| Time |
Nick |
Message |
| 00:03 |
|
fluxflux joined #minetest-dev |
| 01:11 |
|
ANAND joined #minetest-dev |
| 01:25 |
p_gimeno |
ah here's my discussion about sandboxing: https://love2d.org/forums/viewtopic.php?p=221790#p221790 |
| 01:26 |
p_gimeno |
it's not really love specific, the only love specific calls are love.filesystem.load (to load a file using PhysFS, but that's not necessary here) and love.event.quit to exit. |
| 01:28 |
p_gimeno |
love.filesystem.load works like Lua's loadfile |
| 01:32 |
p_gimeno |
it's probably a moot point, unless someone uses deserialize on user input as discussed (I doubt anyone does, but the possibility is scary). |
| 01:44 |
|
pmp-p joined #minetest-dev |
| 02:42 |
|
erlehmann joined #minetest-dev |
| 02:43 |
|
nashimus joined #minetest-dev |
| 03:00 |
|
nepugia joined #minetest-dev |
| 03:37 |
|
Taoki joined #minetest-dev |
| 03:47 |
|
ssieb joined #minetest-dev |
| 05:31 |
|
ANAND joined #minetest-dev |
| 07:34 |
p_gimeno |
about #9369: setfenv(f, safe and {loadstring = function() end} or env) - don't create the table on the fly, that's a waste of resources |
| 07:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/9369 -- Fix misleading documentation on minetest.deserialize(). by luk3yx |
| 07:38 |
p_gimeno |
oh maybe the table is necessary, but the function definitively isn't |
| 07:52 |
p_gimeno |
but either both tables are created on the fly, or none is. Does the function need an environment shared between calls? |
| 07:52 |
p_gimeno |
the function -> the code to deserialize |
| 08:00 |
|
erlehmann joined #minetest-dev |
| 08:25 |
|
pmp-p joined #minetest-dev |
| 08:38 |
|
ShadowNinja joined #minetest-dev |
| 08:57 |
sfan5 |
p_gimeno: the environment does not need to be shared, but there's no reason to recreate it since it doesn't change |
| 08:57 |
sfan5 |
however... |
| 09:00 |
sfan5 |
> =deserialize("x = 12344") |
| 09:00 |
sfan5 |
nil |
| 09:00 |
sfan5 |
> =deserialize("return x") |
| 09:00 |
sfan5 |
12344 |
| 09:00 |
sfan5 |
..is what happens if you don't recreate the environment |
| 09:00 |
p_gimeno |
yes that was my thought too |
| 09:02 |
p_gimeno |
maybe: safe_env = setmetatable({loadstring = ...}, {__newindex = function() error("Deserialized functions can't create globals") end} |
| 09:02 |
p_gimeno |
) |
| 09:04 |
sfan5 |
what if you modify the loadstring global? is there anything you can do using a function object? |
| 09:05 |
p_gimeno |
an empty one? not that I know |
| 09:05 |
p_gimeno |
you could destroy it for subsequent callers though (DoS) |
| 09:06 |
sfan5 |
well this also goes for the non-safe env, can you modify a function without replacing the reference itself? |
| 09:06 |
p_gimeno |
functions are immutable if that's what you mean |
| 09:08 |
sfan5 |
Hope that's 100% true and not just "true unless you do weird things with it" |
| 09:09 |
sfan5 |
(the debug library can edit functions after they exist, but that's not available here of course) |
| 09:10 |
p_gimeno |
can it? I thought it could modify upvalues, not code |
| 09:11 |
sfan5 |
that's what I was thinking of |
| 09:12 |
p_gimeno |
yeah, not without debug.setupvalue/setlocal |
| 09:15 |
|
realzies joined #minetest-dev |
| 09:45 |
p_gimeno |
I can't help wondering why deserialize needs access to loadstring |
| 09:48 |
sfan5 |
it supports serializing functions |
| 09:48 |
sfan5 |
but in practice that doesn't work anyway since mod security forbids you to load bytecode |
| 09:55 |
p_gimeno |
I'm curious about something. In the nested loadstring call, is loadstring treated as an upvalue or as a global? in the latter case, wouldn't it reference the current environment, infinitely recursing on loadstring()? |
| 09:56 |
p_gimeno |
that would solved by localizing them: local loadstring, setfenv = loadstring, setfenv |
| 10:03 |
p_gimeno |
tested, it seems to work as expected by referencing the environment at definition time |
| 10:08 |
rubenwardy |
Loadstring isn't called in the serialised string |
| 10:08 |
rubenwardy |
The serialised string is a Lua function which returns the table |
| 10:08 |
rubenwardy |
Without the function syntax |
| 10:09 |
rubenwardy |
Nested tables are done using a companion table |
| 10:09 |
sfan5 |
https://github.com/minetest/minetest/blob/master/builtin/common/serialize.lua#L121 |
| 10:10 |
rubenwardy |
That should not be supported |
| 10:11 |
rubenwardy |
From the beginning |
| 10:11 |
rubenwardy |
Why was that considered a good idea |
| 10:34 |
|
T4im joined #minetest-dev |
| 10:38 |
|
df458 joined #minetest-dev |
| 11:27 |
|
texmex joined #minetest-dev |
| 11:27 |
|
kollaps[m] joined #minetest-dev |
| 11:28 |
|
Qiangong2[m] joined #minetest-dev |
| 11:33 |
|
Fixer joined #minetest-dev |
| 11:35 |
|
proller joined #minetest-dev |
| 12:00 |
|
Beton joined #minetest-dev |
| 12:56 |
|
absurb joined #minetest-dev |
| 13:11 |
|
proller joined #minetest-dev |
| 14:09 |
|
Taoki joined #minetest-dev |
| 14:59 |
|
pmp-p joined #minetest-dev |
| 15:06 |
|
calcul0n joined #minetest-dev |
| 15:15 |
|
proller joined #minetest-dev |
| 16:47 |
|
Fixer joined #minetest-dev |
| 17:03 |
|
YuGiOhJCJ joined #minetest-dev |
| 17:25 |
|
Fixer joined #minetest-dev |
| 17:29 |
|
CBugDCoder joined #minetest-dev |
| 17:31 |
|
Krock joined #minetest-dev |
| 17:39 |
|
Fixer joined #minetest-dev |
| 18:25 |
|
Fixer joined #minetest-dev |
| 18:27 |
|
Fixer_ joined #minetest-dev |
| 18:32 |
|
Fixer joined #minetest-dev |
| 18:33 |
|
Fixer_ joined #minetest-dev |
| 19:01 |
|
AntumDeluge joined #minetest-dev |
| 19:28 |
|
ssieb joined #minetest-dev |
| 19:58 |
|
Andrey01 joined #minetest-dev |
| 20:04 |
|
AntumD joined #minetest-dev |
| 20:09 |
|
NoctisLabs joined #minetest-dev |
| 20:11 |
|
NoctisLa1 joined #minetest-dev |
| 20:11 |
Andrey01 |
hello, I`m currently playing around Irrlicht Engine and I have some issues: the animated node looks like very jerky and surprisingly the node disappears if it is at edge of the camera view range |
| 20:12 |
Andrey01 |
does anybody know how to solve these issues? here is a code: https://ufile.io/n6kb0kt7 |
| 20:13 |
|
NoctisLa1 joined #minetest-dev |
| 20:37 |
|
NoctisLa1 joined #minetest-dev |
| 20:40 |
|
NoctisLa1 joined #minetest-dev |
| 20:56 |
|
NoctisLa1 joined #minetest-dev |
| 21:25 |
|
NoctisLa1 joined #minetest-dev |
| 21:28 |
|
NoctisLabs joined #minetest-dev |
| 21:34 |
|
NoctisLa1 joined #minetest-dev |
| 21:45 |
|
paramat joined #minetest-dev |
| 21:45 |
paramat |
will merge trivial game#2588 in 2 hours |
| 21:45 |
ShadowBot |
https://github.com/minetest/minetest_game/issues/2588 -- Fix simple bed height, make player lay down just above it, not inside it by paramat |
| 22:40 |
|
df458 joined #minetest-dev |
| 22:45 |
|
behalebabo joined #minetest-dev |
| 22:52 |
|
proller joined #minetest-dev |
| 23:27 |
|
absurb_ joined #minetest-dev |
| 23:29 |
|
Wuzzy joined #minetest-dev |
| 23:33 |
|
pmp-p joined #minetest-dev |
| 23:51 |
Wuzzy |
any ideas on how to make an unit test for pathfinder? |
| 23:52 |
Wuzzy |
pathfinder relies on a real existing world with real nodes, so it seems a pathfinder unit test would first have to get access to a dummy world ... tricky. what do you think? |
