Context-aware Logging

Lakta uses github.com/Vilsol/slox for context-aware logging. Instead of passing a *slog.Logger around explicitly, slox retrieves the logger from the context.

Why context-aware logging

Middleware (gRPC interceptors, HTTP middleware) can inject structured fields — trace IDs, request IDs, user IDs — into the logger on the context. Any code that uses slox downstream automatically inherits those fields.

Basic usage

import (
    "log/slog"
    "github.com/Vilsol/slox"
)

func (m *MyModule) handleRequest(ctx context.Context, req *Request) {
    slox.Info(ctx, "handling request",
        slog.String("user_id", req.UserID),
        slog.Int("item_count", len(req.Items)),
    )

    if err := m.process(ctx, req); err != nil {
        slox.Error(ctx, "processing failed", slog.Any("err", err))
        return
    }

    slox.Info(ctx, "request complete")
}

Log levels

slox.Debug(ctx, "cache miss", slog.String("key", key))
slox.Info(ctx, "server started", slog.String("addr", addr))
slox.Warn(ctx, "retry attempt", slog.Int("attempt", n))
slox.Error(ctx, "fatal query", slog.Any("err", err))

Injecting fields into the context logger

Middleware can enrich the logger for all downstream calls:

func withRequestID(ctx context.Context, id string) context.Context {
    logger := slox.FromContext(ctx).With(slog.String("request_id", id))
    return slox.WithContext(ctx, logger)
}

Fallback behavior

If no logger has been injected into the context, slox falls back to slog.Default(), so code using slox is always safe to call even in tests without a full runtime.