libp2p_swarm_derive: Add crate parameter to configure libp2p crate name
#3006
Conversation
70604ae to
ed03603
Compare
ed03603 to
3e96486
Compare
3e96486 to
11d8ee8
Compare
crate parameter to configure libp2p crate name
| let user_provided_libp2p_prefix = ast | ||
| .attrs | ||
| .iter() | ||
| .filter_map(get_meta_items) | ||
| .flatten() | ||
| .filter_map(|meta_item| { | ||
| if let syn::NestedMeta::Meta(syn::Meta::NameValue(ref m)) = meta_item { | ||
| if m.path.is_ident("crate") { | ||
| if let syn::Lit::Str(ref s) = m.lit { | ||
| return Some(syn::parse_str::<syn::TypePath>(&s.value()).unwrap()); | ||
| } | ||
| } | ||
| } | ||
| None | ||
| }) | ||
| .next(); |
There was a problem hiding this comment.
This is mostly the same as the logic for out_event, right? What do you think of moving it into a function called in both places?
| #[test] | ||
| fn custom_crate() { | ||
| use libp2p as mylibp2p; | ||
| #[allow(dead_code)] | ||
| #[derive(NetworkBehaviour)] | ||
| #[behaviour(crate = "mylibp2p")] | ||
| struct Foo { | ||
| ping: ping::Behaviour, | ||
| identify: identify::Behaviour, | ||
| } | ||
|
|
||
| #[allow(dead_code)] | ||
| fn foo() { | ||
| require_net_behaviour::<Foo>(); | ||
| } | ||
| } |
There was a problem hiding this comment.
This test also already passes with current master (if you remove line 141).
Correct me if I am wrong, but I believe the current ::libp2p prefix imports libp2p from the crate level, which means that aliases in a module should not be an issue? @mxinden?
I think it is only a problem if the crate is imported in the Cargo.toml with a different name.
serde on the other hand states for their crate attribute that:
This is normally only applicable when invoking re-exported Serde derives from a public macro in a different crate.
I am not against adding this parameter, but could we add a test that fails on current master and would be fixed with this? 🙂
There was a problem hiding this comment.
As far as I understand, the issue is not clashing symbols but renaming the crate in your manifest.
To test that, we'd have to make a separate test crate with a renamed import.
This make me wonder, should we perhaps just require certain symbols to be present for the macro to work?
That would make #3023 easier to implement.
Or perhaps we could make a seperate "derive" prelude? Like libp2p::swarm_derive_prelude that includes all the required symbols. Then we can add a configuration knob here that allows you to change where that prelude is coming from.
There was a problem hiding this comment.
This make me wonder, should we perhaps just require certain symbols to be present for the macro to work?
That would make #3023 easier to implement.
Or perhaps we could make a seperate "derive" prelude? Like
libp2p::swarm_derive_preludethat includes all the required symbols. Then we can add a configuration knob here that allows you to change where that prelude is coming from.
Actually, don't worry about it, I already have a solution to that.
|
If we end up shipping #3055, then this PR is obsolete as the #[behaviour(prelude = "mylibp2p::swarm::derive_prelude") |
|
Closing in favor of #3055 which is about to be merged. |
Description
When importing libp2p with a use statement and a custom alias name, the code generated by the NetworkBehaviour derive macro does not use this alias and does not compile. This PR enhances the derive macro with a new parameter "crate" that can be used to specify the custom alias name.
Open Questions
None
Change checklist