Skip to content

Working directory and paths

This page explains where Prof reads your module, where it writes .prof/, and how paths are laid out under each tag so pprof and your own tooling can find data.

Before you begin

  • You have a go.mod at the directory you treat as the module root (usually the repo root for the module).
  • You run Prof with that directory as your current working directory (cwd), the same way you run go test ./... for that module.

What Prof uses from cwd

Prof uses your current working directory to find the Go module and to write .prof/. Run from the same place you run go test for that module (usually the module root).

  • Benchmark discovery is relative to cwd (same rules as go test).
  • Output goes to .prof/<tag>/ under cwd (terminology).
  • prof config init writes minimal prof.json and commented prof.json.example next to go.mod at the module root. Keep cwd aligned with that root when you expect those files to be found.

Directory layout under .prof/<tag>/

Using Prof creates a .prof/ tree next to your module, one folder per run (tag). Domains describe the data they hold (domain/<BenchmarkName>/artifact).

Path What it is
.prof/<tag>/ One labeled run: profiles, measurements, hotspots, and optional extracts for that tag.
.prof/<tag>/profiles/<BenchmarkName>/ Raw pprof profile binaries (.out); durable source for go tool pprof.
.prof/<tag>/measurements/<BenchmarkName>/ go test benchmark run stats (run.txt: ns/op, allocs).
.prof/<tag>/hotspots/<BenchmarkName>/ Function-ranked stack summaries per profile (cpu.txt, memory.txt).
.prof/<tag>/call_trees/<BenchmarkName>/ Call-tree text (pprof -tree) per profile.
.prof/<tag>/source_lines/<profile>/<BenchmarkName>/ Per-function pprof -list extracts when configured.
.prof/<tag>/call_graphs/<profile>/<BenchmarkName>/ Optional Graphviz PNG call graphs when installed.
.prof/<tag>/data_mapping/<BenchmarkName>/map.json Machine-readable index of artifacts for this benchmark (paths, semantics, top symbols, function inventory).
.prof/<tag>/notes.txt Short tag-level note (placeholder until you edit it).
prof.json Active config next to go.mod after prof config init or Manage configuration in prof ui.
prof.json.example Commented reference (not loaded); copy optional sections into prof.json. See Configure — generated files.

Details on collection flags and behavior: Collect profiling data. Configuration keys: Configure collection.

Configuration files

Both files live beside go.mod at the module root:

  • prof.json — active config (valid JSON). Created minimal by prof config init; add a collection section as needed.
  • prof.json.example — commented reference with doc links; not loaded by prof.

Testing / verify

From the module root, after a successful collect, you should see a new directory .prof/<your-tag>/ with at least profiles/<BenchmarkName>/, measurements/<BenchmarkName>/, and hotspots/<BenchmarkName>/ populated for the profiles you enabled.

If .prof/ never appears, see Troubleshooting.

Next steps