Documentation
¶
Overview ¶
Coverage overlay for btest.
G1: runCoverage runs "go test -coverprofile=.blit/coverage.out ./..." and
parses the resulting coverage profile.
G2: renderCoveragePanel prints a summary to the terminal: total %, top 5
least-covered files, and a mini histogram of per-file coverage buckets.
G3: readCoverage reads the most recent .blit/coverage.out without re-running
tests and renders the same panel.
G4: Both render paths dogfood the charts subpackage (Bar histogram + Gauge).
Command blit is a thin wrapper around `go test` for blit-powered test suites. It adds flags that map to blit features: -update to regenerate snapshots, -junit/-html to emit reports, -filter to pick specific tests, -parallel to set parallelism, and -watch to re-run on file changes.
Subcommands:
blit diff <testname> show the failure diff for a named test blit review [root] interactive review of pending .golden.new snapshots
Usage:
blit [flags] [packages...] blit record <name> -- <command> [args...] blit replay [--speed 1x] <name> blit review [root]
Packages default to "./..." when none are provided. The default reporter is the vitest-style runner already wired into the test code.
Examples:
blit # go test ./... blit -filter TestHarness ./btest/... # run tests matching TestHarness blit -update ./btest/... # regenerate snapshots blit -junit out/junit.xml -parallel 4 # parallel run + junit output blit -watch # re-run on file changes (1s poll) blit diff TestFoo # show diff for TestFoo failure blit record dashboard -- ./bin/dashboard blit replay dashboard --speed 2x
review.go implements `blit review` — an interactive TUI that presents a queue of pending .golden.new snapshot diffs so the developer can accept or reject each one without leaving the terminal.
Key bindings:
a accept current item (rename .golden.new → .golden atomically) r reject current item (delete .golden.new) s skip current item (move to next without changing disk) n next item p prev item q quit
Package main — watch.go implements the interactive watch menu for btest.
Running `blit -watch` drops into a bubbletea-powered UI with:
- A status bar showing active filters and run state
- A filter panel showing the current p/t/l settings
- A LogViewer streaming test output in real time
Key bindings (Vitest-style):
p prompt for path filter t prompt for test-name regex u toggle snapshot update f rerun only failed tests a rerun all tests l cycle log level: quiet → normal → verbose c clear filters (and log) q quit