SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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


    • To: Julian_Satran@il.ibm.com
    • Subject: Re: iSCSI - decimal coded binary strings - a proposed resolution
    • From: Paul Koning <ni1d@arrl.net>
    • Date: Fri, 12 Jul 2002 16:49:30 -0400
    • CC: ips@ece.cmu.edu
    • Content-Transfer-Encoding: 7bit
    • Content-Type: text/plain; charset=us-ascii
    • References: <OF215A0B29.49DEF9F0-ONC2256BF4.004F78DF@telaviv.ibm.com>
    • Sender: owner-ips@ece.cmu.edu

    Sorry, I meant to reply earlier.
    
    >>>>> "Julian" == Julian Satran <Julian_Satran@il.ibm.com> writes:
    
     >> 2. On the other hand, for numbers, the proposed rule says
     >> that you can encode the value in decimal whenever the value
     >> happens to be less than 2**64.  If 5 minutes later that same
     >> parameter happens to have a value 2**64 or greater, then you
     >> cannot encode that particular value in decimal.  So the
     >> encoding rule here is tied to the value, not the parameter; a
     >> given parameter sometimes permits decimal and sometimes not.
    
     >> That's what I meant: it is extra complexity to have an
     >> encoding rule for a variable that isn't the same for all
     >> possible values of the variable.
    
     Julian> +++ that is only an issue for the encoder - and encoding is
     Julian> not an issue for any of the encoding methods - you send
     Julian> starting (supposedly) from an internal value of defined
     Julian> length and may use whatever is fit but we may tight the rule
     Julian> to allowed values not actual values.  Is that acceptable?
     Julian> +++
    
    No, the issue is in the decoder.
    
    If I have a bignum parameter, I have to be prepared to get a decimal
    encoding coming in.  I'd have to feed that to atoi() (or its 64-bit
    equivalent), take the resulting 64-bit value and supply the high order
    0 bits to store it into my bigger-than-64-bits variable.
    
    That's extra work for no purpose.
    
           paul
    
    


Home

Last updated: Sun Jul 14 19:18:55 2002
11318 messages in chronological order