00 Hunterz joined #minetest-dev
12:02 Hunterz joined #minetest-dev
12:03 kaeza joined #minetest-dev
12:05 Hunterz joined #minetest-dev
12:15 jin_xi joined #minetest-dev
12:16 crazyR joined #minetest-dev
12:21 PilzAdam joined #minetest-dev
12:23 IS_No2035 joined #minetest-dev
12:39 Issue2035 joined #minetest-dev
13:36 zat joined #minetest-dev
13:53 luizrpgluiz joined #minetest-dev
13:54 VargaD joined #minetest-dev
13:56 VargaD_ joined #minetest-dev
14:01 luizrpgluiz left #minetest-dev
14:18 luizrpgluiz joined #minetest-dev
14:24 Darcidride joined #minetest-dev
14:24 iqualfragile joined #minetest-dev
14:29 iqualfragile joined #minetest-dev
14:32 nore joined #minetest-dev
14:40 gregorycu joined #minetest-dev
14:42 gregorycu nrzkt: Are you the performance monkey around here?
14:43 gregorycu Hoping that someone who knows how to profile can try #3331, provided they are not using MSVC
14:55 kaeza joined #minetest-dev
15:01 JohnnyComeL8ly joined #minetest-dev
15:40 rubenwardy joined #minetest-dev
15:43 est31 joined #minetest-dev
15:54 CraigyDavi joined #minetest-dev
16:02 hmmmm joined #minetest-dev
16:04 blaze joined #minetest-dev
16:14 Soni joined #minetest-dev
16:36 luizrpgluiz left #minetest-dev
16:52 Krock joined #minetest-dev
16:56 jin_xi joined #minetest-dev
17:01 DFeniks joined #minetest-dev
17:03 Amaz joined #minetest-dev
17:05 Hunterz joined #minetest-dev
17:48 ElectronLibre joined #minetest-dev
18:00 Robert_Zenz joined #minetest-dev
18:20 Supertanker2 joined #minetest-dev
18:23 proller joined #minetest-dev
18:36 Soni joined #minetest-dev
18:43 Darcidride joined #minetest-dev
18:50 nrzkt joined #minetest-dev
19:03 proller joined #minetest-dev
19:08 younishd joined #minetest-dev
20:01 paramat joined #minetest-dev
20:03 paramat hmmmm okay will test 3339 soon. how can that be the cause of mapgen memory leak when the 'place schematic' api isn't used in core mapgen?
20:05 paramat PilzAdam is this correct for when the default is the absence of all flags? https://github.com/minetest/minetest/bl​ob/master/builtin/settingtypes.txt#L937
20:06 PilzAdam if the default setting is the empty string, then it should be just the empty string
20:06 PilzAdam so only two spaces between "flags" and "julia"
20:06 paramat ok
20:06 paramat also how do we cope with v3fs? i've added them in as 'todo'
20:07 PilzAdam use the type "string" for now
20:07 PilzAdam and add a description of the format in the comment
20:07 paramat ok
20:16 Lunatrius joined #minetest-dev
20:20 Warr1024 joined #minetest-dev
20:46 paramat left #minetest-dev
21:48 Darcidride joined #minetest-dev
21:55 hmmmm paramat, well maybe a common default mod uses place_schematic
21:55 hmmmm like for trees or whatever
21:55 hmmmm yup
21:56 hmmmm games/minetest_game/mods/default/trees.lua usees place_schematic in 4 places
22:19 paramat joined #minetest-dev
22:19 paramat ok. reviewed your PR
22:21 paramat only sapling growth uses minetest.place_schematic, anyway thanks for this and good to see a leak fixed
22:22 hmmmm hmm
22:22 hmmmm about your first review comment
22:22 hmmmm there's this line in the lua doc:  * After execution, any external copies of the VoxelManip contents are invalidated.
22:23 paramat ah
22:24 paramat which is why it can't be before 'set data'? that's not obvious though
22:27 paramat so it's positioned after 'set data' and before 'set lighting' and 'calc lighting'
22:28 PilzAdam about sapling growth: there is a bug where the lighting upgrade sometimes causes the block to turn completely dark
22:29 PilzAdam https://forum.minetest.net/v​iewtopic.php?f=6&amp;t=13536 I also noticed this recently in singleplayer
22:29 PilzAdam it happens when I was mining underground, so it may be caused if the block above the growing tree is not loaded
22:30 nrzkt PilzAdam, notice same problem too
22:30 PilzAdam IIRC I noticed already a year ago, or so
22:30 paramat yeah is known, 'place schematic' uses a lighting update that is known to cause shadow bugs
22:30 PilzAdam dunno if it was fixed and reintroduced or if it's the same bug
22:32 paramat it uses map->updateLighting() which also causes shadow bugs with l-system trees
22:33 paramat growing many saplings rapidly increases the chance of shadows
22:33 paramat this is where 'catch_up = false' would help
22:33 paramat but best to fix the bug
22:36 paramat '// TODO: Optimize this by using Mapgen::calcLighting() instead' that might fix it
22:36 paramat and would be faster
22:38 paramat 'update map' used with the non-mapgen-object vm also uses map->updateLighting()
22:42 hmmmm paramat:  that is non-trivial
22:43 hmmmm Mapgen::calcLighting doesn't do lighting updates for neighboring blocks
22:43 paramat ok
22:43 hmmmm also it assumes an "overtop" block
22:44 rubenwardy joined #minetest-dev
22:49 rubenwardy joined #minetest-dev
23:03 Player2 joined #minetest-dev
23:17 paramat Settingtypes: Correct documentation for mgfractal #3340 https://github.com/minetest/minetest/pull/3340
23:19 paramat will merge that later as is specified by PilzAdam
23:31 paramat but there's a problem. because the default flags for mgfractal is no flags, the dialogue doesn't appear for editing, i'll try fix
23:37 paramat i'll return it to 'nojulia' since that works
23:43 paramat will merge that in 20mins

| Channels | #minetest-dev index | Today | | Google Search | Plaintext