Re: iSCSI: Text/Operational keys: booleans and ranges

[Luben Tuikov <luben@splentec.com>]

Paul Koning wrote:
Paul Koning> 
Paul Koning> >>>>> "Luben" == Luben Tuikov <luben@splentec.com> writes:
Paul Koning>  Luben> How about ranges using the dash?
Paul Koning> 
Paul Koning>  Luben> True, one could _count_ the commas to fugure out if it is a
Paul Koning>  Luben> list (number of commas > 1) or a range (number of commas = 1),
Paul Koning>  Luben> but you have to add code to _count_ the commas.
Paul Koning> 
Paul Koning> That doesn't work, because sets can have two members.

Implied in my original post:

I was merely pointing this out _considering_ the variable
which is currently processed. (i.e. is it a range or integer or string
by it's nature)

E.g. OFMarkInt cf. MaxRecvPDULength cf. TargetName (resp.)!

By the way, it is also my point that it doesn't work quite so well,
and this is why I started this thread.

Paul Koning>  Luben> Using a dash for ranges, the list of numerical values is
Paul Koning>  Luben> handled by default by the already existing code in your
Paul Koning>  Luben> implementation to intesect two strings of comma separated
Paul Koning>  Luben> objects.
Paul Koning> 
Paul Koning> Not true.  Integers and strings are different data types.  Sets of
Paul Koning> strings are in the spec now, sets of integers are not.

Paul, what is the intersection of ``hello,world'' and ``hello''?
How about ``1,20'' and ``1''? Anyway this is not important.

Anyway, back to the main issue: Paul, you agree that using a
different than comma character to denote ranges of integers is
a good thing?

-- 
Luben

``Beauty [Perfection] is achieved not when there is nothing to add,
  but when there is nothing to take away.''
     Antoine de Saint-Exupery