encoding/json.Marshal
standard libraryThe 15th most-called standard-library function by production calls. Called 1,690 times across 226 modules; 1,470 of those run outside tests and generated code.
5% in production calls vs 2026-06-sample (1,404 → 1,470)
- prod calls
- 1,470
- non-test, non-generated
- all calls
- 1,690
- total callsites
- modules
- 226
- distinct callers
- in loop
- 12%
- of prod calls
- in tests
- 10%
- 171 calls
- resolved
- 100%
- to a known target
resolution mix
How each call resolves
How each call was attributed. Only resolved kinds, where the target function is known, count toward the ranking. Interface dispatch and unresolved receivers are reported separately and never guessed. Read the taxonomy →
- Package function100%A direct call to an exported package function, resolved by type.
estimate
Estimated relative cost
Roughly 1.5× the estimated relative cost of fmt.Sprintf . Benchmarked at 642 ns/op × 1,470 production calls, normalized so fmt.Sprintf = 1.0.
This is not dollars and not a verdict. Per-op time depends heavily on inputs (argument sizes, contention, patterns), the benchmark ran on a single machine (darwin/arm64, go1.22.5), and static call counts are not runtime frequency. How this is estimated →
who calls this
Notable callers
Production calls into Marshal from notable Go projects. Open one on the map to see what else it hammers.
- Kubernetes 735 50%
- HashiCorp 441 30%
- Go (golang.org/x) 382 26%
- gRPC-Go 323 22%
where it's called
Callsites in the wild
6 representative callsites of 1,690, prioritizing in-loop and non-test usage. We link to source, never display it.
- in loop
-
k8s.io/apimachinery @v0.29.3
in loop -
spf13/cobra @v1.8.0
in loop -
stretchr/testify @v1.9.0
-
k8s.io/apimachinery @v0.29.3
- test
top callers
The modules that reach for it
The modules that call Marshal most in this snapshot.
- github.com/spf13/cobra59935%
- k8s.io/apimachinery25015%
- github.com/stretchr/testify22513%
siblings
More from encoding/json
The other most-called functions in this package. See all →