Cargo build, cargo install: very different results

I have taken a multi-year Rust hiatus, but I am trying again to use it.

I have a Rust application that I build and install with

cargo install --path . --force

when cd-ed to the application's directory. This results in the application building, installing and running correctly.

But now I want to run clippy. Again, cd-ed to the application's directory,



   Compiling gtk-sys v0.17.0
   Compiling glib-sys v0.15.10
    Checking futures-util v0.3.1
error[E0432]: unresolved import `futures_macro::_proc_macro_hack_join`
  --> /home/dca/.cargo/registry/src/
78 |     pub use futures_macro::join;
   |             ^^^^^^^^^^^^^^^^^^^ no `_proc_macro_hack_join` in the root

error[E0432]: unresolved import `futures_macro::_proc_macro_hack_try_join`
  --> /home/dca/.cargo/registry/src/
81 |     pub use futures_macro::try_join;
   |             ^^^^^^^^^^^^^^^^^^^^^^^ no `_proc_macro_hack_try_join` in the root

error[E0432]: unresolved import `futures_macro::_proc_macro_hack_select`
   --> /home/dca/.cargo/registry/src/
162 |     pub use futures_macro::select;
    |             ^^^^^^^^^^^^^^^^^^^^^ no `_proc_macro_hack_select` in the root

   Compiling thiserror v1.0.30
   Compiling proc-macro-crate v1.1.3
For more information about this error, try `rustc --explain E0432`.
error: could not compile `futures-util` due to 3 previous errors

Hmm. So I tried

cargo build

Same error.

Noting that cargo install does a --release build, I tried

cargo build --release

Same error.

I then sent the output of cargo build --release and cargo install (both done after cargo clean) to files, which I then sorted, because I noted that things get compiled in different orders in the two. Diff-ing the sorted output files reveals that the versions of many of the dependencies are different in the two, those in the (error-free) cargo install run being later (higher versions).

So while my application works, my problem is that

a. I can't use clippy because of this issue.
b. Why do these two commands that I think can reasonably expected to do the same build actually do very different builds, with different dependency versions?

I should add that none of the version differences involve direct dependencies of my application, which I specify very tightly in my Cargo.toml file.

Any insights into what's happening here would be appreciated.

cargo install doesn't use any existing Cargo.lock unless you force it --locked. Since the newer versions are working, cargo update should solve your local build.

Indeed it does. Thank you for the help.

This topic was automatically closed 90 days after the last reply. We invite you to open a new topic if you have further questions or comments.