cargo publish --dry-run finishes very early

[JulianKnodt]

Problem

I wanted to know what I was missing before running cargo publish, but did not accidentally want to publish a crate. When I ran cargo publish --dry-run it terminated before the upload, but it does not actually reach the point where it checks the Cargo.toml for validation errors. Thus, it seems to me that --dry-run is kind of pointless.

Proposed Solution

Make --dry-run error out only right before a crate is actually published, rather than before upload.

Notes

I'd be willing to implement this, if it is acceptable.

[ehuss]

Can you say more about which validation errors did not seem to run?

If this is related to validation done on the server side, that would require implementing a new publishing API which is outlined in rust-lang/crates-io-cargo-teams#82. That is likely a large project that would require working mostly with the crates.io team, and possibly an RFC outlining the design.

[JulianKnodt]

Specifically it was:

Caused by:
  the remote server responded with an error: missing or empty metadata fields: description, license. Please see https://doc.rust-lang.org/cargo/reference/manifest.html for how to upload metadata

Just checked the other issues and I understand why this is not something that can be resolved client-side, but maybe it would be worth it to differentiate server-side and client-side validation, even if there is an API for doing so?

[weihanglo]

Thank you for the reply! crates.io is only one implementation of remote registries. Other registries may have different validations and requirements. Cargo at this moment sticks to a common minimal set of validations across different registries. I feel like it might break other registries publishing process for such kind of change.

Going to close this in favor of #5941, as more discussions happened there. If it is wrong, comment below and we can re-open this.