feat: introduce ID crate

[daniel-noland]

See docs in crate for details

I hope to make this useful in other crates shortly.

[qmonnet]

Suggested change

	/// if you need to use `MySpecialType` instead of [`Uuid`] for your special type of tagged type.
	/// </div>
	/// if you need to use `MySpecialType` instead of [`Uuid`] for your special type of tagged type.
	/// </div>

https://doc.rust-lang.org/rustdoc/how-to-write-documentation.html?highlight=warning#adding-a-warning-block:

Please note that if you want to put markdown in the HTML tag and for it to be interpreted as such, you need to have an empty line between the HTML tags and your markdown content.

[qmonnet]

For unknown reasons the doctests crash on the id crate in CI.

I don't have time to debug at the moment.

I note we don't run doctests anywhere else at the moment; dataplane's pipeline module has examples, but the package is a binary and the module is not public, so its examples aren't run as part of the doctests (rust-lang/rust#50784).

The error says:

Couldn't run the test: Permission denied (os error 13) - maybe your tempdir is mounted with noexec?

So maybe we should try removing the noexec in the justfile? I think you had that in a PR recently, didn't you? Or was it a previous iteration of this one, maybe?

[qmonnet]

The code looks OK to me, but could you provide more details on the use case you have in mind, please?

[mvachhar]

The code looks OK to me, but could you provide more details on the use case you have in mind, please?

You can take a look at #320 where I will move from my internal version of this to a similar construct. But perhaps a better description in the PR is warranted.

[daniel-noland]

The code looks OK to me, but could you provide more details on the use case you have in mind, please?

Sure.

The motivation is to distinguish different types of IDs which we expect to use in configuration.

I am confident we will end up with multiple tables of configuration data, and keeping the IDs of VPCs compile time distinct from the IDs of nat rules can make things easier and less error-prone. A function which returns or accepts Id<Vpc> is clearer than one which just returns Id. Id of what?

[mvachhar]

Let's give a better PR message for motivation and then I am good to merge this. I'll probably add some stuff to this crate for mapping IDs to human readable strings.

[qmonnet]

OK sounds good, thanks a lot!

[qmonnet]

We'll need to take a look into these doctests issue at some point though, or we won't be able to test examples in the docs 😢

[mvachhar]

The motivation is to distinguish different types of IDs which we expect to use in configuration.

I am confident we will end up with multiple tables of configuration data, and keeping the IDs of VPCs compile time distinct from the IDs of nat rules can make things easier and less error-prone. A function which returns or accepts Id<Vpc> is clearer than one which just returns Id. Id of what?

In the future, we can put the motivation in the commit log. I don't think it is worth holding up this PR just for that though. Something to keep in mind for the future.