feat(shim): support archive layers #121
Conversation
vdice
force-pushed
the
feat/archive-layers
branch
5 times, most recently
from
4a24750
to
acb11b9
Compare
vdice
force-pushed
the
feat/archive-layers
branch
4 times, most recently
from
1d14ea7
to
42df2a3
Compare
vdice
force-pushed
the
feat/archive-layers
branch
from
42df2a3
to
4a5e1da
Compare
vdice
force-pushed
the
feat/archive-layers
branch
from
4a5e1da
to
3f3d291
Compare
|
In 8e83775 I added a new |
vdice
force-pushed
the
feat/archive-layers
branch
5 times, most recently
from
2a331be
to
8e83775
Compare
Signed-off-by: Vaughn Dice <vaughn.dice@fermyon.com>
… testing Signed-off-by: Vaughn Dice <vaughn.dice@fermyon.com>
vdice
force-pushed
the
feat/archive-layers
branch
from
8e83775
to
bfae3fb
Compare
vdice
requested review from
Mossaka,
kate-goldenring,
radu-matei and
devigned
as code owners
| @@ -40,10 +40,6 @@ const SPIN_HTTP_LISTEN_ADDR_ENV: &str = "SPIN_HTTP_LISTEN_ADDR"; | |||
| const RUNTIME_CONFIG_PATH: &str = "/runtime-config.toml"; | |||
| /// Describes an OCI layer with Wasm content | |||
| const OCI_LAYER_MEDIA_TYPE_WASM: &str = "application/vnd.wasm.content.layer.v1+wasm"; | |||
There was a problem hiding this comment.
suggestion: make spin_oci::client::WASM_LAYER_MEDIA_TYPE public and use it in the Spin Shim.
There was a problem hiding this comment.
In fact, Spin itself is waiting on the canonical type/string upstream: https://github.com/fermyon/spin/blob/main/crates/oci/src/client.rs#L34-L35. To be defined in https://github.com/bytecodealliance/registry perhaps? Therefore expecting both Spin and the shim would pull this from the upstream crate.
| // which will fall back to /tmp if TMPDIR is not set. /tmp is either not found or not accessible | ||
| // in the shim environment, hence setting to current working directory. | ||
| if env::var("TMPDIR").is_err() { | ||
| log::debug!( |
There was a problem hiding this comment.
When do we expect TMPDIR to be set? I am wondering if this log is necessary.
There was a problem hiding this comment.
Good point... it appears it isn't ever set, currently. Glad to remove this debug log if you'd like.
There was a problem hiding this comment.
I think we would want to leave it, otherwise we might overwrite it. I don't know what scenario it would be set in, but if it was set and we overwrite I don't think that would be an expected behavior?
|
Thanks, all. Please merge when ready. |
Closes #28
Depends on fermyon/spin#2523
Draft for now as 2523 is based on Spin's main branch which has since bumped wasmtime to 21.0.1. Can either wait for that same bump to occur in this project (presumably in tandem with pinning to the next Spin release with said version) or can explore backporting 2523 to the v2.5 branch and cutting a patch release with it.
TODO