Add a Vec<Span> to YaccErrorKind::Duplicate* variants

[ratmice]

It seems like it is best to add other YaccParserErrorKind variants for a nice clean history/merge. (At least as long the variant isn't turning into a any kind of grand adventure).

• [X] DuplicateRules https://github.com/softdevteam/grmtools/commit/36534ca45807579adbc2907bf0aee6705dd3e1dc (In a previous PR #307),
• [X] DuplicateAvoidInsert
• [X] DuplicatePrecedence,
• [X] DuplicateImplicitTokensDeclaration,
• [X] DuplicateExpectDeclaration,
• [X] DuplicateExpectRRDeclaration,
• [X] DuplicateStartDeclaration,
• [X] DuplicateActiontypeDeclaration,
• [X] DuplicateEPP,

[ltratt]

Please squash.

[ratmice]

Let me know if you want me to keep adding variants to this series for less merge commits, or if we want to just merge now. Otherwise it is squashed.

[ltratt]

Ah there are other places where we need to do the same thing? Then let's add extra commits here -- it makes sense to do it in one PR.

[ratmice]

Will do, I'll just add a task list to the initial pull-request and perhaps rename the PR.

[ltratt]

Looks good to me. Are we ready for squashing?

[ratmice]

About half-way done with the variants, but I think we are good with these. I wouldn't mind reordering/squashing these commits, then updating the top comment.

[ltratt]

Let's keep going with the commits as-is, then we can do squashing at the end.

[ltratt]

Works for me! Any duplicate-possibilities left?

[ratmice]

Just 3 more, DuplicateStart, DuplicateActionTypeDeclaration, and DuplicateEPP. After that I think it's just lrlex::LexErrorKind::DuplicateName, not sure if we want to handle that here, or in another PR.

Start is taking a bit because of its pervasiveness across the library/testsuite.

[ltratt]

Agreed: lrlex is probably best left to another PR.

[ratmice]

Alright, well that is it I believe. At least I am not aware that the others can return multiple locations, but I haven't done a thorough examination there.

The last one, I ended up adding a second span, just because I thought it would be really confusing if I were to put a tuple of String and Span together if the Span was not actually for the String, but the key.

Edit: There are a few issues in the commit messages, So i'll edit those when the time comes.

[ltratt]

Please squash.

[ratmice]

Squashed, but do let me know if you prefer it squashed further into one commit. Edit: and I'm super confused, because these 11 commits don't look at all like the 8 commits i'm staring at locally.

[ltratt]

It looks like they're correct now!

bors r+

[ratmice]

My fault pushed the wrong branch accidentally, fixed.

[bors[bot]]

Build succeeded:

• buildbot/buildbot-build-script