attachment-0001
<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">
<HTML>
<HEAD>
<META content="text/html; charset=iso-8859-1" http-equiv=Content-Type>
<META content='"MSHTML 4.72.3110.7"' name=GENERATOR>
</HEAD>
<BODY bgColor=#ffffff
style="FONT: 10pt Arial; MARGIN-LEFT: 2px; MARGIN-TOP: 2px">
<DIV>I've added a few comments below.</DIV>
<DIV><FONT style="BACKGROUND-COLOR: #ffffff">-Hugo</FONT><BR><BR>>>>
"McDonald, Ira" <imcdonald@sharplabs.com> 03/02/00 10:19AM
>>><BR>> [Sorry for earlier empty reply - sticky
mouse]<BR>><BR>> Hi Paul,<BR>><BR>> We could certainly use
'collection' to hold the values in<BR>> 'application/ipp' encoding (that is,
the URI and zero<BR>> or more of their associated metaparameters for
security,<BR>> authentication, etc.).</DIV>
<DIV> </DIV>
<DIV>Tom is right in saying that the decision of how to represent this
information in SLP and LDAP can be made independently of how we choose to
represent it in IPP. And I think we're converging on Ira's proposal for
the SLP and LDAP representation.</DIV>
<DIV> </DIV>
<DIV>Except ... there's real urgency in fixing the IPP definition now given the
impending publication of the Set-Printer-Attribute operation. Fixing the
IPP definitions for these parallel attributes will be much harder after the
"Set" operation is out (unless the current attribute threesome is
marked as "Read-Only" in the spec with a note that a standard way for
setting those attributes will be available in the future).</DIV>
<DIV><BR>><BR>> But in SLP, LDAP, and a lot of other directory
interfaces,<BR>> we need a simple string encoding (see the Subject of<BR>>
this message).<BR>><BR>> In the SLP and LDAP encoding of the three
original<BR>> parallel attributes, we already use the '>' (greater<BR>>
than) as the URL delimiter (per RFC 2396 and advice<BR>> from Larry
Masinter).<BR>><BR>> If we use 'collection' syntax in
'application/ipp',<BR>> the new combined attribute (now a collection)
won't<BR>> be accessible to IPP/1.0 and IPP/1.1 existing software<BR>>
(some of which IS extensible for new attributes with<BR>> simple
datatypes).</DIV>
<DIV> </DIV>
<DIV>By pre-fixing collection attribute names we achieve the same level of
interoperability with existing IPP clients as the one you imply. Namely,
the reading of a semantically opaque attributes with a known syntax.
Notice that IPP clients from the beginning have been advised to skip over
unknown attribute syntaxes. Current clients will have no harder time
ignoring the 'start-of-collection' and 'end-of-collection' delimiters than
ignoring the new 'xri' syntax.</DIV>
<DIV> </DIV>
<DIV>Again, we don't have to worry about current clients being able to set the
new collection syntax because the "Set" operation is not out
yet. Hence the time-sensitive nature of making this decision real soon.
</DIV>
<DIV> </DIV>
<DIV>> Thus, I was thinking of simple<BR>> multi-valued strings (an actual
new 'xri' syntax isn't<BR>> strictly necessary for the IPP encoding), for
better<BR>> interworking.</DIV>
<DIV>><BR>> Cheers,<BR>> - Ira McDonald<BR>></DIV></BODY></HTML>