Re: paradox

[Barney Wolff <barney@databus.com>]

On Tue, Jul 11, 2006 at 10:28:52AM -0700, Glen Zorn (gwz) wrote:
> 
> >> Just to be clear, I don't think I've claimed that there is a
> >> document, just that there are known methods that (slightly extended)
> >> to deal with the problem.  Off the top of my head, RFC 2868
> >> specifies an (admittedly unsophisticated) method to group related
> >> attributes together using "tags".  Although that document only uses
> >> tags to group attributes of different types, I don't see why a
> >> similar technique would not work with attributes of the same type
> >> with the addition of the rule that all instances of the same
> >> attribute with the same tag be concatenated in order.  A problem
> >> would arise if you wanted to group long attributes with others but I
> >> didn't think that we were talking about that...          
> > 
> > Well yes, I suppose that would work, for long attributes restricted
> > to ASCII printable strings.  
> 
> Actually, I'm not sure why that restriction is necessary?

You need to know whether the first byte is a tag or not.  I suppose
one could instead require that any attribute of length 253 always
have a tag.  But in that case the actual tag is irrelevant.  So here's
an alternative proposal for overcoming the 253 byte limit:

If value length <253, entire value in one AVP.
If value length >=253, first 252 bytes in first AVP, suffixed with ignored
byte.  Repeat until the entire value has been consumed.
In other words, a length field of 255 becomes special, indicating both
that the last byte of the value is to be ignored and that the next instance
of the same attribute type is a continuation.  There is a backwards
compatibility risk, that an old sender will generate an AVP with length
field 255 for a 253-byte value.

-- 
Barney Wolff         I never met a computer I didn't like.

--
to unsubscribe send a message to radiusext-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/radiusext/>