nirvana-labs/nirvana-go v1.82.0 on GitHub

1.82.0 (2026-05-14)

Features

client: optimize json encoder for internal types (7e861d5)

This pull request is managed by Stainless's GitHub App.

The semver version number is based on included commit messages. Alternatively, you can manually set the version number in the title of this pull request.

For a better experience, it is recommended to use either rebase-merge or squash-merge when merging this pull request.

🔗 Stainless website
📚 Read the docs
🙋 Reach out for help or questions

Greptile Summary

This release (v1.82.0) introduces a JSON encoding optimization for internal SDK types: a new WithSkipCompaction option bypasses the redundant appendCompact scan that was applied after every json.Marshaler call, reducing O(keys²) iteration cost on nested SDK structs.

internal/encoding/json: appendCompact signature is widened from a plain escape bool to a full encOpts struct; a new opt.go file adds the Option functional-option pattern and WithSkipCompaction; RawMessage (which relied on the stdlib) is commented out from stream.go.
packages/param/encoder.go: MarshalWithExtras and MarshalUnion now forward WithSkipCompaction(true) to the shim encoder when marshaling SDK types, trading JSON-validity re-scanning for performance.
packages/param/encoder_test.go: 176 lines of new tests verify the optimization across compacted/non-compacted cases; the shimjson-marshal sub-test uses t.Logf on mismatch (see inline comment).

Confidence Score: 4/5

Safe to merge — the optimization is scoped to SDK-internal types and all existing callers pass the correct argument, but the WithSkipCompaction API has a latent defect that warrants a follow-up fix.

The core optimization logic is sound and well-tested. The one real defect is that WithSkipCompaction(b bool) always enables the optimization regardless of the value of b, so a caller passing false would get the opposite behavior with no indication of the bug. Today every call site passes true, so there is no runtime breakage, but the public API is wrong and could confuse future callers.

internal/encoding/json/opt.go — the WithSkipCompaction bool parameter is dead code and should be fixed before the API surface hardens.

Important Files Changed

Filename	Overview
internal/encoding/json/opt.go	New file introducing `Option`/`WithSkipCompaction`; the bool parameter to `WithSkipCompaction` is always ignored, so passing `false` still enables the optimization.
internal/encoding/json/encode.go	`Marshal` gains variadic `opts ...Option`; options are applied before marshaling; `marshalerEncoder`/`addrMarshalerEncoder` now pass full `encOpts` to `appendCompact`.
internal/encoding/json/indent.go	Signature of `appendCompact` widened from `(escape bool)` to `(opts encOpts)`; `Compact` updated accordingly; logic is equivalent to before.
internal/encoding/json/stream.go	`RawMessage` commented out (it depended on the stdlib `encoding/json`); only `packages/param/param.go` references it via the standard-library import, so no breakage.
internal/encoding/json/time.go	Single-line update to pass `opts` instead of `opts.escapeHTML` to `appendCompact`, matching the updated signature.
packages/param/encoder.go	`MarshalWithExtras` and `MarshalUnion` now pass `WithSkipCompaction(true)` when calling the shim encoder, enabling the new optimization for SDK types.
packages/param/encoder_test.go	Comprehensive new tests for the compaction optimization; `shimjson-marshal` subtest uses `t.Logf` for assertion failures, so mismatches do not fail CI.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["shimjson.Marshal(v, opts...)"] --> B["apply encOpts"]
    B --> C["marshalerEncoder / addrMarshalerEncoder"]
    C --> D["v.MarshalJSON() → raw bytes"]
    D --> E{opts.skipCompaction?}
    E -- true --> F["append raw bytes directly\n(no scan, no validation)"]
    E -- false --> G["appendCompact\n(scan + remove whitespace + escape HTML)"]
    F --> H["final JSON output"]
    G --> H

    style F fill:#d4edda,stroke:#28a745
    style G fill:#fff3cd,stroke:#ffc107

Prompt To Fix All With AI

Fix the following 2 code review issues. Work through them one at a time, proposing concise fixes.