Time  Nick           Message
06:21 hmmmm          hrm
06:21 hmmmm          I wonder if people are okay with map generation limits being so rough
06:22 asl97          rough?
06:23 hmmmm          by default it's "something a little below 31000"
06:23 Megaf          can they be increased now?
06:23 hmmmm          what the real map limit is depends on a bunch of things like block offsets and chunk sizes
06:23 Megaf          since now it is possible for us to limit it
06:24 hmmmm          no, it doesn't work that way.
06:28 asl97          uh....  it depend on the chunksize?   i thought it depend on the 16bit int thing?
06:29 hmmmm          it depends on a lot of things tbh
15:05 est31          about #3202, it seems almost good for me
15:05 ShadowBot      https://github.com/minetest/minetest/issues/3202 -- Small Bugfix throw error and not make cheesy maps for minetest/minetest#3196 by netinetwalker
15:06 est31          we should only call freeReplyObject(reply);
15:06 est31          and add braces ofc
15:08 est31          gonna merge it with my changes
15:22 est31          or...
15:23 est31          gonna try first whether it still compiles
18:03 hmmmm          hrmm
18:03 est31          hi hmmmm!
18:03 est31          time to look at PRs?
18:04 hmmmm          findSpawnPos is FUBAR
18:04 nrzkt          hmmmm: agree :p
18:04 hmmmm          sure, we can do PRs
18:04 hmmmm          hrmm
18:05 hmmmm          I think we're going to have to limit the spawn range to a much smaller area than (-1000, -1000) (1000, 1000)
18:06 hmmmm          as it happens right now, most of those further out positions wouldn't ever be valid if they hadn't been explored by other players yet
18:06 hmmmm          because they'd be filled with CONTENT_IGNORE
18:07 hmmmm          so another downside to this is that new players spawning ends up bloating a world by filling it up with many blank but existing blocks
18:07 hmmmm          on the other hand, if I were to really allow it to generate map at those positions, you'd have an even more incredible amount of map bloat due to player spawning
18:07 est31          it always puts me onto a valid position when I spawn in singleplayer
18:07 hmmmm          and it would take an unreasonable amount of time to find the spawn
18:08 hmmmm          yes, it just so happens that your map generator produces sane output, but there's no guarantee at all
18:08 hmmmm          I wonder, does anybody have any objections to making the spawn range much smaller than it is right now
18:08 hmmmm          ?
18:09 nrzkt          hmmmm agree to -500 <-> 500 range it's a 4x little zone
18:09 hmmmm          way smaller than that.
18:09 hmmmm          more like -100 to 100
18:09 nrzkt          sometimes you have a lot of water in this area
18:10 est31          hmmmm, is it configureable with a setting?
18:10 Calinou        should be
18:10 hmmmm          you could set static_spawnpos
18:10 est31          if not, it should be
18:10 Calinou        -250/250 should be OK.
18:10 Calinou        by far
18:10 est31          then we can use some default you deem as ok, i dont really care
18:11 est31          from what I see, it does start mapgen, no?
18:11 est31          i mean players dont spawn in empty world do they
18:11 hmmmm          no
18:11 est31          also, what would change if they spawn closer to each other
18:11 hmmmm          that's all done thanks to GetNextBlocks
18:11 est31          the first one still would have a shitty experience
18:11 nrzkt          hmmmm: yes, but if i doesn't want to have a static spawnpoint and i get many water it's a problem
18:15 hmmmm          i'm retrofitting findSpawnPos with emerge callbacks now
18:16 est31          wtf
18:16 hmmmm          the strategy for finding a spawn position is now to treat CONTENT_IGNORE as an invalid block to spawn on
18:16 est31          that would mean it bloats the candidate area, wouldnt it
18:16 hmmmm          with map_generation_limit the player can spawn outside of the map entirely
18:16 est31          except we have a way to emerge without getting a block saved
18:16 hmmmm          this would also fix the age-old problem of players spawning inside trees
18:17 hmmmm          well the whole idea to reduce bloatage is to try the entire range of non-blank existing blocks first
18:17 hmmmm          if that fails, queue a generation and have a callback
18:18 hmmmm          on the callback, add a "hey we generated a new potential area for spawning"
18:18 hmmmm          notification to the server
18:18 hmmmm          and then next server step it'll try again with those pending new areas it knows about
18:19 est31          that strategy is bad
18:19 hmmmm          so in order to make this efficient, attempted spawn locations need to be centered around a known existing area
18:19 hmmmm          you got something better?  speak up
18:19 est31          if you take existing blocks on the first step, it will spawn players where other players already were
18:19 hmmmm          that's exactly how it works right now
18:20 hmmmm          it's just that i'm going to exhaust the entire range of already existing blocks instead of just choosing a blank block to throw them in
18:20 hmmmm          which may or may not be part of the map
18:20 hmmmm          or it might be a tree for all you know
18:39 paramat        mgv7 has terrain scales up to 500 so i would say a spawn range of -250 to +250 minimum
18:47 paramat        with a smaller range we might need to remove the requirement that spawn is at a low altitude (< y=8 or whatever it is)
18:49 paramat        i'd like to merge #3203 later, any approvals?
18:49 ShadowBot      https://github.com/minetest/minetest/issues/3203 -- Decorations: Remove error message 'chunksize not divisable by sidelen' by paramat
18:49 est31          okay for me, if what you say in the PR is right
18:50 est31          (which likely is the case)
19:00 est31          okay hmmmm what about #3192
19:00 ShadowBot      https://github.com/minetest/minetest/issues/3192 -- Clarify radii and distance types in documentation by t4im
19:01 est31          I've checked the radii they seem to match what t4im said
19:01 est31          and the gitignore commit is ok for me as well
19:02 hmmmm          maximum metric?
19:02 hmmmm          ahh square
19:02 hmmmm          okay, sounds good
19:05 paramat        sfan5 ShadowNinja any comments for game#682 game#681 ?
19:05 ShadowBot      https://github.com/minetest/minetest_game/issues/682 -- Default: Remove light source from mese ore by paramat
19:05 ShadowBot      https://github.com/minetest/minetest_game/issues/681 -- Fire: Add eternal flame node for creative and mod use by paramat
19:31 est31          hmmmm,  https://github.com/minetest/minetest/pull/3076
19:32 est31          and second https://github.com/minetest/minetest/pull/3104
19:33 hmmmm          3076 looks good I guess
19:33 hmmmm          I don'
19:33 hmmmm          I don't know much about how all that stuff works
19:33 hmmmm          but it's reasonable
19:35 hmmmm          3104 LGTM
19:35 est31          map->getNodeNoEx returns in the second argument if the load succeeded
19:35 hmmmm          before you merge 3076 can you fix the typos in the comment
19:36 est31          ?
19:37 hmmmm          there is a spurious comma and "collision" is misspelled
19:37 hmmmm          also the comment would be better in // format
19:38 hmmmm          with /**/ style comments you mess up other people trying to comment out a block of code
19:38 est31          lol collusion
19:38 hmmmm          yeah i loled too
19:38 est31          use #if 0
19:38 est31          that can be stacked unlike /**/
19:38 hmmmm          with /**/ commenting out it colors the text green so you can immediately see what's not active
19:39 est31          #if 0 makes it gray for me as well
19:39 est31          (comments are gray for me)
19:39 hmmmm          hmm
19:39 est31          in kate
19:39 hmmmm          comments are green for me
19:39 est31          well, will make it // style
19:41 est31          but the commit message is really good now :)
19:45 est31          #3199
19:45 ShadowBot      https://github.com/minetest/minetest/issues/3199 -- split map_generation_limit into x/y/z components by EUGD
19:45 est31          I see some indentation mistakes for the if clauses
19:48 est31          what do you think, is map_generation_limit_v better?
19:48 est31          bc "volume" can be mistaken with pregeneration
19:50 est31          #3163
19:50 ShadowBot      https://github.com/minetest/minetest/issues/3163 -- fix minetest.afters calling minetest.after by HybridDog
19:52 hmmmm          hmmm
19:53 hmmmm          doesn't matter to me
19:53 paramat        'map generation edge'? i don't like volume
19:54 hmmmm          no idea about 3163
19:54 paramat        'edge' is consistent with other code for a max edge / min edge
19:55 est31          edges always come in two versions
19:55 est31          min edge and max
19:56 est31          but here we only have one
19:56 est31          where the other one is just -1 times it
20:05 paramat        yes, maybe 'vector' 'vertex' 'extent' ...
20:06 est31          extent
20:08 paramat        seems good
20:14 est31          paramat, can you comment?
20:14 paramat        yeah
20:15 paramat        .. i can suggest 'extent' in the thread
20:16 est31          done already
20:20 paramat        added comment of support
21:24 neti_netwalker est31: hello thank you for the merge
21:24 neti_netwalker est31: but your fix is a problem
21:24 est31          why
21:24 neti_netwalker new chunks will not gen
21:25 neti_netwalker i think the retrun "" is needed
21:25 neti_netwalker new chunks stops the server ...
21:25 est31          ahhh
21:25 est31          lemme see how to fix it
21:25 neti_netwalker *g*
21:26 est31          neti_netwalker, what is the error it shows
21:26 neti_netwalker i am currently testing it with a friend
21:26 neti_netwalker the problem is i don't see the error it on his client
21:27 neti_netwalker but i can provide a patch for it if you have one moment
21:27 est31          well there are two ways to patch it
21:27 est31          I have added that error throwing below for a reason
21:27 est31          dont remove it again
21:29 est31          better way to fix it is by adding yet another case
21:29 neti_netwalker https://github.com/minetest/minetest/pull/3206
21:31 neti_netwalker est31: this is working an removes the chance to provide cheesy maps
21:32 neti_netwalker hmm i think a case for key not found in redis is missing in the loadBlock function
21:32 neti_netwalker that can be a better solution
21:32 est31          what about this https://github.com/est31/minetest/commit/6c719aeccc555269ba5cdb524ff0d0f60d9f5fc8
21:32 est31          does it work
21:33 neti_netwalker i can test it tomorrow on my testserver
21:33 neti_netwalker est31: that looks good
21:33 est31          neti_netwalker, do you know the error message it gives server side
21:34 est31          I'm not sure whether it returns nil, or error
21:34 est31          or sth else
21:34 neti_netwalker *g*
21:35 est31          it will print something like "returned invalid reply type 43"
21:35 neti_netwalker i reed the doc it should null or nil
21:37 neti_netwalker currently building
21:37 neti_netwalker i will test it in 1 minute
21:42 neti_netwalker est31: nice! yes https://raw.githubusercontent.com/est31/minetest/6c719aeccc555269ba5cdb524ff0d0f60d9f5fc8/src/database-redis.cpp is working for the new chunk problem
21:42 est31          thanks neti_netwalker merging the commit then
21:42 est31          thanks for testing
21:44 est31          pushed
21:47 neti_netwalker tested with restarting redis and starting redis without waiting ... minetest server hangs but no cheesy map is generated
21:48 neti_netwalker hmm if it shows a stacktrace an cloeses would be better
21:50 neti_netwalker est31: any idear t
21:50 est31          I think it should give you an error, not hang
21:50 neti_netwalker error yes
21:51 neti_netwalker but you cannot do anything anymore
21:51 est31          like?
21:51 neti_netwalker thats ok for a problem with a storage engine but shutting down would be better
21:52 est31          when precisely is it hanging
21:52 neti_netwalker 2015-09-26 23:45:37: ACTION[Main]: Server for gameid="minetest" listening on 0.0.0.0:30000.
21:52 neti_netwalker 23:45:38: WARNING: Undeclared global variable "unified_inventory" accessed at ...in/../mods/technic/technic/machines/register/recipes.lua:54
21:52 neti_netwalker 2015-09-26 23:45:43: ERROR[Emerge-0]: loadBlock: loading block (-9,-4,22) failed: LOADING Redis is loading the dataset in memory
21:52 neti_netwalker 2015-09-26 23:45:43: ERROR[Emerge-0]: ERROR: An unhandled exception occurred: Redis command 'HGET %s %s' errored: LOADING Redis is loading the dataset in memory
21:52 neti_netwalker 2015-09-26 23:46:07: ACTION[Server]: Janakas [192.168.0.42] joins game.
21:52 neti_netwalker 2015-09-26 23:46:07: ACTION[Server]: Janakas joins game. List of players: Janakas
21:52 neti_netwalker ah
21:52 neti_netwalker sorry
21:53 neti_netwalker you can join but you can not do anything on the server
21:54 est31          if the load is successful, does it work then?
21:54 est31          eg if you wait and rejoin
21:54 est31          not restart minetest, only the client
21:58 neti_netwalker if it hangs it hangs it doesn load any wourld chunks anymore
21:58 est31          yeah I see how it happens
21:58 neti_netwalker if it starts succesful all is fine
21:58 est31          hmmmm, what is a good way to crash the server?
21:58 est31          currently there is assert(0)
21:59 hmmmm          huh?
21:59 est31          but that isnt good, because release builds ignore it
21:59 neti_netwalker stacktrace an stop ?
21:59 hmmmm          FATAL_ERROR_IF()
21:59 hmmmm          sanity_check()
21:59 est31          divide through zero?
21:59 est31          END_DEBUG_EXCEPTION_HANDLER
21:59 hmmmm          that wouldn't be thrown as a C++ exception
21:59 est31          emerge.cpp line 573
22:00 est31          (and debug.h line 152)
22:00 hmmmm          what are you trying to accomplish exactly
22:00 est31          concerning the bug neti_netwalker has
22:00 est31          that if redis is still loading, the server should crash
22:01 hmmmm          so then use FATAL_ERROR_IF()
22:01 hmmmm          i don't get the problem
22:01 est31          but right now, it brings the server into a state where the emergethread has ended
22:01 est31          and super-lag
22:02 hmmmm          well
22:02 hmmmm          i don't quite know what to tell you
22:02 hmmmm          FATAL_ERROR_IF() should abort in every thread
22:03 hmmmm          do you want the entire server to crash, or do you want the emergethread to keep on running?
22:03 est31          where is fatal_error_fn defined
22:03 hmmmm          debug.h..
22:04 est31          I want to fix the regression
22:04 est31          added by zeno, as he removed the assert macro
22:04 est31          or sb else, but I think it was him who removed that macro
22:04 est31          bc it is meant to crash the entire server
22:05 T4im           do you mean segfaulting when saying crash? O_o
22:05 est31          sort of
22:05 est31          rust would call it panic
22:05 est31          we dont know whether the error is of a kind where a simple restart of the according thread is enough
22:05 T4im           I've seen: * ( volatile int * ) 0 = 0x12345678;
22:06 T4im           for dev purposes though :o
22:27 est31          hmmmm, -> https://github.com/est31/minetest/commit/e7282816fffc433861e7e83b0bf1de701d3ea2d1
22:28 hmmmm          oh no don't do that
22:28 hmmmm          that's bad
22:29 hmmmm          er wait a minute
22:29 hmmmm          oh i see that's fixing an old commit
22:29 est31          the assert(0) is totally pointless
22:29 est31          its only called in release mode
22:29 hmmmm          yeah i know
22:29 hmmmm          alright yeah that should be alright then
22:30 est31          okay pushing
22:30 hmmmm          wow I must've never noticed that because i always run in debug mode
22:37 VanessaE       can someone give #3166 some attention before it becomes unusable?
22:37 ShadowBot      https://github.com/minetest/minetest/issues/3166 -- Send to clients only changed node metadata instad of whole mapblock by RealBadAngel
22:37 VanessaE       (including the bug est31 says needs fixed)
22:38 paramat        i guess we need RBA
22:38 est31          yea
22:38 est31          its said that RBA has no internet
22:39 VanessaE       is THAT what's going on?
22:39 est31          dunno just heard it from a good friend of RBA
22:40 VanessaE       hm, ok
22:40 paramat        will merge #3203 once rebased
22:40 ShadowBot      https://github.com/minetest/minetest/issues/3203 -- Decorations: Remove error message 'chunksize not divisable by sidelen' by paramat
22:46 paramat        wow almost at commit 5000
22:47 est31          round number
22:49 est31          34% by c55, top 10 without c55 provide 32.6%
22:57 paramat        now merging 3203
23:01 paramat        complete