Time  Nick          Message
07:44 fruitsnack    hey guys, why the engine doesn't respect param1 probability of schematic type decoration nodes? It's not implemented or I'm doing something wrong
07:45 fruitsnack    all nodes are always placed unless the probability is 00
14:21 lllI1I        are there plans for / would anyone be opposed to the ability to keep some node and item definitions "in reserve" so that not all node and itemdefs are sent at start, and some might be sent on a "need-to-know" basis to prevent e.g. spoilers maybe?
14:24 nepugia       Registering items after start makes sense certainly to me
14:25 Krock         interests: yes, plans: no.
14:29 lllI1I        how does one get to brainstorming something like that
14:29 lllI1I        I'm guessing (99% sure) it'd need a protocol update
14:30 lllI1I        and probably won't get in until minetest 6.0
14:31 lllI1I        though maybe it just so happens that the protocol is vague enough about this point that it just so happens that every implementation would support it
14:31 lllI1I        but I doubt that
14:35 lllI1I        could also save on initial connection time if there are loads of mods on the server
14:36 lllI1I        probably would need to classify definitions as "likely" or "unlikely" to appear commonly
14:36 lllI1I        e.g. good chance "default:stone" or "default:dirt_with_grass" will be encountered
14:38 lllI1I        less likely "supertools:megadiamond_pingspoofer" will show up, so less of a problem if there's a tiny "loading texture" texture for a fraction of a second when it gets found
14:39 lllI1I        guessing default would be assume it's common, more rare stuff (thus being sent later) can be flagged as such
14:47 nepugia       I'd just mark stuff as to be downloaded, or initiate downloads specifically inside mods
14:49 nepugia       This would probably work best in tandem with clientside mods
14:49 nepugia       (i.e some code might be there already, but meshes or textures only get send over once they become relevant)
14:59 Krock         lllI1I: it would work without a proto bump. Older clients would then just not receive the new content
15:00 nepugia       Krock: that doesn't sound like "working" to me
15:01 nepugia       in any case, some clients simply not supporting functionality that is expected to work seams to be undesirable, so to me a protocol bump makes sense
15:01 Krock         it works for clients that support it
15:01 Krock         don't expect forwards compat in old clients
15:01 Krock         how even
15:02 nepugia       you *can't* make it compatible with old clients
15:02 nepugia       which is exactly why you should bump the protocol version...
15:03 Krock         oh. I wasn't clear there. It requires a proto bump, but not an entire backwards incompatible one (i.e. not v6.0)
15:04 Krock         protocol version bump *
15:04 Krock         older clients would then see unknown items and unknown nodes
15:05 Krock         or such with outdated properties
15:06 nepugia       Krock: dynamic textures when? :(
15:07 Krock         how dynamic?
15:07 nepugia       allow me to place *any* texture on my block, no matter when i made it
15:07 nepugia       on any side i want :)
15:07 Krock         nodemeta-specified texturing would skip PLENTY of drawing optimizations
15:08 Krock         since metadata is never read in the render
15:08 nepugia       i never sais nodemeta-specified
15:08 Krock         how else?
15:08 nepugia       but justifying not implementing something with how the design is now is flawed logic
15:08 nepugia       of course it doesn't do a different usecase perfectly to the one it is ment to do now
15:08 nepugia       it doesn
15:09 nepugia       't matter how it is specified really
15:09 nepugia       if you want you could do it based on adjacent blocks too
15:09 nepugia       point is that that should be supported, even if it would be in the nodemeta
15:09 Krock         it matters because somehow this data has to be stored in the map database
15:09 nepugia       same for dynamically adjusting voluminocity of objects
15:09 nepugia       sure, but how it gets stored is of no consequence to me :)
15:09 nepugia       and matters little at runtime
15:09 Krock         apparently.
15:11 nepugia       in any case, the node swap thing seems like a bad hack to me
15:11 nepugia       don't want to register 16 versions of a block just for a texture or voluminocity change
15:13 Krock         how so? it's just a loop
15:13 nepugia       it's putting all my code 16 times in ram for no reason at all
15:13 nepugia       say, what if my block should do like a rainbow animation, rgb like
15:14 nepugia       that would need 256* 256 *256 copies of the code to do that
15:14 Krock         if you want animations, don't use swap_node. Use tile animations
15:14 nepugia       if i want arbitrary ones i have to use swap
15:15 nepugia       in any case, swap_node sucks :P, doesn't matter to me whether you like the api interface or not, programtically there should be no difference between a lit furnace and a non-lit furnace, the code is pretty much the same for both
15:15 nepugia       just that it may play some animations, or change the ligting
16:05 lllI1I        "API interface..."
16:06 lllI1I        unless you mean an interface to get the API or through which the API goes, etc
16:07 lllI1I        like an ATM machine would produce ATMs
16:10 nepugia       bah, my mistake
16:10 nepugia       accept my cincerest apology for this grammatical error
16:49 MinetestBot   02[git] 04acmgit -> 03minetest/minetest_game: Remove stack_max from bucket:bucket_empty 1383fb6fe https://git.io/fjXGv (152019-07-13T16:47:45Z)
20:16 DS-minetest   can I compare a macro with a char in a preprocessor #if?
20:16 DS-minetest   (I want code only be executed if DIR_DELIM is not "/")
20:19 Krock         only if that's a macro too
20:19 Krock         integer comparison works, so why not string too? *shrug*
20:19 DS-minetest   and if the "/" is a constant?
20:20 Krock         it has to be known to the compiler before the actual code is parsed
20:20 DS-minetest   ah, I see
20:20 Krock         it's not a macro? weird, it should be.
20:26 DS-minetest   meh, that doesn't work:
20:26 DS-minetest   #define SLASH "/"
20:26 DS-minetest   #if DIR_DELIM != SLASH
20:26 DS-minetest   ...
20:26 DS-minetest   #endif
20:26 DS-minetest   #undef SLASH
20:26 lllI1I        that doesn't sound like a job for the prepocessor
20:26 DS-minetest   (I hope, it wasn't too long for irc)
20:27 DS-minetest   should I use #ifdef _WIN32?
20:27 lllI1I        is it a compiletime check or a runtime check?
20:27 DS-minetest   a compiletime check I guess
20:28 DS-minetest   it should compile different code based on whether DIR_DELIM is "/" or not
20:28 lllI1I        I guess it would help if we knew what it was you were trying to do
20:29 DS-minetest   I want to send a path via network to another computer, which might use a different DIR_DELIM
20:29 DS-minetest   so, I always send with "/"s
20:30 * DS-minetest will just compare at runtime and hope that the compiler will optimize
20:30 lllI1I        yeah I wouldn't worry too much about optimizing until and unless it becomes an issue
20:37 * DS-minetest ended up using #ifdef _WIN32 because of warnings and errors
21:06 lllI1I        where is the best description of the network protocol
21:07 DS-minetest   in networkprotocol.h?
21:07 DS-minetest   clientface.h might be interesting, too
21:08 lllI1I        yeah I looked at that first, wanted to know the more lower level to get a better understanding and then forgot about it
21:08 lllI1I        thanks
21:11 Lone-Star     say im creating a fence node, with the fixed post being wood and connected nodes to be steel. how would i define the textures of the connecting nodes?
21:12 DS-minetest   you can't
21:12 DS-minetest   only the boxes change
21:12 DS-minetest   the tiles are applied like to every other nodebox
21:44 lllI1I        TYPE_ORIGINAL seems under documented
21:45 lllI1I        I'm seeing a lot of those packets in wireshark and no idea what they are meant to be
21:46 lllI1I        wait I might be dumb
21:48 lllI1I        yeah 00 01 is server's peer id, not channel 00, type 01
21:56 lllI1I        why does my client appear to send TOSERVER_INIT before receiving a peer id, only to resend it with the peer id after?
21:59 lllI1I        sorry if there's a better channel to ask these questions in
21:59 epoch_        lllI1I: what are you wanting to do with the network protocol?
22:03 DS-minetest   does the client even know its peer_id?
22:04 DS-minetest   (probably yes, sorry, I wish there was a way to add [off] afterwards)
22:06 lllI1I        at this point, it doesn't know its peer id
22:06 lllI1I        epoch_: thinking of making special "NPC" clients for my server
22:08 lllI1I        not just using a mod because I want to keep CPU cycles low on the server, so in case the "NPC"s are doing a lot of calculations it won't slow the server down
22:09 epoch_        that'd be neat.
22:09 epoch_        like IRC bots
22:09 lllI1I        and I can add more NPC's with more VPS's at no computational cost to the server
22:10 lllI1I        well the reason I used quotes is because I'm thinking of offloading quite a bit to these clients
22:10 DS-minetest   haha, multithreading in minetest
22:10 lllI1I        so that they don't have to even be loaded on the server's memory when players aren't present in the area
22:10 DS-minetest   =outsourcing
22:11 epoch_        I was looking into the protocol a while ago because I was wanting a little program that'd tell me what version the server was running
22:11 lllI1I        yes it works
22:11 epoch_        so I could point the little program at any minetest server and the proper version would get launched.
22:11 lllI1I        really the only thing that's tripping me up is 1 undocumented byte
22:11 lllI1I        actually 2 I think
22:15 epoch_        are you working on a wireshark dissector for minetest packets?
22:18 DS-minetest   (I'll leave this here: https://github.com/minetest/minetest/blob/master/util/wireshark/minetest.lua)
22:19 epoch_        oh boy.
22:19 lllI1I        thought about writing a wireshark plugin for it, but I didn't understand the protocol
22:19 lllI1I        but if I learn the protocol I wouldn't need the plugin shfkksv
22:32 lllI1I        plugin works, thanks DS-minetest
22:34 lllI1I        oh I see now
22:35 MinetestBot   02[git] 04paramat -> 03minetest/minetest_game: New lighter, greyer permafrost texture 133bca295 https://git.io/fjXnx (152019-07-13T22:35:25Z)