Time  Nick       Message
07:21 nerzhul    hello
07:21 nerzhul    can someone take a look at #9769 please, i need to continue my refactors for a more important plan on server
07:21 ShadowBot  https://github.com/minetest/minetest/issues/9769 -- Server class code cleanups by nerzhul
09:00 ANAND      #9705, anyone? :)
09:00 ShadowBot  https://github.com/minetest/minetest/issues/9705 -- set_fov: Add support for time-based transitions by ClobberXD
13:17 nerzhul_   ANAND: approved
13:24 ANAND      nerzhul_: Awesome, thanks! :)
13:29 ANAND      I think you forgot to add the relevant label, nerzhul_
13:32 nerzhul_   which one ? i am the only one to approve
13:44 ANAND      Huh, I thought that label was removed earlier
13:44 ANAND      Nvm
13:51 ANAND      Krock: Hi, did you take a look at https://github.com/minetest/minetest/pull/8755#issuecomment-620328708?
13:52 ANAND      Also, I've made the changes you've requested in #8817. The PR is ready for review; it's a lot simpler now.
13:52 ShadowBot  https://github.com/minetest/minetest/issues/8817 -- Settings: Prevent unnecessary writes by checking for modification by ClobberXD
16:01 rdococ     I went looking into the code, and apparently it turns out that flowing water vertices don't have proper normals on purpose, so that they aren't subjected to the face-based shading that other nodes are
16:01 rdococ     which is, like... super weird. why not just have an exception in the face-based shading code?
17:34 ANAND      #9781
17:34 ShadowBot  https://github.com/minetest/minetest/issues/9781 -- Pick node by middle click
17:41 Krock      will merge game#2666 in 10'
17:41 ShadowBot  https://github.com/minetest/minetest_game/issues/2666 -- Cart inventory image was rendering the side twice by Zughy
17:42 sfan5      ANAND: perhaps middle click should just have no functionality and be up to games to implement
17:43 Krock      ^
17:43 sfan5      it's not like it can really work client-side anyway since the client has to wait for the server to send the inventory change
17:45 rubenwardy yeah
17:51 Krock      merging
17:54 nerzhul_   middle click is used to split inventory stacks by 2 no ?
17:54 nerzhul_   or is this a game feature ?
17:54 nerzhul_   can someone take 5-10 mins to check #9769 please ?
17:54 ShadowBot  https://github.com/minetest/minetest/issues/9769 -- Server class code cleanups by nerzhul
17:54 Krock      nerzhul_: talking about walking-mode, not formspec-mode
17:55 sfan5      in Minecraft, if you are in creative mode it will automatically grab the node you are pointing at from the creative inventory and put it in the current slot
17:56 Krock      explains where the idea comes from
17:57 ANAND      That makes sense
17:58 ANAND      As a callback? on_pick?
17:58 Krock      some kind of user input callback
17:58 Krock      this flows kinda into the keypress callback requset
17:59 ANAND      Yeah, that'd be nice
17:59 nephele    sfan5, i would like even more than just middle click to be available for me to implement :P
18:00 ANAND      nephele: See #8437
18:00 ShadowBot  https://github.com/minetest/minetest/issues/8437 -- Discussion regarding implementation of custom keypress sending
18:00 ANAND      You can have all the keys
18:00 nephele    "sending"
18:00 sfan5      ANAND: more like minetest.register_on_middleclicknode
18:00 nephele    not what i'm looking for, i need csm support
18:00 nephele    :)
18:00 sfan5      that is more in line with how existing node interaction works
18:01 ANAND      Makes sense
18:01 nerzhul_   i think it's a good idea ot have a register for middleclick, as we have for other buttons
18:01 nephele    ANAND, also, that api sucks, i need to be able to have a "some key was pressed" state where i can check which are currently pressed
18:01 nephele    so i can deal with shift and so
18:02 nephele    if that proposed api is used i need to keep lots of extra data around to see if shift is still pressed and so
18:02 ANAND      nephele: It's assumed that get_player_control would contain all these keys as well
18:02 ANAND      Oops, I digressed
18:02 sfan5      yes, you can relyon get_player_control in punchnode/dignode/... callbacks
18:03 nephele    why bother with that api then?
18:03 Krock      basically in all callbacks
18:03 ANAND      Callbacks are useful for certain usecases, while polling is useful for certain other usecases
18:03 sfan5      nephele: I don't know, haven't read the issue ANAND linked :)
18:04 sfan5      but yeah avoiding polling would be one reason
18:04 nephele    Callbacks are usefull, but it's not usefull for me to regiter a callback for every key seperately only to be able tosee keypreses
18:04 ANAND      Huh, it just occurred to me that I'd have to implement middle clicks on Android :D
18:04 nephele    then mt needs to keep around my callback 126 times
18:04 ANAND      Or does middle clicking functionality already exist on Android?
18:05 nephele    I want to have an api callback for "on mouse state change" and "keyboard state change"
18:05 nephele    but anyway, i need this in csm
18:05 nephele    networking does not help at all
18:06 nephele    I don't think sending keypresses over the network makes sense in any case... communicating actions to the server is one thing, but not like this
18:06 ANAND      Implementing custom keypresses in CSM would be the next step, but probably only after someone implements SSCSM first :)
18:07 nephele    although, maybe server send csm mods are a requirement
18:07 nephele    i nominate krock
18:07 nephele    just because :D
18:07 ANAND      lol
18:07 sfan5      SSCSM is on my list of things I want to implement actually
18:07 ANAND      Oh nice!
18:07 sfan5      but there's a few things that have to be decided and solved before that can happen
18:08 rubenwardy yeah
18:08 nephele    i can decicde stuff with coins :D
18:08 ANAND      I was once planning to give it a shot too, but then realised it's way out of my league :)
18:10 rubenwardy namely: #7041
18:10 ShadowBot  https://github.com/minetest/minetest/issues/7041 -- Client-side API sandboxing
18:10 rubenwardy others: How to deal with proprietary client-scripts, and consent?  Policies for vulnerabilities (I support a soft kill switch)
18:11 nephele    >Disable LuaJIT and JIT compiling
18:11 nephele    why?
18:11 rubenwardy LuaJIT is not secure
18:11 sfan5      JIT is potentially insecure*
18:11 nephele    in what way?
18:11 sfan5      well there's another very important thing: you don't want to crash the entire client in case of malfunctioning mods
18:12 nephele    That's more of a multithreafing thing no? if luajit could potentially crash or leak i would just run it in a different namespaced prcoess per-mod (where this can be done, i.e os support)
18:12 nephele    >Seperate mods into their own lua_state's and block access to certain functions unless the mod receives permission from the user (e.g. screenshot api)?
18:12 rubenwardy not multithreading, but Lua environments
18:12 sfan5      "just"
18:13 nephele    sfan5, yes, just, the linux namespacing api is not that hard to use ;)
18:13 sfan5      but yes this is a sorta requirement for it to actually be sure
18:13 nephele    if the os does not have that then don't try it
18:13 rubenwardy this is still my position: https://github.com/minetest/minetest/issues/7043#issuecomment-504223589
18:13 sfan5      nephele: not that, you cannot "just" trivially replace blocking callbacks into Lua with the appropriate IPC
18:13 nephele    sfan5, i was talking about luajit
18:13 p_gimeno   nephele: do you mean to run Lua in a process separate from Minetest?
18:14 p_gimeno   LuaJIT*
18:14 sfan5      nephele: that doesn't change anything?
18:14 nephele    The claim was that luajit is insecure, not that the generated code is
18:15 p_gimeno   nephele: FFI allows you to access any C library function, including e.g. fopen, fwrite
18:16 nephele    p_gimeno, you don't need to use FFI
18:16 nephele    I don't see what the point of allowing FFI to the c lib would be either for client mods
18:17 nerzhul_   running lua in a separate process or thread has zero sense as the mods relied on core sync things, like the map and the environment.
18:18 rubenwardy there's event-based scripting which is still relevant
18:18 nerzhul_   you will just add interthread communication with extract contracts to add to talk between threads, or many locks
18:18 rubenwardy for example, client-side community
18:18 rubenwardy *prediction
18:18 rubenwardy things like rendering will be difficult
18:18 nerzhul_   event based it was we implemented with the on_* calls, it's on events
18:18 rubenwardy you'll get tearing unless you block for sync
18:19 nerzhul_   i want to see core and a mod modifying a block each one together and see the result haha
18:19 nerzhul_   no don't loose effort on thread Lua, no sense in our context
18:19 nerzhul_   just optimize the more needed calls and reduce the lua vm overhead (in all lua contexts, not only lua-dead-jit)
18:21 nerzhul_   thanks Krock for the review
18:22 Krock      yw
18:26 ANAND      Discussion in #4064 seems to be going a tad offtopic
18:26 ShadowBot  https://github.com/minetest/minetest/issues/4064 -- Add support/API for mods written in C++
19:05 rubenwardy updated #8592
19:05 ShadowBot  https://github.com/minetest/minetest/issues/8592 -- Add core.open_url() to main menu API by rubenwardy
19:05 rubenwardy and the ci failed
19:07 sfan5      isn't that a good thing?
19:08 rubenwardy yeah, it's good to have the CI catch things
19:44 rubenwardy https://github.com/rubenwardy/minetest/projects/1
19:45 rubenwardy discovered a thing
19:45 rubenwardy you can link PRs in other repos, meaning I can have my own project to order things by priority without spamming the main repo or having others modify it
19:46 rubenwardy thanks to ANAND
19:47 sfan5      since you're here right now: what are your thoughts on having falling sand be stuck on the players (and mobs) heads?
19:47 rubenwardy in terms of gameplay, I think it should suffocate players (in MTG)
19:47 rubenwardy in terms of an engine default, I'm not sure
19:48 rubenwardy having players get stuck is not good
19:48 sfan5      this would be necessary in order to make falling stacks of sand work properly
19:48 sfan5      the sand gets stuck on the player, not the player on the sand
19:48 sfan5      thought you can't jump if that's what you mean
19:48 sfan5      though*
19:48 rubenwardy I see you've moved it to a separate method on the entity, this means it'll be easier for a mod to change anyway. A player blocking it is a sensible default
19:49 sfan5      to change that you would edit the initial_properties
19:50 rubenwardy oh, using physical?
19:50 rubenwardy thought you were referring to the sand placing itself
19:51 sfan5      sorry for not being precise, I was referring to collide_with_objects
19:58 Krock      rubenwardy: #9780 zeuner is using a newer build than what you linked
19:58 ShadowBot  https://github.com/minetest/minetest/issues/9780 -- Player disappears while being punched: server crashes due to assertion failure
19:59 rubenwardy my bad
20:13 rubenwardy omg, Mike Pall is developing LuaJIT again!
20:24 p_gimeno   rubenwardy: reference?
20:26 p_gimeno   I see a merge in the repo from time to time
20:36 rubenwardy https://github.com/LuaJIT/LuaJIT/commits/v2.1
20:41 sfan5      /home/runner/work/minetest/minetest/src/porting.cpp:716:78: error: cast from ‘HINSTANCE’ {aka ‘HINSTANCE__*’} to ‘int’ loses precision [-fpermissive]
20:41 sfan5      716 |  return (int)ShellExecuteA(NULL, NULL, url.c_str(), NULL, NULL, SW_SHOWNORMAL) > 32;
20:41 sfan5      why in gods name would ShellExecuteA return a HINSTANCE but only on win64
20:41 sfan5      oh wait
20:41 sfan5      rubenwardy: cast it to intptr_t instead
20:42 rubenwardy the docs say to cast to ent?
20:42 rubenwardy > It is not a true HINSTANCE, however. It can be cast only to an int and compared to either 32 or the following error codes below.
20:42 sfan5      when they say int they mean integer
20:42 rubenwardy oh right