| 4 Nov 2024 |
 Nathan Petrangelo | * Dynamic linking is probably only conceptually possible if you don’t use generics. Trait objects could probably be allowed though, because they only carry a pointer and a vtable. That would probably be the primary way of carrying flexible type safety through the ABI boundary. | 05:36:35 |
 Ribbon |
wake up
the day is very good
you study about patents on open-source software
day ruined
| 05:37:28 |
| Ribbon | *
- wake up
- the day is very good
- you study about patents on open-source software
- day ruined
| 05:37:40 |
| Nathan Petrangelo | In reply to @holobrine:matrix.org
So a dynamically linked library could have functions with trait objects as parameters when they don’t want to constrain the allowed types to just theirs So in conclusion, if dynamic linking ever came to Rust, you’d pay in small performance cost for large compile time gains | 05:43:04 |
| Nathan Petrangelo | In reply to @holobrine:matrix.org
So a dynamically linked library could have functions with trait objects as parameters when they don’t want to constrain the allowed types to just theirs * So in conclusion, if dynamic linking ever came to Rust, you’d probably pay in small performance cost for large compile time gains | 05:43:24 |
| Nathan Petrangelo | In my way of thinking, that's desirable for dev builds, but I want the performance for production | 05:50:32 |
| Ribbon | My focus is stability. | 05:52:17 |
| Ribbon | That's why I fucking love Redox. | 05:52:27 |
| Ribbon | Ron Williams: What is the hardest thing on math for you? | 05:56:29 |
 Ron Williams | In reply to @draft_isa:matrix.org
Ron Williams: What is the hardest thing on math for you? Explaining things | 06:01:14 |
| Ribbon | In reply to @rw_van:matrix.org
Explaining things For me is calculation with letters. | 06:01:55 |
| Ribbon | Hidden relations... | 06:02:03 |
| Ribbon | "Determine the value of x, but y is not known"
😶 | 06:03:02 |
| Ribbon | * "Determine the value of x, but y is not known" | 06:03:18 |
| Ribbon | 😐️ | 06:03:22 |
| Ron Williams | This was something I found very hard to deal with:
https://en.wikipedia.org/wiki/Jacobian_matrix_and_determinant | 06:04:53 |
| Ribbon | 
Download image.png | 06:07:20 |
| Ribbon | https://research.nvidia.com/person/tero-karras | 06:18:24 |
| Nathan Petrangelo | In reply to @rw_van:matrix.org
This was something I found very hard to deal with:
https://en.wikipedia.org/wiki/Jacobian_matrix_and_determinant Please tell me you've heard of 3Blue1Brown, but just in case you haven't, his math explainers are top tier
https://youtu.be/Ip3X9LOh2dk?si=_uFCXsJ08RvthC1y | 06:18:41 |
| Nathan Petrangelo | I still remember what a determinant is, conceptually, because he explained it so well | 06:20:07 |
| Ron Williams | In reply to @holobrine:matrix.org
I still remember what a determinant is, conceptually, because he explained it so well It took me a long time to get an understanding of eigenvalues and eigenvectors, when really it's the name that's confusing, not the concept. | 06:21:19 |
| Nathan Petrangelo | That happens a lot in math lol | 06:21:45 |
| Nathan Petrangelo | While we're talking about math, I must also share about geometric algebra
https://youtube.com/playlist?list=PLVuwZXwFua-0Ks3rRS4tIkswgUmDLqqRy&si=UTZwVPvkVm5su7IG | 06:23:00 |
| Nathan Petrangelo | Bivectors shed a more intuitive light on determinants | 06:24:03 |
| Nathan Petrangelo | and the wedge product is better than the cross product (it actually scales to more dimensions) | 06:26:59 |
| Nathan Petrangelo | As it happens, complex numbers are isomorphic to the even subalgebra in two dimensions (basically i is a bivector), and quaternions are isomorphic to the even subalgebra in 3 dimensions (j and k are also bivectors) | 06:31:19 |
| Nathan Petrangelo | so geometric algebra unlocks intuitions about quaternions | 06:31:59 |
| Ribbon | 
Download image.png | 06:38:46 |
| Ron Williams | In reply to @holobrine:matrix.org
so geometric algebra unlocks intuitions about quaternions I don't like to think too deeply about math anymore. I basically want to know enough math to understand neural networks at a conceptual level. I wish I had a better handle on Fourier transforms and cosine transforms. But the only real math I do now is budgets. | 06:39:07 |
| Ribbon | In reply to @draft_isa:matrix.org
sent an image. I generated this image with Stable Diffusion on AI Horde, very nice!! | 06:39:09 |
