Time  Nick         Message
04:09 Wuzzy        paramat: I object to #6629.
04:09 ShadowBot    https://github.com/minetest/minetest/issues/6629 -- Biome dust node: Only place on 'walkable' cubic non-liquid drawtypes by paramat
13:54 kilbith      https://cdn.discordapp.com/attachments/332193918940872705/381072944530456576/image.jpg
13:54 kilbith      known issue on windows?
13:56 Krock        https://github.com/minetest/minetest/pull/5795#issuecomment-305400108
13:57 Krock        I guess it wasn't fixed properly then
13:57 Krock        nvm. these changes do not affect the HUD API
14:02 kilbith      guess the UTF-8 support is half assed
14:10 sfan5        unicode support is a mess in minetest anyway
14:10 sfan5        we use wchar_t* but then there's wide_to_narrow so you can't even be sure that char* is just ascii
14:10 sfan5        it makes no sense really
14:14 kilbith      I've forwarded to my team
14:15 Fixer        another example: https://cloud.githubusercontent.com/assets/4017302/26666984/1bf69286-46cd-11e7-8cdf-d66e8138f7cb.png
14:16 Fixer        i can imagine that in 2010, but in 2017?
14:49 kilbith      https://cdn.discordapp.com/attachments/350623618339962880/381081996350324736/erreur_vector_access.jpg
14:49 kilbith      what if datasize equals 2, you copy 0 elements to m_data[0]
14:50 kilbith      then it crashes when you try accessing it
14:51 kilbith      if your command equals to 2 bytes, without params
14:52 nerzhul      kilbith, yeah it's normal as datasize is sizeof(u8)
14:53 nerzhul      sizeof(data)
14:53 nerzhul      2 + networkpacket data (2 = opcode)
14:53 sfan5        does &m_data[0] work on zero-size vectors?
14:54 sfan5        i don't see why this would crash otherwise
14:54 nerzhul      sfan5, as memcpy use address it's a valid address then yes, because it's the pointer to the start of the range, empty or not
14:54 celeron55    sfan5: visual studio debug builds crash in an assert when you do that
14:54 nerzhul      memcpy(x,x,0) is a NOP
14:54 celeron55    nothing else does
14:54 sfan5        hmm
14:55 celeron55    wait, was that on strings?
14:55 celeron55    yeah it was a string thing
14:55 celeron55    on vectors it's obviously not allowed anywhere
14:55 sfan5        nerzhul: in theory a vector does not need to have memory region when you .resize(0) so m_data does not necessarily have an address
14:55 nerzhul      sfan5, exact
14:55 sfan5        but it's a no-op as you said so it shouldn't matter
14:56 nerzhul      i also think it's a nonsense to have m_data < 3 because you always have something to tell
14:56 nerzhul      there is no command without arg in MT protocol
19:05 paramat      Krock sfan5 can i merge #6629 now? semi-trivial but i'd rather have another +1
19:05 ShadowBot    https://github.com/minetest/minetest/issues/6629 -- Biome dust node: Only place on 'walkable' cubic non-liquid drawtypes by paramat
19:07 Krock        how about fences and nodeboxes? wouldn't it be better to check the node definition?
19:08 paramat      see my discussion in the issue
19:09 paramat      you would need to check the top face of every nodebox (there could be very many) and such cases would be rare
19:09 paramat      what about fences? they shouldn't have dust
19:10 Krock        instead of checking the collision box, the definition given from the node registration could be used. Just an idea
19:11 paramat      ?
19:11 Krock        but I guess that's not required to add if checking against the drawtypes works quite good to
19:12 paramat      fences are narrow so dust would not look right floating above them
19:14 Krock        right. The code looks good but I've seen that Wuzzy would like to test it first. Perhaps give them 24 hours and merge it afterwards?
19:15 Wuzzy        if you tested it already, i have no objections
19:15 Krock        ah well, then merge it, paramat.
19:15 paramat      no need, i've tested it. i want to merge #6632 now before conflicts arise so i want to merge 6629 with it
19:15 paramat      ok
19:15 ShadowBot    https://github.com/minetest/minetest/issues/6632 -- Ore API documentation: Update and improve by paramat
19:16 paramat      merging those 2 now
19:18 paramat      oh and #6412
19:18 ShadowBot    https://github.com/minetest/minetest/issues/6412 -- [2nd] Load textures from the subfolders in texturepacks by numberZero
19:25 Wuzzy        thanks
19:40 paramat      ok done back later
21:27 ThomasMonroe #6637
21:27 ShadowBot    https://github.com/minetest/minetest/issues/6637 -- Custom Rendering Engine or Possible Irrlicht "replacement"?
21:37 nerzhul      another time the same discussion
21:46 ThomasMonroe wydm nerzhul?
21:50 nerzhul      sorry i don't understand this acronym
21:51 Krock        "what do you mean"
21:51 Krock        ThomasMonroe, I'm pretty sure we already have an issue about this. It was discussed previously too
21:52 Krock        like forking irrlicht or using another engine
21:52 ThomasMonroe ah ok, lol we thought we had had this amazing idea that noone had ever thought of b4, should have known better
21:52 ThomasMonroe its not like its a bad idea though, just its alot of work
22:06 rubenwardy   ThomasMonroe, comes up every few months or so
22:06 rubenwardy   https://github.com/minetest/minetest/issues/5365
22:07 sfan5        hey a snarky reply by me
22:07 rubenwardy   :D
22:08 numzero      sfan5
22:08 numzero      I have a question on #6621
22:08 ShadowBot    https://github.com/minetest/minetest/issues/6621 -- Fix dark liquids by numberZero
22:08 sfan5        yes
22:09 numzero      What is better: 1. remove LightPair and return to bitwise operations,
22:09 numzero      2. Change LightPair to provide getters/setters
22:10 numzero      3. Keep as is as the compilers are VERY unlikely to break that
22:10 sfan5        neither
22:11 numzero      (for the latter case, I can write a standard-compliant class with compatible interface that might be used if anything seems to fail, but it would be slow and heavy)
22:11 sfan5        just make the u16 constructor deconstruct the param with bitwise ops
22:11 sfan5        and then you can remove the union{}
22:11 numzero      okay
22:31 paramat      i'm working on a 'water sounds' mod for MTG, triggered by flowing water. because it does a node search around players it is optimum to move fire sounds into it as well, to avoid 2 separate node searches
22:32 paramat      so i thought this might as well be an 'ambience' mod, which then later can be updated to use the ambience sounds API if/when that appears
22:37 paramat      the sound code of the fire mod could remain but be default disabled. to allow using the fire mod alone
22:38 paramat      i'm still working on a skins mod for MTG, it's a big task, working through the database
22:41 sfan5        you plan to include those in 0.5, correct?
22:45 paramat      yeah
22:46 paramat      i just want MTG to not be too embarassing for 0.5.0: skins mod, water ambience and a hostile mob
22:48 paramat      i also have a bunch of other improvements coming, a few month's work
22:49 paramat      i think it's important that android is improved before 0.5.0 so i'd like to see 0.5 maybe 2 months into 2018, to give lordfingle a chance to work on it (he says he may be able to in 2018)