Re: iSCSI - decimal coded binary strings - a proposed resolution

[Paul Koning <ni1d@arrl.net>]

>>>>> "Julian" == Julian Satran <Julian_Satran@il.ibm.com> writes:

 Julian> Considering the widespread use of decimal encoding and the
 Julian> perceived difficulty of mapping them to a binary string -
 Julian> mainly due to the difficulty of matching lengths I suggest
 Julian> the following:

 Julian> Decimal encoding for binary strings will be used only where
 Julian> the length of the binary string is explicitly defined (by a
 Julian> constant or a rule that does not use the encoding itself)


 Julian> The definition for decimal constant becomes:

 Julian> decimal-constant: an unsigned decimal number - the digit 0 or
 Julian> a string of 1 or more digits starting with a non-zero
 Julian> digit. This encoding is not used for numerical values equal
 Julian> or greater than 2**64. Decimal-constants are used to encode
 Julian> numerical values or binary strings. Decimal constants can be
 Julian> used to encode binary strings only if the stringlength is
 Julian> explicitly specified. There is no implicit length for
 Julian> deci-mal strings.

The point about "no implicitic length" certainly helps.

I like David's suggestion (DLB's comments T.12) to disallow decimal
for data items that are capable of being larger than 2**64 -- rather
than tying it to whether the particular value being encoded happens to
be bigger or smaller than 2*64.  That way it essentially becomes a
property of the data type, which seems like a logical way to go.

Text to do that would be something like "This encoding is not used for
parameters that may have a value >= 2**64" or alternatively "This
encoding is used only with parameters whose largest permitted value is
less than 2**64".

	 paul