Re: [PATCH][RFC] vsprintf: unify the format decoding layer for its 3 users

[Linus Torvalds <torvalds@linux-foundation.org>]

On Sat, 28 Feb 2009, Frederic Weisbecker wrote:
> 
> You're right, that's much proper.
> See the V2 below:

Good. This looks better.

One final (?) nit:

> +		default: {
> +			enum format_type type = spec.type;
> +
> +			if (type == FORMAT_TYPE_LONG_LONG)
> +				num = va_arg(args, long long);
> +			else if (type == FORMAT_TYPE_ULONG)
> +				num = va_arg(args, unsigned long);
> +			else if (type == FORMAT_TYPE_LONG)
> +				num = va_arg(args, unsigned long);
> +			else if (type == FORMAT_TYPE_SIZE_T)
> +				num = va_arg(args, size_t);
> +			else if (type == FORMAT_TYPE_PTRDIFF)
> +				num = va_arg(args, ptrdiff_t);
> +			else if (type == FORMAT_TYPE_USHORT)
> +				num = (unsigned short) va_arg(args, int);
> +			else if (type == FORMAT_TYPE_SHORT)
> +				num = (short) va_arg(args, int);
> +			else if (type == FORMAT_TYPE_UINT)
> +				num = va_arg(args, unsigned int);
> +			else
> +				num = va_arg(args, unsigned int);
> +
> +			str = number(str, end, num, spec.base,
> +				spec.field_width, spec.precision, spec.flags);

You've got three copes of these (in the three different users), and that 
just looks horrid. I see why you did it, but even something like

		case FORMAT_TYPE_LONG_LONG:
			num = va_arg(args, long long);
			goto handle_number;
		case FORMAT_TYPE_ULONG:
		...
		case FORMAT_TYPE_INT:
			num = va_arg(args, int);
	handle_number:
		str = number(str, end, num, &spec);
		break;

would be better.

And yes, please do pass in "spec" to the "number()/pointer()/string()" 
routines instead of exploding it into the individual numbers. When 
cleaning up, let's just do it properly..

		Linus