RE: Use of BITS construct in table indices

 
Yes this makes sense to me now as an INDEX value of 0 (no bits set) is legal.
 
Thanks
Jay

-----Original Message-----
From: Wijnen, Bert (Bert) 
Sent: Thursday, November 14, 2002 3:41 PM
To: Daniel Chuang
Cc: Knotts, Jay B (Jay); [email protected]
Subject: RE: Use of BITS construct in table indices


Aha... right. that is the one. I had forgotten to start counting at zero... Oops

Thanks,
Bert 

-----Original Message-----
From: Daniel Chuang [mailto:[email protected]]
Sent: donderdag 14 november 2002 20:14
To: Wijnen, Bert (Bert)
Cc: Knotts, Jay B (Jay); [email protected]
Subject: Re: Use of BITS construct in table indices


I believe it is  N = (M/8) + 1.

Thanks,
Daniel

Wijnen, Bert (Bert) wrote:


Actaully I think it is:

   N = (7 + M)/8  where M is highest bit value

Thanks,
Bert 


-----Original Message-----
From: Knotts, Jay B (Jay) [ mailto:[email protected] <mailto:[email protected]> ]
Sent: donderdag 14 november 2002 15:45
To:  [email protected] <mailto:[email protected]> 
Subject: RE: Use of BITS construct in table indices



This is the correct formula for calculating string length of 
BITS? Just wanted to make sure. 



I am inclined to believe that it is considered a 

variable-length string,

and subject to being encoded with (N+1) subidentifiers where
N = (M + (M % 8))/8, M = highest bit value supported by an 

implementation,

unless it is an IMPLIED index. 

Jay Knotts 


-----Original Message-----
From: Juergen Schoenwaelder [ mailto:[email protected] <mailto:[email protected]> ]
Sent: Thursday, November 14, 2002 9:06 AM
To:  [email protected] <mailto:[email protected]> 
Cc:  [email protected] <mailto:[email protected]> 
Subject: Re: Use of BITS construct in table indices




Michael Kirkham writes:

Michael> RFC 2578 is a bit vague on the legality and implications of
Michael> using objects of type 'BITS' as an INDEX.  It's not
Michael> explicitly forbidden, and seems to be legal given that it is
Michael> otherwise indistinguishable from an OCTET STRING.  If it's
Michael> legal, then, for purposes of encoding the instance identifier
Michael> for a columnar object, is it considered a fixed-length or a
Michael> variable-length OCTET STRING (since new bits can be added
Michael> over time)?

RFC 2578 section 7.7 page 28:

   Since a single value of a Counter has (in general) no information
   content (see section 7.1.6 and 7.1.10), objects defined using the
   syntax, Counter32 or Counter64, must not be specified in an INDEX
   clause. If an object defined using the BITS construct is used in an
   INDEX clause, it is considered a variable-length string.

I think t

he last sentence provides a clear answer to your question.

/js

-- 
Juergen Schoenwaelder    

 <http://www.informatik.uni-osnabrueck.de/schoenw/> <http://www.informatik.uni-osnabrueck.de/schoenw/>