Error message for obsoleted protocol composition syntax should say "types" instead of "protocols"

[AnthonyLatsis]

The following syntax error is emitted when the parser spots obsoleted protocol composition syntax:

typealias C = protocol<P, Q> // error: 'protocol<...>' composition syntax has been removed; join the protocols using '&'

The problem with this message is that "protocols" is not an accurate or prudent description of the components at this compilation stage:

• Protocol compositions are not limited to protocols.
• A component can be written in terms of a member type, such as a nested typealias.
• The parser has no knowledge of whether the components are legal, namely, whether they really reference protocols or classes.

A more neutral "types" is preferable.

[hborla]

Hmm, I'm not sure that "types" is the right word to use here. The things that are joined together with & in a composition are constraints. Compositions can be used for existential types, or for generic requirements on type parameters. We've been trying to guide people away from thinking of protocols and other constraints as types.

[AnthonyLatsis]

Good point, wording them as constraints (or type constraints per the language guide) sounds better.

[IceCurrent]

So we change the word "types" to "type constraints"?
Also is it the right way to do it, as I've done ? (because it is failing the smoke test)
@AnthonyLatsis @hborla

[AnthonyLatsis]

So we change the word "types" to "type constraints"?

Sounds good to me. It’s totally fine to do this in the same pull request btw, no need to close it.

Also is it the right way to do it, as I've done ? (because it is failing the smoke test)

Yeah, you simply forgot to update a test somewhere — check out the logs.

[Rajveer100]

Have done the necessary changes @AnthonyLatsis

[AnthonyLatsis]

Have you checked in with @IceCurrent? They might still be interested in wrapping this up.

[Rajveer100]

I didn't check with him, but looks like he closed his pull request around 2 months back and didn't revert, so I just did the changes.

[Rajveer100]

Also, I actually by mistake committed and pushed changes of this issue in the same branch of the previous pull request and since the branch is same, it's reflected in the previous issue too...
What's the best I can do to resolve this...

[AnthonyLatsis]

Soft-reset those changes, stash them, then unstash them in a new branch, then force-push the other branch.

[Rajveer100]

Thanks!

[Rajveer100]

@AnthonyLatsis I have committed the changes and also did the testing, can I open a pull request?

[AnthonyLatsis]

I cannot forbid you to, but I will suggest to at least try and reach out to the person that last worked in this. We do not want fixing issues to be a race between contributors (unless it is a race that has been agreed on 😅).

[Rajveer100]

Lol!

[Rajveer100]

Also, I see the commits of my ongoing PR in the new PR, is that normal to happen, because when I researched, it seemed that it happens a lot and it's a Git issue?

[AnthonyLatsis]

That means that your new branch branches off from the branch associated with the ongoing PR. You should set the new branch to track main instead and rebase.

[Rajveer100]

@AnthonyLatsis I have created a pull request, please review it.

[AnthonyLatsis]

@IceCurrent Are you interested in wrapping this up?