[4/n] add stricter typing in JsonPathStack#15
Merged
sunshowers merged 11 commits intoMay 12, 2026
Conversation
Created using spr 1.3.6-beta.1
Created using spr 1.3.6-beta.1 [skip ci]
sunshowers
changed the title
Created using spr 1.3.6-beta.1 [skip ci]
Created using spr 1.3.6-beta.1
ahl
approved these changes
May 2, 2026
Collaborator
ahl
left a comment
ahl
left a comment
There was a problem hiding this comment.
Didn't look very carefully; please let me know if something needs particular scrutiny
sunshowers
commented
May 5, 2026
sunshowers
commented
May 5, 2026
sunshowers
changed the base branch from
sunshowers/spr/main.2n-add-stricter-typing-in-jsonpathstack
to
main
Created using spr 1.3.6-beta.1
sunshowers
added a commit
that referenced
this pull request
May 5, 2026
Detect cycles by checking for path segments (with a trailing `/`) rather than just using a raw string prefix. We hit this in Omicron where `AuditLogEntry` was incorrectly detected as a prefix of `AuditLogEntryActor`, hiding incompatible changes. This is a minimal fix that (I believe) is correct given the current structure. I'm reworking this in #15 to have stronger types so we can think about this in a more principled manner.
Created using spr 1.3.6-beta.1 [skip ci]
Created using spr 1.3.6-beta.1
sunshowers
changed the base branch from
main
to
sunshowers/spr/main.4n-add-stricter-typing-in-jsonpathstack
sunshowers
changed the title
sunshowers
changed the base branch from
sunshowers/spr/main.4n-add-stricter-typing-in-jsonpathstack
to
main
sunshowers
changed the base branch from
main
to
sunshowers/spr/main.4n-add-stricter-typing-in-jsonpathstack
Created using spr 1.3.6-beta.1 [skip ci]
Created using spr 1.3.6-beta.1
sunshowers
commented
May 8, 2026
Comment on lines
+69
to
+81
| /// Lifecycle of a schema comparison, keyed by `VisitedKey` in | ||
| /// [`Compare::visit_state`]. | ||
| #[derive(Clone, Debug)] | ||
| pub(crate) enum VisitState { | ||
| /// Comparison is in progress somewhere up the call stack. Re-entering a | ||
| /// key in this state means the schema graph has a cycle. | ||
| Visiting, | ||
| /// Comparison has completed. | ||
| Completed { | ||
| /// Whether the two schemas were determined to be equal. | ||
| equal: bool, | ||
| }, | ||
| } |
Contributor
Author
There was a problem hiding this comment.
In this PR I replaced the existing cycle detection check, which relied on looking at the stack of schemas, with a more explicit DFS postorder check.
This is equivalent to the old code:
if old_schema.context().stack().contains_cycle()
&& new_schema.context().stack().contains_cycle()
{
return Ok(true);
}Why? Basically because having a prefix in the stack is equivalent to being in the visiting state while performing the DFS. The tests added in #23 don't have any output differences, which is a reasonable sign that there aren't any behavior changes here.
I like how explicit the postorder check is, and it also fits quite nicely into visit_state.
sunshowers
commented
May 8, 2026
Comment on lines
+205
to
+217
| // Returning `true` on `Visiting` is sound only because the compare | ||
| // methods call `schema_push_change` eagerly on detecting a change | ||
| // rather than making decisions based on the value returned from this | ||
| // function. If that weren't the case -- for example, if there were | ||
| // code which did something like: | ||
| // | ||
| // let inner_eq = self.compare_schema(...)?; | ||
| // if !inner_eq { | ||
| // self.schema_push_change(...); | ||
| // } | ||
| // | ||
| // Then, relying on `true` here would result in a false negative. | ||
| // This invariant must be upheld by all compare methods. |
Contributor
Author
There was a problem hiding this comment.
Note this block -- the old code also returned true, which is correct, but I wanted to reason about this more clearly.
sunshowers
added a commit
that referenced
this pull request
May 12, 2026
Rename `tests/cases/cycle-shared-prefix` to `tests/cases/cycle-detection` to reflect the broader scope, and extend the fixture with a few different kinds of cycles: - mutual recursion through `$ref` - three-way cycle - mutual recursion through `allOf` wrappers - asymmetric tests where the old and new schemas have different shapes Also add tests to verify outputs. (We're going to rework cycle detection a bit in #15 and want to ensure that PR doesn't introduce any behavior changes.)
sunshowers
changed the base branch from
sunshowers/spr/main.4n-add-stricter-typing-in-jsonpathstack
to
main
Created using spr 1.3.6-beta.1
sunshowers
deleted the
sunshowers/spr/2n-add-stricter-typing-in-jsonpathstack
branch
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A
JsonPathStackalways has an endpoint at the bottom and a series of schemas above that. We're going to rely on this in upcoming work -- express this constraint in the type system.