Codecov Report

Patch coverage: 100.00% and project coverage change: +7.77 🎉

Comparison is base (04567e9) 39.94% compared to head (1bee6fd) 47.72%.

❗ Current head 1bee6fd differs from pull request most recent head 54634d3. Consider uploading reports for the commit 54634d3 to get more accurate results

@@            Coverage Diff             @@
##            trunk      #41      +/-   ##
==========================================
+ Coverage   39.94%   47.72%   +7.77%     
==========================================
  Files          19       19              
  Lines        1860     1869       +9     
==========================================
+ Hits          743      892     +149     
+ Misses       1117      977     -140     

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

Anything else you want to make a PR of or shall I release 0.0.12 with these changes?

This is no longer adding dependent lifetimes, that are needed at least in my case. The docs have

The lifetimes passed to this function will automatically depend on any other lifetime this struct or enum may have. Example:

The struct is struct Foo<'a> {}
You call `generator.impl_for_with_lifetime(“Bar”, &[“b”])
The code will be impl<'a, 'b: 'a> Bar<'b> for Foo<'a> {}

https://docs.rs/virtue/latest/virtue/generate/struct.Generator.html#method.impl_for_with_lifetimes

From my example use, the diff after this change is:

-impl < 'de : 'a, 'a > crate :: sshwire :: SSHDecode < 'de > for UserauthPwChangeReq < 'a >
+impl < 'de, 'a > crate :: sshwire :: SSHDecode < 'de > for UserauthPwChangeReq < 'a >

(that's for https://github.com/mkj/sunset/blob/c97ff133d51ae482cbb61ae355dafdaa08dd3a34/sshwire-derive/src/lib.rs#L392 targetting struct https://github.com/mkj/sunset/blob/c97ff133d51ae482cbb61ae355dafdaa08dd3a34/src/packets.rs#L164)

I ran into this in bincode as well. I solved it with

generator.impl_for_with_lifetimes(format!("{}::BorrowDecode", crate_name), ["__de"])
    .modify_generic_constraints(|generics, where_constraints| {
        for lt in generics.iter_lifetimes() {
            where_constraints.push_parsed_constraint(format!("'__de: '{}", lt.ident))?;
        }
        Ok(())
    })?

I'm thinking the new behavior is correct because not everyone will want this behavior, but maybe we can add a helper method that does this?

Thanks, that works for me. I agree it shouldn't be the default. Maybe just adding an example in the impl_for_with_lifetimes() documentation would be enough.

Opened https://github.com/bincode-org/virtue/issues/51