bevy_crevice derive macro doesn't work properly in external crate

[alice-i-cecile]

Bevy version

main

Proplem

The bevy_crevice_path macro returns the render_resource path segment two times if one uses the bevy_derive macros in a local project.

Proposed patch

diff --git a/crates/bevy_crevice/bevy-crevice-derive/src/lib.rs b/crates/bevy_crevice/bevy-crevice-derive/src/lib.rs
index e0cc77a8..421c4aa9 100644
--- a/crates/bevy_crevice/bevy-crevice-derive/src/lib.rs
+++ b/crates/bevy_crevice/bevy-crevice-derive/src/lib.rs

@@ -47,14 +47,17 @@ fn bevy_crevice_path() -> Path {
                 segments,
             }
         })
-        .or_else(|| bevy_manifest.maybe_get_path(crate::BEVY_RENDER))
-        .map(|bevy_render_path| {
-            let mut segments = bevy_render_path.segments;
-            segments.push(BevyManifest::parse_str("render_resource"));
-            Path {
-                leading_colon: None,
-                segments,
-            }
+        .or_else(|| {
+            bevy_manifest
+                .maybe_get_path(crate::BEVY_RENDER)
+                .map(|bevy_render_path| {
+                    let mut segments = bevy_render_path.segments;
+                    segments.push(BevyManifest::parse_str("render_resource"));
+                    Path {
+                        leading_colon: None,
+                        segments,
+                    }
+                })
         })
         .unwrap_or_else(|| bevy_manifest.get_path(crate::BEVY_CREVICE))
 }

Additional information

Submitted on behalf of a user on Discord.

[DJMcNab]

@alice-i-cecile I know you're the ticketmain, but you are also allowed to create pull requests

That would make this proposed patch easier to review

Fwiw, I'm surprised by this patch; I'd expect the fix to be making the bevy::render::render_resources which the map is operating on be just bevy::render

But again, I can't review it easily on mobile, so it's challenging to check my assumptions there.

[DJMcNab]

As I mentioned on discord (in the linked thread) the correct fix is to just delete line 44 (I think), as explained above.

I'm not too bothered about attribution for that fix, but can create a pr if no-one else does before I get around to it.

[mockersf]

oh I was testing that fix locally too @DJMcNab 👍