SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    Re: effect of initializing CRC reg to 1's depends on implementati on? iSCSI



    --- Mark Bakke <mbakke@cisco.com> wrote:
    > This is becoming somewhat disturbing, given the comment
    > about
    > the Ethernet CRC.
    
    Which comment did you have in mind?
    
    >  The intent of the document is
    > absolutely,
    > positively to keep the CRC identical to Ethernet, with
    > the
    > exception of the generator, just as Paul had said.
    
    Then we should JUST say THAT in the iSCSI spec.
    I think that it is enough.
    
    > If someone has an
    > implementation
    > that does not match the examples, they simply try a few
    > variations until they get it right.
    
    Mark, iSCSI draft/rfc is supposed to be THE AUTHORITY
    reference for iSCSI implementation.
    
    You cannot say ``just try a few variations until you get it
    right''!
    
    >  This has worked very
    > well
    > for everyone I have talked to who is implementing the
    > CRC.
    
    The reason is that they know and/or have a preconception
    about what is being talked about. I.e. the current
    description is very IMPLICIT. As part of a definition
    document it has to be EXPLICIT.
    
    You need an outsider, an impartial party to scrutinize the
    draft. All parts of it.
     
    > Here is some text that I had suggested adding instead of
    > (or 
    > in addition to) the mathematical description a while ago:
    
    I liked the one you had above better.
     
    > The above, plus the examples, have seem to work out just
    > fine
    > for the implementors.  Getting the math right would be
    > nice, but
    > most of us implementors don't look at it anyway, and
    > prefer the
    > more practical description.  If we are going to include
    > the math,
    > it has to match the examples.
    
    Most certainly agree! Including math is not necessary. We
    can prepare a document (what I and Vince are doing) and
    make a reference to it in the draft for all the math,
    optimization, derivation, equivalence and implementaion
    details.
     
    > Anyway, my guess is that is what you are discussing; it
    > was just
    > disturbing to think that anything but the Ethernet + new
    > poly
    > was being considered, since that has been agreed on for a
    > long
    > time, and is being widely implemented.
    
    ... and not being mentioned anywhere in the draft!
    
    -l
    
    
    
    
    =====
    --
    
    __________________________________________________
    Do You Yahoo!?
    Check out Yahoo! Shopping and Yahoo! Auctions for all of
    your unique holiday gifts! Buy at http://shopping.yahoo.com
    or bid at http://auctions.yahoo.com
    


Home

Last updated: Mon Dec 17 15:17:41 2001
8105 messages in chronological order