yamt opened PR #4451 from wasi-nn-wit
to main
:
it imports from "wasi-nn" module with no version.
(unlike "wasi_ephemeral_nn")
is it expected? i couldn't find a way to override it.i suspect that the alignment of tensor-data, which is of list<u8>,
is just 1. isn't it better to use a larger alignment of tensor-data
to increase the chance for the receiver to process it in-place?<!--
Please ensure that the following steps are all taken care of before submitting
the PR.
[ ] This has been discussed in issue #..., or if not, please tell us why
here.[ ] A short description of what this does, why it is needed; if the
description becomes long, the matter should probably be discussed in an issue
first.[ ] This PR contains test cases, if meaningful.
- [ ] A reviewer from the core maintainer team has been assigned for this PR.
If you don't know who could review this, please indicate so. The list of
suggested reviewers on the right can help you.Please ensure all communication adheres to the code of conduct.
-->
bjorn3 submitted PR review.
bjorn3 created PR review comment:
Maybe a new version of the
wasi-nn
crate could be created which doeswit_bindgen_rust::import!("spec/wasi-nn.wit.md");
internally? That way uses of the wasi-nn interface don't have to copy/submodule the wasi-nn specification.
abrown created PR review comment:
cc: @brianjjones; I opened https://github.com/bytecodealliance/wasi-nn/issues/64 about this a bit ago. I would hope that
import!
would be a viable approach.
abrown submitted PR review.
yamt submitted PR review.
yamt created PR review comment:
Maybe a new version of the wasi-nn crate could be created which does wit_bindgen_rust::import!("spec/wasi-nn.wit.md"); internally?
maybe.
yamt submitted PR review.
yamt created PR review comment:
i guess it isn't a big problem as the submodule can be shared with https://github.com/bytecodealliance/wasmtime/tree/main/crates/wasi-nn once it's adapted.
Last updated: Nov 22 2024 at 16:03 UTC