-
Notifications
You must be signed in to change notification settings - Fork 37
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update to wasmtime 3.0 and enable wasi-nn calls #209
Conversation
8c09cbb
to
245390e
Compare
27b3eaf
to
1a2b813
Compare
I think we'll want to guard the availability of wasi-nn behind a CLI flag, so that (as our testing tool) we don't create the expectation that wasi-nn is available in production |
There are now two flavors of the handle request call. One is used for trap_test and the other is used by the Viceroy framework. Also extended the call to bubble up the error so it could be easily included in the response body.
I have implemented enabling experimental modules via flags and refactored the trap test so that it uses the body instead of headers. |
Also switches backtrace output to stderr.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hey @geekbeast just one change needed for the error output and then a question on a thing that seemed turned off. This looks great otherwise. If there's a test you can do to show it working that would be great, but given that it's experimental support I'm not going to require it to approve!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for taking the time to get all these in!
The main purpose of this PR was to support wasi-nn calls inside Viceroy. As the necessary changes weren't available until the wasmtime 3.0 release, I also made the updates to to make it compatible with the broader changes in that specific release.
The two big changes that came in were changes to the pattern for PoolingAllocationConfig and how error/traps are handled. There was also one minor change in that it looks like the linker is no longer under the tokio namespace.