Historical v0.2.0 evaluation cases

Evaluation partition only. These cases come from the immutable v0.2.0 release and are never pooled with development cases.

Exact means the tool returned the complete reviewed location set with exact token ranges. A disagreement is a measured contract result, not automatically a defect verdict; the reviewed source contract remains the reference.

Strict-contract overview

Reference profile	Shared	Both exact	Bifrost only	Reference only	Neither
gopls 0.23.0	12	7	1	2	2
Pyright 1.1.411	12	6	1	4	1
TypeScript language server 5.3.0	12	5	5	2	0
Total	36	18	7	8	3

Exact only for Bifrost

Reference profile	Case file	Case
gopls	go-02.yaml	real-project-v1-go-02-2
Pyright	python-03.yaml	real-project-v1-python-03-1
TypeScript language server	typescript-01.yaml	real-project-v1-typescript-01-2
TypeScript language server	typescript-02.yaml	real-project-v1-typescript-02-1
TypeScript language server	typescript-04.yaml	real-project-v1-typescript-04-1
TypeScript language server	typescript-04.yaml	real-project-v1-typescript-04-2
TypeScript language server	typescript-04.yaml	real-project-v1-typescript-04-3

Exact only for the reference server

These eight cases are the clearest current Bifrost parity backlog within this evaluation slice.

Reference profile	Case file	Case
gopls	go-01.yaml	real-project-v1-go-01-1
gopls	go-02.yaml	real-project-v1-go-02-3
Pyright	python-01.yaml	real-project-v1-python-01-1
Pyright	python-03.yaml	real-project-v1-python-03-3
Pyright	python-04.yaml	real-project-v1-python-04-2
Pyright	python-04.yaml	real-project-v1-python-04-3
TypeScript language server	typescript-01.yaml	real-project-v1-typescript-01-1
TypeScript language server	typescript-02.yaml	real-project-v1-typescript-02-3

Neither exact

Three cases are non-exact for both Bifrost and the corresponding reference server. They remain visible in the raw reports and count in the denominator; they are not silently dropped from the comparison.

The broader historical development case comparison covers ten language-server profiles and 131 shared cases. It is useful for regression history but has a different selection and review boundary.