| Time |
Nick |
Message |
| 00:05 |
|
Eragon joined #minetest-dev |
| 05:00 |
|
MTDiscord joined #minetest-dev |
| 05:06 |
|
AntumDeluge joined #minetest-dev |
| 08:06 |
|
whosit joined #minetest-dev |
| 12:06 |
|
Desour joined #minetest-dev |
| 12:45 |
|
guest52 joined #minetest-dev |
| 13:15 |
|
TheCoffeMaker joined #minetest-dev |
| 14:14 |
rubenwardy |
bump for #14964 |
| 14:14 |
ShadowBot |
https://github.com/minetest/minetest/issues/14964 -- SDL: Use scancodes for keybindings by y5nw |
| 14:14 |
rubenwardy |
I'll try and test it over the hols |
| 14:49 |
|
ivanbu joined #minetest-dev |
| 15:42 |
|
Desour_ joined #minetest-dev |
| 15:55 |
|
TheCoffeMaker joined #minetest-dev |
| 18:33 |
MTDiscord |
<abdu7944> Do the generated artifacts of this PR have SDL enabled znd uses by default? |
| 18:35 |
sfan5 |
yes |
| 18:35 |
sfan5 |
the mingw artifact matches what we would ship on a release |
| 18:42 |
|
SFENCE joined #minetest-dev |
| 18:50 |
|
SFENCE joined #minetest-dev |
| 19:06 |
MTDiscord |
<luatic> merging #15550 in 10m |
| 19:06 |
ShadowBot |
https://github.com/minetest/minetest/issues/15550 -- Restore `.x` models to working state by appgurueu |
| 19:07 |
MTDiscord |
<exe_virus> On the rotation subject, it would be extremely expensive for collision boxes if they become non-AABB. That said, you can define multiple collision boxes for an entity, keep them AABB, and rotate their centroids as the entity rotates, that's very much more performant. I'm focusing on AABB performance improvements next (with SIMD), so it might be reasonable to consider that in the pipeline |
| 19:08 |
MTDiscord |
<luatic> I think performance will survive so long as a bounding box is used |
| 19:08 |
MTDiscord |
<luatic> e.g. bound the rotated box with an AABB and use that to save yourself 99% of the costly rotated AABB collision calculations |
| 19:10 |
MTDiscord |
<exe_virus> That would also help, agree. Here's my half way solution: |
| 19:10 |
MTDiscord |
<exe_virus> https://cdn.discordapp.com/attachments/747163566800633906/1319018918559613001/Rotations_and_AABB.gif?ex=67646f98&is=67631e18&hm=2779f384b5eaa8aa04a3afeb6be95a174109acceb90d143a4c7261931ed6548d& |
| 19:10 |
MTDiscord |
<exe_virus> (unsure how that looks on irc) |
| 19:15 |
pgimeno |
I haven't looked, but I presume MT doesn't use swept collisions, right? |
| 19:26 |
Desour_ |
@exe_virus: aha, I see a reuse! https://github.com/minetest/minetest/issues/12372#issuecomment-1136689810 |
| 19:53 |
sfan5 |
wow the irrlicht gui code is awful |
| 19:53 |
Krock |
it's the most beautiful and efficient code I ever saw |
| 19:53 |
Krock |
buttons are like 8 drawcalls |
| 20:38 |
MTDiscord |
<exe_virus> Thanks for finding it, desour, I forgot about that conversation, and now remember we only allow a single collision box per entity. |
| 20:38 |
MTDiscord |
<exe_virus> So my idea is moot |
| 20:39 |
MTDiscord |
<exe_virus> And more expensive likely, maybe, than one rotated box collision. Of course... 500 entity collisions, is not uncommon. That makes this much harder either way |
| 21:55 |
|
hwpplayer1 joined #minetest-dev |
| 22:40 |
MTDiscord |
<landarvargan> Is it possible for mapblock sending to hog all of the server step? |
| 22:47 |
sfan5 |
generally it's possible for anything to hog all of everything |
| 22:47 |
sfan5 |
you will have to find out if that's the case using the engine profiler or other means |
| 22:50 |
MTDiscord |
<landarvargan> y |
| 22:50 |
MTDiscord |
<landarvargan> oops, I shall test once I upgrade my VPS |
| 23:33 |
|
panwolfram joined #minetest-dev |
