[wasm-metadata] parse OCI author custom section #1925
Conversation
There was a problem hiding this comment.
Thanks for this! What I might recommend is to take the &[u8] entire wasm in memory as opposed to just a custom section. That way you can bake in how to handle a missing section, duplicated sections, the section name, etc. Internally you can use a BinaryReader to parse the binary data if necessary, but yeah the Parse-trait side of things is sort of there with FromReader but it's no the same as the encoding side of things.
yoshuawuyts
force-pushed
the
refactor-registry-metadata
branch
from
801338f to
06f9aa0
Compare
| rewrite_wasm( | ||
| &self.name, | ||
| &Producers::from_meta(self), | ||
| &self.author, | ||
| self.registry_metadata.as_ref(), | ||
| input, | ||
| ) |
There was a problem hiding this comment.
The number of function arguments will continue to increase until morale improves.
More seriously though: once I finish adding the custom sections, I plan to refactor the way we handle this. I suspect we can combine rewrite_wasm, Metadata, and AddMetadata into a unified structure, allowing us to tidy this up.
This PR adds support for parsing the
authorcustom section, implementing parts of #1922. Thanks!Draft Notes
I'm filing this PR early and as a draft because I'm not 100% sure I'm following best practices here. It's my first time using
wasmparserandwasm-encoder, and I'm for example not sure whetherparse_wasmshould take aBinaryReaderorCustomSectionReader.Also: I was looking for something like a
Parsetrait, but I don't thinkwasmparserhas one? I quite liked the encoder side of things.Finally: in this PR I'm basically just wrapping
CustomSectionin a newtype - however that means storing the internal data as a byte array rather than a string. How do we feel about that?I'm asking these questions since I'm planning to repeat this pattern a number of times for the other custom sections, and it seems worth making sure it's early on. Thanks!