Intermediate ~15 min

Errors with Gin

Stdlib writes http.Error. Gin aborts with JSON and stops the chain:

r.GET("/tickets/:id", func(c *gin.Context) {
    if c.Param("id") != "7" {
        c.AbortWithStatusJSON(404, gin.H{"error": "not found"})
        return
    }
    c.JSON(200, gin.H{"id": 7})
})

The return after Abort is load-bearing — without it Gin double-writes.

From one handler to all of them

The AbortWithStatusJSON you write here handles one failure in one handler. But a ticket API has dozens of handlers, and each one can fail the same ways — not found, bad input, something broke. Copying the envelope into every handler rots the same way duplicated middleware does.

The way out is the post-Next phase from the middleware lesson, turned into a pattern. Handlers stop answering their own failures: they record them with c.Error(err) and keep going. One middleware, attached with r.Use(), runs after c.Next() returns, inspects the collected c.Errors, and writes the single envelope:

func ErrorHandler() gin.HandlerFunc {
    return func(c *gin.Context) {
        c.Next() // handlers run first
        if len(c.Errors) > 0 {
            c.JSON(500, gin.H{"success": false, "message": c.Errors.Last().Err.Error()})
        }
    }
}

Same factoring-out as middleware itself, one level up: per-handler aborts become per-request error collection.

One related trap: gin.CustomRecovery() takes a func(c *gin.Context, recovered any) for panics, and inside it you must call an abort method yourself — c.AbortWithStatus(), for example. Forget it and subsequent handlers keep executing right past the panic.

Further reading

Hint

Check c.Param("id"), AbortWithStatusJSON(404, ...) + return, else c.JSON(200, ...).

Task

Fill in NewRouter in errors.go.

Check

go test ./exercises/17_frameworks/06_errors/ -v

Source: exercises/17_frameworks/06_errors/README.md · Inspired by Mainmatter’s 100 Exercises to Learn Rust (CC BY-NC 4.0) — all Go prose here is original, free for non-commercial use.