Flag that the development page's PHP row predates its audit - #167
Merged
Conversation
The page reports PHP as ten shared cases with nine exact for both sides and none exact for the language server alone, which reads as Bifrost satisfying every PHP case in the 24 July run. That cannot describe the corpus as it stands. The PHP ground-truth audit landed on 22 July, two days before that run, and rewrote php-property-access and php-parity-static-property-access to expect sigil-inclusive ranges. Every Bifrost version measured since fails at least one of them: both under v0.10.1, and the static-property case alone under v0.10.5, which corrected the declaration side but not the static usage side. So the row describes a corpus from before the audit, or a shared denominator that dropped those two cases for a reason the page does not give. It is not evidence that Bifrost passed them. This is not hypothetical. Reading that row as current evidence produced a false regression hypothesis that took a source-level comparison across two Bifrost tags to disprove, and a round trip to another repository. A caution on the page is cheaper than the next person repeating it. The numbers are left untouched. The page is historical evidence and the point is that its provenance is unclear, not that the figures should be restated. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes #157.
The page reports PHP as 10 shared / 9 both exact / 0 exact-for-LSP-only, which
reads as Bifrost satisfying every PHP case in the corrected 24 July run. That
cannot describe the corpus as it stands.
The PHP ground-truth audit (
f19c801) landed 22 July, two days before that run,and rewrote
php-property-accessandphp-parity-static-property-accesstoexpect sigil-inclusive ranges. Every Bifrost measured since fails at least one:
php-property-accessphp-parity-static-property-accessSo the row reflects a pre-audit corpus, or a shared denominator that excluded
those two for a reason the page does not record.
Why bother
Reading that row as current evidence is what produced the false "PHP
regression between v0.8.8 and v0.10.1" hypothesis — disproved only by checking
visit_property_declarationat both tags, after a round trip to anotherrepository. A caution costs less than the next person repeating it.
The numbers are untouched. This is historical evidence; the point is that its
provenance is unclear, not that the figures should be restated.
links:checkpasses 22 routes.🤖 Generated with Claude Code