jackc/pgx · error

invalid digitsRead: (this can't happen)

Error message

invalid digitsRead: %d (this can't happen)

What it means

While accumulating nbase digit groups into a big.Int, the loop selects a power-of-10000 multiplier based on how many digits the helper read (1-4). This guard fires when digitsRead is outside 1-4 — an internal invariant violation the message itself labels 'this can't happen'. It indicates a bug in the digit-reading helper or memory corruption, not bad user input.

Solutions

1. Report as a pgx bug with the numeric value that triggered it
2. Upgrade to the latest pgx — the nbase accumulation loop has had correctness fixes
3. Try text format for the column to bypass binary decoding

When it happens

Trigger: Thrown at pgtype/numeric.go:676 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of jackc/pgx@ec1a0befd2 (2026-08-04). Data as JSON: /api/errors/e173577cc72b2f91. Report an issue: GitHub.

Appendix: source

Thrown at pgtype/numeric.go:676

	// int(ndigits) before the addition: ndigits is a uint16, so ndigits+3
	// would wrap for counts above 65532 and skip the loop entirely.
	for i := 0; i < (int(ndigits)+3)/4; i++ {
		int64accum, digitsRead := nbaseDigitsToInt64(r, int(ndigits)-i*4)

		if i > 0 {
			var mul *big.Int
			switch digitsRead {
			case 1:
				mul = bigNBase
			case 2:
				mul = bigNBaseX2
			case 3:
				mul = bigNBaseX3
			case 4:
				mul = bigNBaseX4
			default:
				return fmt.Errorf("invalid digitsRead: %d (this can't happen)", digitsRead)
			}
			accum.Mul(accum, mul)
		}

		accum.Add(accum, big.NewInt(int64accum))
	}

	if err := r.Finish(); err != nil {
		return fmt.Errorf("numeric: %w", err)
	}

	exp := (int32(weight) - int32(ndigits) + 1) * 4

	if dscale > 0 {
		fracNBaseDigits := int(ndigits) - int(weight) - 1
		fracDecimalDigits := fracNBaseDigits * 4
		dscaleInt := int(dscale)

View on GitHub (pinned to ec1a0befd2)