Fail CI if lock file isn't updated.

[nihohit]

No description provided.

[jaymell]

I'm not sure I fully understand what --locked does in the cargo test context. Should we consider doing this in the other runs as well? If we don't, are we still going to potentially hit the same compatibility issues with older Rust versions?

[nihohit]

https://doc.rust-lang.org/cargo/commands/cargo-build.html#manifest-options

--frozen
--locked
Either of these flags requires that the Cargo.lock file is up-to-date. If the lock file is missing, or it needs to be updated, Cargo will exit with an error. The --frozen flag also prevents Cargo from attempting to access the network to determine if it is out-of-date.
These may be used in environments where you want to assert that the Cargo.lock file is up-to-date (such as a CI build) or want to avoid network access.

It does the same thing in cargo test as in cargo build - fail the build if the lock file is out of date.
I thought it makes sense to put it in the tests, since that will run on the oldest Rust version. We can move it to a separate step, if you prefer it out of the makefile.

Should we consider doing this in the other runs as well?

Not sure which runs are you referring to. In the makefile we hit 8 out of 9 CI workflows. Would you like this also in the lint workflow? That workflow is only tested on the most recent Rust, AFAIK.

are we still going to potentially hit the same compatibility issues with older Rust versions?

I don't think so, since this makefile is called from the workflows with the older rust version matrix.

[jaymell]

Sorry, I mean the specific test invocations within the Makefile. The first one -- "Testing Connection Type TCP without features" -- doesn't test any of the features. I was thinking that we're more likely to have issues w/ the subsequent invocations that explicitly use --all-features and bring in all the additional dependencies. But maybe it doesn't matter?

[nihohit]

no, that's a good point. Maybe we'll just start with cargo build --locked --all-features?

[jaymell]

Sounds good to me 👍

[nihohit]

@jaymell round

[jaymell]

Again, this could just be flawed understanding, but if we don't run each iteration of the tests in the Makefile with --locked, are we not still likely to encounter an issue related to unlocked dependency versions bumping MSRV?

EDIT: To be clear, what I'm asking is, should we add --locked to each test iteration in the Makefile?

[nihohit]

I'm not sure that it's possible, but adding the flag doesn't cost us money, so why not?