New week, new Rust! What are you folks up to?
I thought I’d rewrite my FFI guide to be more like one big extended tutorial instead of a bunch of disparate articles. The new tutorial is from the perspective of someone writing a REST client which has a Qt frontend and uses
reqwest under the hood.
I’ve passed the most difficult part, calling
qmake and making sure everything compiles and runs without any linker errors.
If there’s anything anyone wants me to talk about regarding FFI and integrating Rust with other languages, let me know via the associated issue.
I’m also spending some more time with
env_logger. I think we should work towards another release soon.
I added some more fundamental client methods to
elastic over the last week and will work on support for bulk next and generating some documentation for the raw endpoints.
Trying to get at least a prototype of dependency license bundling working in
cargo-lichking. Will definitely need some more work before it’s “done”, but once I fix the performance issue with checking license content I think it’ll be good enough for an initial release.
Also just released Ammonia 1.0.0-rc2, which will probably be the last release candidate before the final release (unless new flaws are found).
I’m going to start working on simple motion blur support for my renderer:
It looks like a small task but the test scene needs some HDR lighting (lacking a proper sun&sky simulation) and the dependency on the openexr crate might be a problem for e.g. Travis CI. We will see …
Finishing up DropoutLayer implementation and starting to to implement LongTermShortTerm/RNN networks for juice / coaster-nn - if I feel mathish I might tackle CPU based calculation of gradients for nd-convolutions
Finally got my Piet interpreter cleaned up and pushed out the door. It works with a few initial test programs. Still have to do some thinking about how to handle large numbers.
This week I’m providing feedback to jturner314 who is making an awesome feature for ndarray.
Before, we supported multidimensional slicing like
.slice(s![1..5, 3..5, 1..5]) and now ndarray will support mixing indexing with slicing like
.slice(s![1..5, 0, 0]), and yes the dimensionality of the result is known statically.
(I’m also doing a few other small things.)