Closed virtualritz closed 1 day ago
It seems something like
theon
is needed...
Yeah, I've felt this way for some time too! For intermediate library crates, I really dislike making this choice for users. That's why theon
exists! It's a really difficult problem to solve (thoroughly) though. 😬
Is this dead?
Only insofar that I've stretched myself a bit thin with projects. 😅 I don't consider it dead and I've been thinking about this crate recently. I've been planning on publishing some significant changes to decorum
first before starting work on theon
again, since it integrates with decorum
. From there, I'd like to turn my sights back to plexus
.
That all being said, I'm not sure when any of this will actually happen at the moment.
Speaking of which, bumping this to latest
nalgebra
is the only dependency that breakstheon
...
That sounds about right. In my experience, nalgebra
has had the most API churn and is the most involved crate to maintain a reverse integration with.
I've published version 0.1.0
, which mostly just updates the Rust Edition, MSRV, dependencies, and all integrations. This bring much better LAPACK platform support. Features now use the corresponding crate name per convention (there is no longer a "geometry" prefix). The APIs are yet unchanged, but I have plans.
Awesome! Thanks heaps. ☺️
On that note, there is yet another one, lina
, by Lukas, used by his lox
crate, which seems to be similar to plexus
. lox
came out of his master thesis which curiously mentions plexus
btw.
It seems something like
theon
is needed more than ever with the number of lin. algebra crates exploding and every other crate using something else. Especially since the popularnalgebra
becomes slow as molasses indebug
builds. If more crates relied on sth. liketheon
this would be a non-issue.Speaking of which, bumping this to latest
nalgebra
is the only dependency that breakstheon
, all othergeometry-*
features work fine against the resp. latest geometry crates.