Time  Nick         Message
03:28 hmmmm        celeron55, can you explain to me what exactly happens when a block is marked as underground?
03:30 hmmmm        all i can actually see is that it's used in Map::isNodeUnderground() and it's marked as underground in a diagnostic string description of a block
03:30 hmmmm        like, other than that, is it totally useless...?
03:47 hmmmm        I don't understand why you need to emerge the neighbors of what's being generated either
03:55 hmmmm        why is m_map->emergeBlock(obj->m_static_block) called without allow_generate = false in ServerEnvironment::removeRemovedObjects()?  surely an object won't be in a nonexistent block...?  or not?  I'm thinking of the case where someone has noclip on, and they go into some far off area and the block hasn't been generated yet - you drop an object, such as a pickaxe, it falls, and it gets removed by something
03:56 hmmmm        now i can get that, but this is being called in the main server thread, no?  not the emerge thread where it's supposed to be, so the main thread will be held up by this douche who dropped his pickaxe in the wrong spot and it's being removed
03:57 hmmmm        I personally think it should be changed to createBlock()
03:57 hmmmm        do you forsee any undesirable side effects from this?
04:02 hmmmm        i can't really continue until this is resolved somehow
04:47 VanessaE     darkrose, thexyz:  Can someone push this to upstream?  It does look a bit better than what we have now.
04:47 VanessaE     https://github.com/jordan4ibanez/minetest/commit/37d6c3a2f19854caf6dfeb352cb9f6c50bb52757
04:50 * VanessaE   wanders off to bed.
04:58 OldCoder     VanessaE, good night
07:01 celeron55    hmmmm: i am not exactly sure if it is used for something else, but i think this is the main thing:
07:01 celeron55    can you think of how the game can figure out from where the sunlight comes from?
07:03 celeron55    it does it so that if a block is not underground and there is no generated stuff above it, it assumes sunlight is coming from the top of it; otherwise it is not
07:03 celeron55    that may be a bit obscured or not used currently because it generates so large areas at a time; i am not exactly sure
07:04 celeron55    hmmmm: you need to get the neighbors because parts of trees and things like that need to be able to be placed on them
07:04 hmmmm        ?!?
07:05 hmmmm        are you sure about that?  from what i'm reading, you just scan downward until you find a node which has sunlight_propogates = false
07:05 celeron55    i can't remember that much about what the code currently does; i just know the ideas behind the stuff
07:08 hmmmm        indeed, it has nothing to do with isNodeUnderground
07:08 celeron55    there are probably three or more sunlight-related algorithms for different situations (node placed, mapblock generated, miscellaneous area generated; or something alike)
07:11 hmmmm        if it's not too slow, i'd eventually like to have the sun at an angle (maybe like 8 directions total?) instead of directly above
07:11 hmmmm        but, later
07:12 celeron55    it's probably possible; just statically offset the lookup of "above" every 8 nodes in the y direction
07:13 celeron55    or whatever
07:13 celeron55    ah, you meant to calculate 8 different lighting values?
07:13 hmmmm        no
07:13 hmmmm        what you said first was what i had in mind
07:13 celeron55    ok; it's worth a try
07:14 hmmmm        but i mean the two ideas could be combined
07:14 hmmmm        any thoughts on this?  [22:47] <hmmmm> I don't understand why you need to emerge the neighbors of what's being generated either
07:14 celeron55    i answered that already
07:15 hmmmm        oh
07:15 hmmmm        didn't see that
07:15 hmmmm        if the tree placement and what not gets put in lua, you know SOMEBODY is going to make a tree larger than 16 blocks in some direction though
07:15 hmmmm        which might break things
07:16 celeron55    minecraft has it's own way of making that happen: it generates only a single "mapblock" at a time, but generates caves and trees and things like that in a larger area around it to make sure everything that might end up in it gets put in it
07:16 celeron55    and discards the leftovers
07:17 hmmmm        uber inefficient
07:17 celeron55    maybe; maybe not; depends on the details
07:17 hmmmm        the cave generation algorithm i still don't entirely understand... the bottom line is that it's probably really slow and it's probably going to stay that way for some time
07:18 hmmmm        i like the current caves anyway
07:19 celeron55    optimizing generation for speed only is not wise in the long run; the priority should be on the output of it
07:21 celeron55    people don't care if they get 10000000000 boring nodes per second; they care if they get whatever amount of awesome stuff at almost whatever speed
07:21 hmmmm        it's awesome to have both, though. :)
07:22 celeron55    but if it's either or, then it must be the "not speed" one
07:22 hmmmm        i'll agree with that
07:23 hmmmm        now what do you think about emergeBlock being called with allow_generate = false?
07:23 hmmmm        there are only two instances of that being used, both in environment.cpp
07:23 hmmmm        i'd like to really get rid of any non-emerge-thread map generations for various reasons
07:24 hmmmm        er, allow_generate = true*
07:24 celeron55    the same thing is done by the map generator/loader thread too, and it is invoked in more places afaik
07:24 celeron55    it doesn't use emergeBlock
07:25 celeron55    umm...
07:26 hmmmm        the thing is i don't know where i'd plop the mapgen object for emergeBlock generation calls
07:26 hmmmm        i intended for each emerge thread to have one mapgen object
07:27 celeron55    i see more than two instances of it being used with and with no allow_generate in environment.cpp
07:28 celeron55    and what is the problem with allow_generate=false compared to ture?
07:28 celeron55    true*
07:28 hmmmm        lines 1275, 1518, 1768
07:28 hmmmm        yea i missed one
07:28 hmmmm        and there's one with it explicitly true in server.cpp
07:29 hmmmm        ....which is in findSpawnPos
07:29 celeron55    ^ answer the question
07:29 hmmmm        [02:26] <hmmmm> the thing is i don't know where i'd plop the mapgen object for emergeBlock generation calls
07:29 celeron55    so it applies to false and true?
07:29 hmmmm        just true, because that's the only case where it'd be using the mapgen
07:31 celeron55    i think all cases could work so that if the block isn't generated, they'd ask for the generation and come back polling for it later until it is
07:31 hmmmm        ekh
07:31 hmmmm        i wonder how it'd carry out the polling
07:32 hmmmm        for example in findSpawnPos the only reason it uses emergeBlock is to ensure that the spawn position actually exists
07:32 celeron55    i think all of the places can just cancel what they are doing and automatically come back to try again, as they currently are
07:32 hmmmm        but with the new mapgen setup i'd just add that block to the emerge queue
07:32 celeron55    findSpawnPos can't do that as-is, yes
07:33 celeron55    ehm... and ServerEnvironment::addActiveObjectAsStatic is some kind of a mapgen-with-mobs helper
07:33 celeron55    ...i guess
07:33 hmmmm        and the other two are removeRemovedObjetcts and deactivateFarObjects apparently
07:33 celeron55    but for example ServerEnvironment::removeRemovedObjects can just queue, cancel and come back later
07:33 hmmmm        right
07:33 hmmmm        now why can't i just use createBlock for that?
07:34 hmmmm        that particular one
07:34 celeron55    maybe
07:35 hmmmm        i was saying before that it tries to emerge m_static_block, which is "the block frmo which the object was loaded from, and in which a copy of the static data resides"
07:35 celeron55    i don't see why that couldn't be done
07:35 hmmmm        so if it didn't exist, then...
07:35 hmmmm        basically
07:35 hmmmm        is there something like createBlock except it'll fail if it's not in memory and not on disk?
07:36 celeron55    emergeBlock with allow_generate=false?
07:36 celeron55    dunno
07:36 hmmmm        oh that's true
07:36 hmmmm        hah
07:37 hmmmm        yeah, so i'll just add ", false" there, and that should do the trick
07:37 hmmmm        and there are the two remaining which will poll like you said
07:38 celeron55    it could be useful to have a working system for creating blocks without generating the node data, and have it work properly (leaving everything else in it intact) when it finally is generated
07:39 celeron55    including non-generator node data
07:39 celeron55    it probably works already though
07:39 celeron55    because tree-like thing scurrently need that
07:39 celeron55    s<<1
07:46 hmmmm        will do
07:46 hmmmm        but i'll have to do it in a different commit
07:46 hmmmm        this is already ginormous
07:47 hmmmm        'night
14:37 RealBadAngel hi
15:37 PilzAdam     Im currently developing a farming mod for upstream
15:38 PilzAdam     should there be some functions like register_plant() or should I define every growing step by hand?
15:39 celeron55    register_plant() sounds like the way to go
15:39 celeron55    make it public so other lazy mods can use it
15:40 PilzAdam     i already have some test code for this: https://github.com/PilzAdam/minetest_game/blob/farming/mods/farming/init.lua
15:48 RealBadAngel Hi celeron55
15:49 RealBadAngel what do you think about my modifications to the formspec?
17:16 hmmmm        i was looking at addActiveObjectAsStatic again, and i've gotta wonder
17:16 hmmmm        why must the emerge succeed?
17:17 hmmmm        when would it possibly be valid for a static active object to be added to a nonexistant block?
18:03 hmmmm        would you say that a definition of all the available biomes belongs in ServerEnvironment?
18:04 PilzAdam     OldCoder, https://github.com/downloads/PilzAdam/minetest_game/farming.zip
18:04 OldCoder     Hi
18:04 OldCoder     Thank you!
18:04 OldCoder     Working
18:07 PilzAdam     oops, that was the wrong channel
18:07 OldCoder     I got it anyway
18:09 celeron55    hmmmm: is addActiveObjectAsStatic even called anywhere
18:09 celeron55    appears it is not
18:09 hmmmm        oh
18:09 celeron55    throw it in the bin
18:10 hmmmm        that's true :)!
18:10 hmmmm        well wait, what does it actually do?  what's a "static" object?
18:11 celeron55    a static object is an active object that is stored in the non-active state... the state in which they are on disk, and the state in which they are when they are far away from players, to not have the full infinite world active at the same time
18:12 celeron55    it is basically a serialized AO
18:12 celeron55    stored in the mapblock, not environment
18:13 hmmmm        hmm, i can't think of any situations where that'd be helpful
18:13 hmmmm        what were you thinking of at the time when you wrote it?
18:14 celeron55    i... think it was used for at least testing the generation of mobs in the mapgen
18:14 celeron55    not sure
18:14 celeron55    but it's kind of useless for that too
18:14 hmmmm        i don't think mobs will ever be a part of the engine
18:14 hmmmm        again
18:14 hmmmm        alright, to the trash it goes
18:15 PilzAdam     what are the params active_object_count and *_wider that are passed to abms?
18:18 hmmmm        active_object_count is the number of objects in that block
18:18 hmmmm        active_object_count_wider is the total number of objects of that and all the neighbors
18:19 PilzAdam     thx
18:19 PilzAdam     there should be a hint in lua-api.txt
18:20 hmmmm        i think someone was supposed to make a lua api documentation wiki
18:20 hmmmm        all this crap should be added there
18:21 celeron55    well... cornernote had it, and he just quit
18:21 hmmmm        pft
18:21 celeron55    the data has been copied, but we don't have the platforms he used
18:21 celeron55    http://api.minetest.net/ currently serves a static copy from thexyz's servers
18:24 celeron55    would be quite convenient to just add it now here, but well...  http://api.minetest.net/type/view/name/abm.html
18:25 celeron55    thexyz: by the way, can you make that one available as a .tar.gz?
18:31 OldCoder     The API site is back?
18:35 celeron55    it's in fact back by cornernote too, but he filled it with the vegan crap he started flooding everywhere 8)
18:36 hmmmm        why did he quit again?
18:36 PilzAdam     people sometimes go crazy
18:36 hmmmm        it seems like it's a little better than it was, but people leave all the time
18:36 hmmmm        like too often
18:36 celeron55    he was burned out and trolled out and got finally completely pissed off when thexyz kicked him
18:37 hmmmm        this is basically the only minecraft clone project that's worth a shit
18:37 hmmmm        i don't know where else he'd go to
18:37 PilzAdam     RL?
18:38 hmmmm        RL is overrated
18:38 PilzAdam     +1
18:38 celeron55    i hope he first takes a break and then re-figures out his time management so he can possibly focus on something without going crazy
18:39 hmmmm        i actually have a huge project due on the 17th for school but i'm procrastinating with minetest
18:39 celeron55    i seriously hope people don't screw up their lives because of minetest
18:39 celeron55    it's not worth it 8D
18:40 PilzAdam     btw: if someone is interested in the farming mod I currently develop for upstream: https://github.com/PilzAdam/minetest_game/tree/farming/mods/farming
18:40 hmmmm        nah if it weren't this it'd be some other video game
18:57 OldCoder     celeron55, MT probably is not the leading cause of cancer :p
18:57 OldCoder     If anything it is a stress reducer
18:58 OldCoder     Except for those who must keep it working O_o
18:58 OldCoder     :-)
18:58 OldCoder     Calinou, tell me true // if you // can connect
18:58 Calinou      nope
18:58 OldCoder     PilzAdam, I will update farmingminus in a bit
18:58 OldCoder     Calinou, Hmm?!
18:58 OldCoder     I will connect myself, then
18:59 Calinou      europe.minetest.org port 30010
18:59 OldCoder     No!
18:59 OldCoder     There is only 30000
18:59 OldCoder     That is a separate VPS
18:59 OldCoder     No connect to the first dozen or so worlds
18:59 OldCoder     The entire point is that that VPS is in Europe
18:59 OldCoder     For our Europe friends
19:00 hmmmm        ermm let's not get too carried away with the off-topic conversation, okay?
19:00 Calinou      ok
19:00 OldCoder     I may have other Europe VPS lined up
19:00 celeron55    OldCoder: you're on #minetest-dev
19:00 OldCoder     celeron55, yes?
19:00 OldCoder     If that is not core dev
19:00 OldCoder     I was simply answering his question
19:00 celeron55    that discussion started in #minetest
19:00 OldCoder     Hmm?
19:00 OldCoder     I am confused; one moment
19:01 OldCoder     I have lost track of things celeron55 and am needed elsewhere regardless. If the load is slow on with the show it is regrettable.
19:01 * OldCoder   will return
19:02 celeron55    to put it more simply: talk about your servers on #minetest, because this is simply not meant for those, especially as there is other discussion in here tonight
19:02 OldCoder     <PilzAdam> OldCoder, https://github.com/downloads/PilzAdam/minetest_game/farming.zip
19:02 OldCoder     celeron55, I am very very literal
19:02 OldCoder     I regret this but it is true
19:03 OldCoder     If somebody addresses me I do answer
19:03 OldCoder     Your own remarks are not entirely on-topic as well; but of course as the head it is your privilege
19:03 celeron55    but calinou asked you on #minetest, not here
19:03 OldCoder     That quote is from right here
19:03 OldCoder     One hour ago exactly
19:04 OldCoder     I will add, Sir, that some of my server issues are core dev issues as they are about testing core chanegs
19:04 OldCoder     *changes
19:04 celeron55    ummmmmm
19:04 OldCoder     Farming is not the case for that
19:05 OldCoder     But for farming I was answering here
19:05 OldCoder     A question that was asked here
19:05 hmmmm        what the hell
19:05 OldCoder     Hmpph.
19:05 OldCoder
19:05 celeron55    i don't understand; you haven't replied anything to that line of PilzAdam
19:05 hmmmm        just don't talk about it any more
19:05 OldCoder     I am confused; which line?
19:05 celeron55    ah, you did, but immediately after it
19:05 celeron55    i didn't mean those when i asked you to move to #minetest
19:05 celeron55    i meant the stuff you talked about with calinou
19:06 OldCoder     Looking up, that appears to be an honest mistake. I am in about 30 channels ATM.
19:06 OldCoder     It is not as you characterized it. I have no intent to be argumentative.
19:06 OldCoder     More than 30 it appears. I have no idea how you more experienced IRC people do this.
19:07 OldCoder     celeron55, mistakes are going to be made. I feel there is no need to emphasize the issue.
19:07 PilzAdam     OldCoder, just stop talking in -dev
19:07 celeron55    OldCoder: i am not trying to emphasize it, but rather i haven't yet really seen you understand what i said
19:07 OldCoder     As you wish, Mr. Adam. celeron55 please ban me from this channel.
19:07 celeron55    apparently you did understand it then
19:08 OldCoder     Will you do as I ask?
19:08 OldCoder     Ban now, please
19:08 celeron55    wtf
19:08 OldCoder     There is no need for a kick
19:08 celeron55    i have no reason to ban you
19:08 OldCoder     I am exitinhg
19:08 PilzAdam     cornernote?
19:08 OldCoder     Hmm?
19:08 OldCoder     Nothing to do with him
19:08 OldCoder     Though I see something of his point of view
19:08 celeron55    this doesn't make any sense
19:08 OldCoder     <PilzAdam> OldCoder, just stop talking in -dev
19:08 OldCoder     Honestly, Pilz, was that necessary?
19:09 celeron55    http://weknowmemes.com/wp-content/uploads/2012/04/what-the-fuck-man.jpg
19:09 celeron55    i'm actually laughing here, this has become hilarious
19:10 celeron55    i'm fucking glad i have zero money or anything involved in this :------D
19:12 celeron55    oh, anyway; back to figuring whether i can whip up a quick fix for the bad bandwidth throttling
19:13 celeron55    hmmmm: i assume you are making some kind of both-hands-and-feet facepalms in there and thus not being able to comment? 8)
19:14 PilzAdam     ...
19:14 hmmmm        i'm actually watching an episode of the 'minecraft world tour'
19:14 * PilzAdam   doesnt understand what is going on with people
19:14 hmmmm        what's wrong with the bandwith throttling?  i didn't even know that there was bandwith throttling
19:15 celeron55    there has to be if you have the intention of stuffing up miscellaneous blobs of data through UDP
19:15 celeron55    as in moving media from the server to the client
19:16 celeron55    otherwise you'll just cause most of the packets to be dropped and you've basically failed networking
19:18 celeron55    in largeish servers it would make much more sense to use TCP for those, but it'll hurt small-scale hosting and testing and... ehm... well, this is an another case of "it's just not that simple"
19:19 hmmmm        the original Starcraft has a udp transfer system much like this
19:19 PilzAdam     would it be possible (in theory) to add a setting in config to switch UPD and IP?
19:19 hmmmm        i remember transferring maps that were > 500KB being horribly slow
19:19 hmmmm        pilzadam, sure... if you feel like rewriting the socket class
19:20 hmmmm        which isn't actually too bad
19:27 celeron55    the correct name for what i talked about appears to be congestion control
19:27 PilzAdam     so we have a name for the topic, now code it ;)
19:28 celeron55    it's really a full-blown research field
19:28 hmmmm        of course
19:28 celeron55    which means i'll just whip up something from the top of my head!
19:29 hmmmm        and if you want congestion control, TCP is where it's at
19:29 hmmmm        seriously - the OS already has all sorts of great algorithms for congestion control and they're really good
19:29 hmmmm        the reason to use UDP would be low latency
19:29 celeron55    oh well; we could use udp and tcp on the same port
19:29 hmmmm        which i can see latency being really bad for FPSes and and racing games or whatever
19:30 hmmmm        but this game... not as much
19:30 hmmmm        right, and like you and pilzadam said, you can easily have both.  i suggest that UDP be used for the regular game commands and a TCP connection is made for bulk transfers
19:31 hmmmm        i think a lot of games do that as well
19:31 PilzAdam     so you have to tell all the n00bs that they have to forward both, TCP and UDP
19:31 PilzAdam     *port forward
19:32 celeron55    i find it somewhat useful, because there are some things that are updated very often and that are not that important to reliably go through (positions of entities and players, or whatever), and if you do that over tcp, you can get stupid hiccups because they didn't get through and tcp ends up re-sending them
19:33 celeron55    packet loss is a real thing on the internet, in any case
19:34 hmmmm        unrelated but
19:34 hmmmm        what's up with the naming scheme for things that are in gamedef?
19:35 celeron55    how so?
19:35 hmmmm        what do you mean by "SomethingDef"
19:35 hmmmm        'definition'?
19:35 celeron55    yes
19:36 celeron55    it's a very common abbreviation almost everywhere in the source
19:37 hmmmm        ahh yes, because whatever derived class that uses that interface is "defined" by that interface
19:37 hmmmm        ?
19:37 celeron55    well... GameDef basically defines what the game is
19:38 celeron55    or, well, everything apart from the mechanisms that the engine does
19:38 celeron55    then there are some odd-but-per-game things like the sound and event thingies on the client
19:39 celeron55    because the pointers are so conveniently transferred in it
19:39 PilzAdam     has anyone anything against this? https://github.com/celeron55/minetest_game/pull/26
19:40 hmmmm        nah, it's fine
19:40 PilzAdam     (except that the commit is gramaticaly nonsense)
19:40 hmmmm        consistency is better than grammar i say
19:40 PilzAdam     (there is a missing "of")
19:40 PilzAdam     k
19:42 celeron55    there are two things that PilzAdam could have meant with "gramaticaly nonsense" 8)
19:43 celeron55    and i think hmmmm understood it the other way than PilzAdam meant
19:43 hmmmm        right, bucket being capitalized in that context, and the commit message
19:43 hmmmm        i realized what he ment after i pressed the enter key
19:43 hmmmm        doesn't matter though
19:43 PilzAdam     well, actualy i mean there has to be an "of" in front of bucket (in the commit message)
19:46 celeron55    i think i'll try how much of effort it is to make minetest set up a TCP connection in addition to the UDP one (well, technically it isn't a connection at UDP level, but minetest makes it be)
19:47 celeron55    this has to be made transparent to the upper layers, because i am lazy
19:47 PilzAdam     in my farming mod I use textures were I cant remember were i got them from; but they are WTFPL iirc
19:47 celeron55    happily there is already enough abstraction to do it transparently :-D
19:47 hmmmm        make a base ISocket class, have UDPSocket a derived class from that, then make another derived class for TCP that's literally just a wrapper for the standard TCP functions
19:47 PilzAdam     can I use them if i want it to be upstream?
19:47 hmmmm        should take like an hour tops
19:47 celeron55    ehm
19:48 celeron55    hmmmm: there is no benefit from making a common interface to them though
19:49 celeron55    they are both handled specially
19:49 hmmmm        yeah i guess you're right
19:49 celeron55    and because minetest is built to use UDP, there is no thing called "connection" at the socket level; it is at the upper level from it
19:50 celeron55    but whatever; it'll turn out somehow
19:50 hmmmm        oh ew, i need to add epoll() and kqueue() for WaitData
19:51 hmmmm        select is ancient technology from the 1970s
19:51 celeron55    you need to make it work on windows in addition to posixy things
19:52 hmmmm        simple, fall back to select where no alternative is available
19:53 celeron55    oh god this Connection class had these funny naming things...
19:54 celeron55    i don't want to touch this :D
19:56 celeron55    there's two generations of interfaces put in one class, and the naming of some public thread-safe things as compared to some internal things differ only in capitalization of first letter, and the public interface has the different convention than minetest mostly has...
20:00 hmmmm        hmmmmmm...?  http://codepad.org/GUbhC4mI   <--- is this not really needed any longer?
20:01 hmmmm        (well, i don't intend on removing it, just wondering if in theory that wouldn't be necessary)
20:02 celeron55    i don't know; i suspect the problem still exists
20:03 hmmmm        if that still exists but there are no static objects being added, i think that'd classify as map corruption
20:06 hmmmm        hrmm, how am i going to poll to see if the block is created or not in deactivateFarObjects, exactly?
20:06 hmmmm        i am _not_ blocking in the main server thread - hell no.
20:08 hmmmm        looking at what calls it, i guess it's alright to block in ~ServerEnvironment() since it's shutting down, but in ServerEnvironment::step, i think i can just return with no ill consequences
20:09 hmmmm        i would rather let the caller know if i had actually completed deactivating far objects though so i'll have it return true or false if it did
20:09 hmmmm        and of course if force_delete = true, it'll always succeed (return true)
20:11 hmmmm        it seems like i'm going to have to make that "emerge blank block" function now
20:13 hmmmm        okay nevermind any of that crap
20:15 hmmmm        here's what i'm going to do:  change emergeThread to:  get from memory, load from disk, and if not existing, create a blank MapBlock with a "needs generating" flag set
20:16 hmmmm        then if the emergeThread loads a mapblock from disk with "needs generating" set, it'll send it to the mapgen
20:48 celeron55    gotta hate these stupid header dependencies
20:48 celeron55    modifying connection.h causes like everything to be rebuilt
20:50 celeron55    well, 17 files out of almost 100... i guess that's reasonable
20:54 hmmmm        speaking of modifying the map format... what do you say to me adding a noiseparams field along with a seed and other metadata
20:54 hmmmm        or would you like me to include those things in the same text file where the seed is
20:57 celeron55    noiseparams field... to what?
20:57 hmmmm        the map database itself
20:57 hmmmm        instead of the map database containing only blocks, why can't it contain metadata as well?
20:58 hmmmm        oh that's what you're asking... noiseparams fields of the terrain, biomes, humidity, temperature, etc.
20:58 celeron55    because it's there to only hold blocks; there is for example the leveldb implementation being developed somewhere
20:58 hmmmm        if this mapgen is going to be modifiable i need to store it so that it stays consistent between people changing those parameters
20:59 hmmmm        if someone modifies their mapgen parameters, that is
21:00 celeron55    (i wonder how hell of a job integrating this https://github.com/minetest/minetest/commits/db_backends to your stuff might be)
21:01 hmmmm        that?  not a problem at all
21:02 hmmmm        so it's settled, i'll store the metadata in map_meta.txt
21:05 hmmmm        and for the mapblock thing, i don't even need to add a flag, i can just use a so-called 'dummy' block
21:13 hmmmm        well... THERE was a bug waiting to happen.  if you tried to serialize a dummy block, it's segfault
21:14 hmmmm        MapBlock::serialize, it tries to copy over the mapnodes without checking if data == NULL first
21:14 hmmmm        so i'm going to add something like   if (isDummy()) flags |= 0x10;  after all
23:07 celeron55    lol i should sleep but i can't go now
23:07 celeron55    SOCKETS, SOCKETS EVERYWHERE
23:08 celeron55    lol minetest serves HTTP now :-----D