| Time |
Nick |
Message |
| 00:18 |
|
Sokomine joined #minetest-dev |
| 00:25 |
|
ShadowBot joined #minetest-dev |
| 00:35 |
|
Sokomine joined #minetest-dev |
| 01:22 |
|
v-rob joined #minetest-dev |
| 02:11 |
|
Alias joined #minetest-dev |
| 02:44 |
|
hrmmmm joined #minetest-dev |
| 03:30 |
|
hrmmmm joined #minetest-dev |
| 04:00 |
|
MTDiscord joined #minetest-dev |
| 04:58 |
|
calcul0n joined #minetest-dev |
| 05:40 |
|
hmmmm joined #minetest-dev |
| 05:41 |
|
hmmmm joined #minetest-dev |
| 08:22 |
|
Warr1024 joined #minetest-dev |
| 09:26 |
|
hmmmmm joined #minetest-dev |
| 09:28 |
|
hrmmmm joined #minetest-dev |
| 10:14 |
|
crazyR joined #minetest-dev |
| 10:49 |
sfan5 |
~tell Desour_ since you're doing some cleanup recently you can also do #12366 if you don't mind |
| 10:49 |
ShadowBot |
sfan5: OK. |
| 10:55 |
|
nrz joined #minetest-dev |
| 12:22 |
sfan5 |
#13200 still needs reviews |
| 12:22 |
ShadowBot |
https://github.com/minetest/minetest/issues/13200 -- Fix missing particlespawner values by velartrill |
| 13:38 |
|
kilbith joined #minetest-dev |
| 14:34 |
|
numzero joined #minetest-dev |
| 16:13 |
|
olliy joined #minetest-dev |
| 16:46 |
|
olliy joined #minetest-dev |
| 16:48 |
|
Desour joined #minetest-dev |
| 17:09 |
|
olliy joined #minetest-dev |
| 17:12 |
Krock |
found a new bug in IMoveAction::apply .. hopefully that one will be done before 5.7.0 |
| 17:51 |
Krock |
will push https://krock-works.uk.to/u/patches/0001-InventoryManager-Fix-crashes-caused-by-set_list-and-.patch in 20 minutes. reproduced with cornernote's 3x3,4x4,5x5 workbench. It's far from perfect but at least in this case Minetest no longer crashes |
| 17:52 |
Krock |
(case being: you place an item into the workbench grid and then attempt to move it) |
| 17:53 |
MTDiscord |
<Niklp> #11805 ? |
| 17:53 |
ShadowBot |
https://github.com/minetest/minetest/issues/11805 -- IMoveAction::apply(): Assertion `move_count <= count' failed |
| 17:54 |
Krock |
ah right. while debugging I ended up doing the same fix there |
| 17:55 |
Krock |
however the actual fix for cornernote's mod is the other part where the pointers get updated |
| 17:56 |
Desour |
interesting more UaFs caused by mod callbacks deleting stuff on which we still have a ref in c++ |
| 17:56 |
Krock |
inventory lists must be replaced with shared pointers at some point to prevent this from happening |
| 17:56 |
Krock |
or never have the pointers cached in first place |
| 17:58 |
Krock |
sfan5: was the 5.7.0 release planned for today or later? |
| 17:58 |
Desour |
I think a mod that deletes a list in allow_inv.*, and still allows moving/putting/taking from/to that list, should crash |
| 17:59 |
Krock |
it should error, not crash. |
| 17:59 |
Desour |
crash as in lua crash/error, yes |
| 18:00 |
Desour |
can actually the inv_from or inv_to also be deleted by a mod? |
| 18:02 |
Krock |
all but player inventories. yes. |
| 18:05 |
Desour |
Krock: list_to also gets accessed in line 385 after allowMove |
| 18:08 |
Krock |
oh right |
| 18:10 |
Krock |
I hate this code |
| 18:15 |
rubenwardy |
before release, it should probably be discussed whether #13242 should be reverted and the _optioanl drawtype deprecated instead |
| 18:15 |
ShadowBot |
https://github.com/minetest/minetest/issues/13242 -- Enable connected glass by default by PrairieAstronomer |
| 18:18 |
rubenwardy |
reading the issue, it seems like the optional drawtype is problematic in itself |
| 18:18 |
rubenwardy |
or texture packs should have some ability to control rendering more than just textures - windows are basically invisible in soothing32 with connected glass |
| 18:18 |
Krock |
reverting is easier |
| 18:19 |
Pexin |
noncore opinion: yes revert 13242. erle's opinion is weird because it's unclear if some people understand connected obsidian glass is SUPPOSED to be invisible. my regular server uses it explicitly for that because we're accustomed to enabling connected, but it should actually be decided by the server itself |
| 18:21 |
|
HuguesRoss9 joined #minetest-dev |
| 18:21 |
|
numzero joined #minetest-dev |
| 18:21 |
Desour |
for backwards compat reasons, it's probably better to revert to the old default. I'd not like a drawtype deprecation |
| 18:23 |
rubenwardy |
deprecation doesn't mean removal |
| 18:23 |
rubenwardy |
it means "stop using this" |
| 18:24 |
rubenwardy |
the eventual route could just be to remove the setting and always use non-framed |
| 18:28 |
Desour |
the optional_ drawtype has its purpose, if the player wants to choose and the server doesn't really care. it exists with the same justification as texture packs. therefore I'd not even deprecate it |
| 18:31 |
Desour |
the default should be defined by the server (or its mod) though, and not hardcoded by the engine. so we should have a field in nodedef or tiledef to give the setting's default (the default of that field can then be changed to connected if we want) |
| 18:40 |
sfan5 |
Krock: the milestone is set for today but I don't think we'll be releasing |
| 18:43 |
|
Noisytoot joined #minetest-dev |
| 18:51 |
Krock |
alright |
| 18:52 |
Krock |
Q: how should we deal with changing InventoryLists within Lua callbacks? would it be okay to prevent resizing/deleting the source and destination inventory list? |
| 19:08 |
|
MTDiscord1 joined #minetest-dev |
| 19:22 |
sfan5 |
sure |
| 19:29 |
Krock |
well then #13360 |
| 19:29 |
ShadowBot |
https://github.com/minetest/minetest/issues/13360 -- InventoryManager: Disallow resizing or deleting inventory lists that are in use by SmallJoker |
| 21:07 |
|
numzero joined #minetest-dev |
| 21:30 |
|
dzho joined #minetest-dev |
| 21:30 |
|
dzho joined #minetest-dev |
| 21:43 |
|
Guest54 joined #minetest-dev |
| 21:56 |
Guest54 |
Pexin elaborate on weird? i am sure that some people exist that understand that connected obby glass is supposed to be invisible, but i have yet to see builds that rely on it. |
| 22:07 |
Guest54 |
rubenwardy the reasons ppl use “optionally connected glass” is as desour says. minecraft has such thing too, as part of a a client-side modification called “optifine” (it does a bunch of other stuff to). regarding rendering, instead of texture pack control to work around long-standing rendering bugs i'd rather have a fix so it works well |
| 22:07 |
Guest54 |
with existing textures of any size. or even a new drawtype that has a connected look but no physically raised frames, because then you could *maybe* just 9-slice existing textures for a seamless glass look … and *maybe* get away with it (i have not tried it and it might be horrible, obviously). |
| 22:07 |
Guest54 |
to → too |
| 22:59 |
|
YuGiOhJCJ joined #minetest-dev |
| 23:10 |
|
jonadab joined #minetest-dev |
