feat: port snapshot_from_lockfile from denoland/deno
#27
Conversation
| "once_cell", | ||
| "spin", | ||
| "untrusted", | ||
| "web-sys", |
There was a problem hiding this comment.
I don't get why they have web-sys here.
There was a problem hiding this comment.
So it seems like deno_lockfile depends on ring, which in turn depends on web-sys (but this isn't included in the compilation, just included in the lock file). briansmith/ring#904 is talking about it
src/resolution/snapshot.rs
Outdated
| } | ||
|
|
||
| // clear the cache to reduce space for cache | ||
| api.clear_cache(); |
There was a problem hiding this comment.
I think we can remove this from here and instead call it after running this function in the CLI
There was a problem hiding this comment.
Sounds good, now clear_cache method is removed from the trait, and it's moved to the CLI code: https://github.com/denoland/deno/pull/20024/files#diff-747e5bf0658bef1ca03cfa6b83c21b7ffc899ce4324557b2cc94356681e76699R841
src/registry.rs
Outdated
| /// A trait for getting package information from the npm registry, with no | ||
| /// internal caching mechanism. | ||
| #[async_trait] | ||
| pub trait NpmRegistryApiWithoutCache: Sync + Send { |
There was a problem hiding this comment.
Perhaps we could just have NpmRegistryApi to have default implementations that don't offer caching so we don't need an extra trait?
| /// The reason why this function is not combined with [`snapshot_from_lockfile`] | ||
| /// is because we want `lockfile` to not live across the `.await` points. This | ||
| /// becomes problematic if `lockfile` is wrapped in [`std::sync::Mutex`] or | ||
| /// similar, which doesn't implement [`Send`]. |
dsherret
changed the title
snapshot_from_lockfile from denoland/denosnapshot_from_lockfile from denoland/deno
This commit brings
snapshot_from_lockfilefunction todeno_npmfrom the maindenorepository.denoland/deno#20024 shows how the caller side will change due to this port.