Assign HirIds in HIR traversal order

[cjgillot]

AST->HIR lowering assigns a HirId to most nodes in the HIR tree. The HirId consists of an OwnerId, which tells to which item (function, trait, associated function...) a node belongs to, and an ItemLocalId which is its index inside that OwnerId.

ItemLocalIds are assigned in the order they are created during lowering, by a simple counter. As a consequence, their values when looking at the HIR tree do not follow a consistent order. Making that order consistent may help simplifying some code that needs to visit HIR when it may just iterate on ItemLocalIds.

Steps:

• in rustc_passes::hir_id_validator, add a check that the ItemLocalId of the parent of each node has a lower value that the ItemLocalId of the node itself;
• change rustc_ast_lowering, in particular calls to lower_node_id and next_id until the tests pass.

Bonus: manage to get the order of ItemLocalIds be the order in which a HIR visitor (defined in rustc_hir::intravisit) visits all the HirIds.

I recommend using -Zunpretty=hir-tree to debug. Please contact me on zulip for any question.

[chenyukang]

@rustbot claim
I'm learning the HIR part, will take some time for this.

[ericmcbride]

@rustbot claim

[DavidCoy77]

Hi! I'd love to get started contributing to Rust, but I'd need a little help getting started since this would be my first ever GitHub contribution. You've included the tags Easy, Help-Wanted, and Mentor, so if you're willing to help, I'm eager to get to work. Thanks!

[the-shank]

This hasn't been assigned to anyone yet. @DavidCoy77 are you working on this? In case you are not working on this, I would like to pick this up.

[axelmagn]

@rustbot claim

Since there has been no activity on this since November, I am assuming that there is no work in-flight. Please let me know if that's not the case.

I am a first-time contributor looking for a good mentorship project, and this fits my interest area.

[axelmagn]

Dropping this issue after conversation with @cjgillot in favor of #99669 (higher priority)

@rustbot release-assignment

[the-shank]

@rustbot claim

@cjgillot I am a first-time contributor and would like to work on this.

[cjgillot]

Closing as not planned. It's really hard (= mostly rewrites lowering to change the id generation order), with very uncertain benefits (= simplify lint level computation).