Improved error message when no baseline version found on crates.io

[era]

Fixes #183.

The new error message now is:

Error: tinylang not found in registry. The registry is crates.io. Crates published on a custom registry cannot be checked using default settings, see https://github.com/obi1kenobi/cargo-semver-checks/issues/166

Where tinylang is the crate I tested the program.

Let me know how you want to change the error message :)

[obi1kenobi]

Neat, thanks for the PR! The new message is certainly a lot better, and merging it would already be an improvement.

In case you're interested in doing one more round of polish, I wanted to suggest a couple of options for further improvements if you'd be open to it:

• The repetition of "registry" in the beginning is a bit verbose. Concise errors are best. Can we shorten it to Error: tinylang not found in registry (crates.io) or similar?
• Do you think linking to the FAQ section on checking non-crates.io crates might be a better option?
• It's possible at this point in the code that the user didn't actually use default settings, and in that case the reference to default settings would be confusing. The underlying problem is that we're trying to look up a crate version in the registry, and failing since the registry doesn't know about the crate. Perhaps we can reuse some language from the FAQ entry and nudge the user toward the workaround of supplying a baseline themselves (and not just a baseline version number).

Don't worry about the nightly failure, that's a "me" problem and not anything you did. I'm on it.

Also, might you be interested in tackling some more issues in the project? If you'd like, I'd be happy to act as concierge and suggest issues if you let me know what sorts of things you find interesting.

[era]

Hi @obi1kenobi , thank you for the quick review.

I tried to reduce the message in this new revision, I hope I was able to still pass the right message. Since the problem is not necessarily the usage of a custom registry, but also the possibility of the user not even publishing on one I tried to follow your advice and be a bit more generic towards the root cause.

I think the FAQ makes sense, the person facing the error will probably go to the GitHub issue and follow the link to the FAQ anyway, so maybe this way we can optimize it a bit.

Let me know if you want me to rephrase something on the error message.

Also, might you be interested in tackling some more issues in the project? If you'd like, I'd be happy to act as concierge and suggest issues if you let me know what sorts of things you find interesting.

I would love to :), I just don't want to delay any important feature, so any task that is not high-priority and I can take my time to work on it would be perfect.

[obi1kenobi]

Awesome, thanks for the help! Depending on your interests, there are at least two flavors of issues to choose from.

One set is issues that directly improve the CLI and ergonomics of using the tool. A few example issues:

• Unsatisfactory UX when items are no longer importable by the old path #35
• cargo semver-checks might exit 0 if it finds only publish = false crates in a workspace #424
• Running cargo semver-checks with no other arguments should default to checking #414
• -v does not include enough info to debug failures #325
• Displayed timing is unclear and only accounts for rule-checking time #298

Another direction is improving our checking capabilities, like adding more lints and improving existing ones. For example:

• Confusing method_parameter_count_changed error message (per_result_error_template) #430
• Missing path in enum_variant_added lint's per_result_error_message. #431
• New schema and lints for const and static values #366, or any other subpart of Tracking issue: Required schema additions to make new lints possible #241
• New lint: trait removed supertrait #232

Sometimes, improving our checking capabilities requires making changes in the trustfall-rustdoc-adapter crate as well. I'd be happy to help you get onboarded — it isn't nearly as hard as it looks — and you could make a seriously valuable contribution by writing up onboarding docs we could use for future contributors.

None of these are particularly time-sensitive, so don't worry about taking your time. We're all volunteers here so things get done when they get done :)