Time  Nick       Message
10:54 Zeno`      hi all
10:54 Zeno`      It has been suggested to me that I should have explained why I removed myself as a dev
10:55 Zeno`      Some people know, because I sent them a memo
10:55 Zeno`      But for those that don't, it can be summed up as that I think: new features should only be added when current features are bug free. I realise that this would upset the users, though
10:55 Zeno`      I have no problem with the project or other devs
10:56 Zeno`      Mainly, though, I felt trapped chasing bugs
10:57 Zeno`      I do think the project is worthwhile though, it's just that at this point in time I can't motivate myself to contribute, so it didn't seem fair that I had commit access
17:10 est        pushing in ~ 20 mins https://github.com/est31/minetest/commit/5643b9b9ed3ec39f90e3a7c9bf09bc255e0bcef3
17:10 est        can be reverted sooner or later
17:10 est        but needing it right now
17:12 nrzkt      est: ok
17:12 est        can I merge nrzkt?
17:13 nrzkt      yes
17:13 est        thx
17:15 est        wanting to merge this too, if travis works https://github.com/minetest/minetest/pull/3422
17:15 est        is that okay?
18:44 est31      anybody here to look at 2 PRs?
18:45 est31      first
18:45 est31      https://github.com/minetest/minetest/pull/3428
18:45 est31      then
18:45 est31      https://github.com/minetest/minetest/pull/3417
18:45 est31      both are okay for me
18:46 est31      for the second, the concerns I had and PA had too are fixed now
20:08 paramat    +1 for 3422 and 3428, perhaps i'll merge these later
20:08 paramat    oh 3428 needs squashing
20:15 paramat    for 3417 i feel someone else should check the code
20:23 VanessaE   #3428
20:23 VanessaE   damn it ShadowNinja fix the bot
20:59 paramat    i'll also merge https://github.com/minetest/minetest/pull/3418 later
21:39 VanessaE   paramat: did you say the other day you had some code around you wanted me to try regarding tree lighting bugs in mgv7?
21:46 paramat    ugh =) i did but further testing showed the fix broke other stuff, sorry
21:46 VanessaE   it's okay
21:46 VanessaE   I'll just harass you about it off and on. ;)
21:47 paramat    see https://github.com/minetest/minetest/issues/3421
21:47 VanessaE   it's probably related to the y=+63 glitch you mentioned before
21:47 * VanessaE looks
21:48 paramat    no that's independent
21:49 VanessaE   "I created a branch with a version of mgv7 without overgeneration and tried the moretrees mod, the shadows were fixed. "  well at least you have a theory
21:49 paramat    although having no overgen fixes the moretrees shadows, the issue should be fixed another way by working on lighting code
21:50 VanessaE   yeah a lot of folks think the lighting code needs .... revamped.
21:50 VanessaE   but that's non-trivial
21:50 paramat    also of course moretrees is using l-system during mapgen, which it shouldn't, so it's partly a mod thing. you're lucky it works in mgv6
21:50 VanessaE   well
21:50 VanessaE   there's a caveat there:
21:51 paramat    well delayed a little
21:51 VanessaE   moretrees can be configured NOT to grow trees immediatley.
21:51 VanessaE   immediately*
21:51 VanessaE   (in that alternate mode, it places saplings that trigger spawn_tree() calls 10-20 seconds after mapgen of that block)
21:52 paramat    i confirmed that the shadows are fixed if the upper mapchunk is generated first, so if you place saplings grown by ABM a few seconds later that might work
21:52 VanessaE   originally I added that feature to reduce mapgen lag, but since my deferred code, it didn't matter.
21:52 paramat    aha
21:52 VanessaE   OH that reminds me:
21:52 VanessaE   moretrees does NOT actually generate *during* mapgen
21:52 VanessaE   not since the deferred code.
21:52 paramat    yep i know
21:52 VanessaE   (including spawn_tree() calls)..  they're done during globalstep
21:52 VanessaE   oh ok
21:53 paramat    but it's still fast enough that the mapchunk above is not generated first
21:53 VanessaE   true
21:53 paramat    try the sapling method in mgv7, that might fix the shadows
21:53 VanessaE   I'll give it a shot
21:53 VanessaE   meanwhile, it almost sounds like you need a more MC-like mapgen method, though maybe not necessarily a full 16 mapblock column
21:54 paramat    ideally 'updateLighting' needs work to be more robust
21:54 paramat    and a veersion of l-system without a lighting update
21:55 paramat    comments in the code show it is a TODO, it was ment to be another decoration type
21:55 VanessaE   that might help but there's a problem - it's not possible to know ahead of time when the very last tree in a chunk will be generated
22:00 paramat    (version)
22:00 paramat    yeah
22:02 VanessaE   I suppose it might be feasible to pre-sort biome_lib's registered actions list to group spawn_tree() calls together.  But how would one know when to actually run one of those calls *with* lighting?
22:02 paramat    since they are often over 16 nodes tall, it's not ideal to place them as decorations withing the mapgen vm because that has limits at mapchunk edges. best is to grow l-system trees from sapling using an ABM, then each tree gets a vm big enough for it's height
22:02 paramat    (within)
22:03 paramat    yeah agreed
22:03 VanessaE   that's what they do in the alternate mode
22:03 VanessaE   they're placed in globalstep, but then an ABM comes around some time later and grows them with spawn_tree()
22:04 VanessaE   maybe I could read the mapgen type and enforce the alternate mode if it isn't mgv6 or something
22:04 paramat    the version without a lighting update should be the one done as a decoration
22:05 paramat    yes all other mapgens other than mgv6 will probably have shadow bugs, so they should use the sapling method
22:07 paramat    the big lighting issue is the y = 63 one, i just can't see what it is about non-mgv6 mapgens that causes it https://github.com/minetest/minetest/issues/2759 it's bad
22:08 VanessaE   yeah, saw the screenshots
22:08 VanessaE   it's like those floating shadows caused by I guess "cave air" up high
22:08 VanessaE   in that they suck :)
22:09 VanessaE   the shadows in your screenshots are easily worse than the moretrees thing.
22:09 VanessaE   btw the moretrees glitch even happens with smaller trees like palms, which barely stand one mapblock tall
22:12 paramat    yes i think that's because l-system automatically emerges a large and tall volume that would touch the mapchunk top
22:12 VanessaE   ah
22:12 paramat    3-4 blocks tall
22:12 VanessaE   I recall RBA saying that by default it emerges 3 mapblocks tall
22:12 paramat    ha
22:12 VanessaE   but I don't remember what he said (if anything) about the horizontal range
22:13 VanessaE   what would happen if you tweaked it to force say, 6 mapblocks in height?
22:13 VanessaE   clearly it would exceed a chunk boundary every time then
22:14 paramat    more chance of shadow i think
22:14 VanessaE   hm no, that wouldn't help anyway (what if the bottom of the stack was just below the lower boundary)
22:15 paramat    well i don't understand it, thought i did, for a day
22:15 VanessaE   it's like quantum mechanics
22:15 VanessaE   if you think you understand it, you haven't studied it enough :)
22:16 paramat    for the y = 63 thing, i even made mgv7 as mgv6-like as possible: removed overgeneration and used the exact lighting calculation mgv6 does, but still shadows. mysterious
22:16 paramat    ahh it's good to see est here
22:16 VanessaE   wb est.
22:16 est31      hi
22:17 est31      updated my pr about luajit detection
22:18 paramat    every dev still active is so appreciated currently
22:18 paramat    okay
22:18 est31      tbh I'm trusting susnux here, but I doubt they do it badly
22:19 paramat    can it be tested?
22:19 paramat    well we'll soon know i guess
22:20 est31      the buildbots say its okay
22:22 paramat    hmmmm you are much missed
22:27 harrison   oberon event will happen in Boston and Zurich simultaneously on some Saturday in April
22:27 harrison   exact date, venue yet to be chosen
22:27 harrison   please join g+ group "Oberon/A2/Bluebottle" to submit talk proposals and plan conference/meetup
22:27 harrison   i am organizing the Boston half of the event
22:27 harrison   and have no experience at all I haven't even ever been to a coding meetup
22:27 harrison   please help me
22:28 est31      harrison, its quite likely that the stuff you talk about is off-topic
22:28 VanessaE   um.... this is minetest-related how?
22:28 VanessaE   ninja'd/
22:30 est31      well, for the case there is sb living near boston wanting to promote minetest, now is your chance, contact harrison.
22:31 VanessaE   haha
22:32 harrison   i know it is not very relevant to the channel
22:32 harrison   i am hoping for genneral advice
22:32 harrison   on organizing a coding meetup / informal conference
22:33 harrison   like, advise me bro
22:37 paramat    -> #minetest
23:34 rubenwardy Will sending particle spawners individually to players using playername cause lag problems? http://rubenwardy.com/minetest_modding_book/lua_api.html#particlespawner-definition-add_particlespawner
23:53 paramat    thanks hmmmmm i will make the shadow bool optional
23:53 est31      yea I too think that optional is better here
23:54 est31      but idk about mapgen code design
23:57 paramat    yep
23:57 paramat    i mean yes optional is best