Hi Tom,
I disagree entirely with your reasoning below.
The main purpose of PSI is print-by-reference. If Send-URI
remains OPTIONAL (and therefore rarely implemented and not
interoperable) in IPP, then gateways based on FSG PAPI/1.0
interfaces between PSI and IPP transports will fail (unless
the gateway fetches the referenced document, which introduces
a different set of security issues).
This isn't a percentage kind of thing. The most important
operation in PSI is AddDocumentByReference. All of the mobile
scenarios depend on it. Mobile devices _cannot_ fetch local
copies of large documents, in order to use AddDocumentByValue.
Cheers,
- Ira
-----Original Message-----
From: Hastings, Tom N [mailto:hastings@cp10.es.xerox.com]
Sent: Thursday, April 24, 2003 8:19 PM
To: printing-architecture@freestandards.org
Cc: 'ipp@pwg.org'; McDonald, Ira
Subject: RE: IPP> FW: Summary of PWG Document Object issues
To the FSG Group,
cc: IPP DL
Ira wrote:
PWG PSI/1.0 (now in working group 'last call') has only
REQUIRED operations (including print-by-reference), but
the IPP Document Object spec defines many operations and
almost all Document Description attributes as OPTIONAL.
This impacts any adoption of PSI by FSG PAPI, because
interworking with IPP-based intermediate systems and
printers will be degraded or will fail in some cases.
So lets list the REQUIRED and OPTIONAL operations in the current spec (not
counting the proposals to increase the conformance that are out for a two
week comment and which have met with objection on the mailing list):
2 REQUIRED Job operation: Close-Job, Get-Documents
3 REQUIRED Document creation operations: Create-Document, Send-Document
(RFC 2911), Send-Data
1 OPTIONAL Document creation operation: Send-URI (RFC 2911)
3 REQUIRED Document operations: Cancel-Document, Get-Document-Attributes,
Get-Documents
2 OPTIONAL Document operations: Delete-Document, Set-Document-Attributes
ISSUE: So which operations would the FSG PAPI not want to include in the
PAPI because they are OPTIONAL in the Document object spec, but would if
they were REQUIRED?
I don't see that any of the OPTIONAL operations would be serious omissions
from the PAPI, if they weren't included in the PAPI.
Also lets see which of the 8 Operation/Document Description attributes are
REQUIRED and OPTIONAL in the Document object spec:
1. ipp-attribute-fidelity (boolean) *
Job Creation
MUST
Allows the client to indicate whether or not the Printer MUST support all
Job Template and Document Template attributes.
2. job-mandatory-attributes (1setOf type2 keyword)
Job Creation
MUST
Allows the client to list which Job Template, Document Template and
Operation attributes MUST be supported.
3. document-charset (charset)
Document Creation
CMUST (The Printer MUST support the "document-charset" operation/Document
Description attributes if the Printer supports a document-format in which
the charset may be ambiguous in the Document content, such as
'application/vnd.hp-PCL' where the charset escape sequence MAY be omitted
from the data)
The charset of the Document Object content
4. document-digital-signature (type2 keyword)
Document Creation
MAY
the type of digital signature, if any used in the Document Object content.
5. document-format (mimeMediaType)
Document Creation
MUST
The document format of the Document Object content
6. document-format-details (1setOf collection)
Document Creation
CMUST (The Printer MUST support the "document-format-details"
operation/Document Description attributes, if the Printer supports a
packaging document format, such as 'application/zip' or 'multipart/related')
The details of the Document Object content, including when its package of
files (zip, multipart/related), such as application and operating system
that created the document, the intended device type (when the format is
device-dependent), the natural languages of the document.
7. document-format-version (text(127))
Document Creation
MAY
The version of the document format.
8. document-natural-language (naturalLanguage)
Document Creation
MAY
The primary language of the Document Object content.
ISSUE: Here the OPTIONAL operation/Document Description attributes can be
without serious problems by a Printer that doesn't support them.
So I don't see a problem with the PAPI including any of the OPTIONAL
Operation/Document Description attributes in the PAPI, even though they are
OPTIONAL.
Tom
-----Original Message-----
From: McDonald, Ira [mailto:imcdonald@sharplabs.com]
Sent: Thursday, April 24, 2003 15:07
To: 'ipp@pwg.org'
Subject: IPP> FW: Summary of PWG Document Object issues
Hi folks,
I sent the summary below to the Free Standards Group
Open Printing Architecture mailing list earlier today.
Most of the issues/topics below were discussed during
this afternoon's continued review of the IPP Document
Object spec in the PWG Semantic Model telecon.
Most of these issues remain unsolved.
Comments?
Cheers,
- Ira McDonald
High North Inc
-----Original Message-----
From: McDonald, Ira [mailto:imcdonald@sharplabs.com]
Sent: Thursday, April 24, 2003 1:12 PM
To: printing-architecture@freestandards.org
Subject: [Printing-architecture] Summary of PWG Document Object issues
Hi,
At today's FSG OP Architecture telecon, Claudia asked that I
send out a summary of recent issues for IPP Document object.
I used letters for the points below to avoid ambiguity with
the many email notes on the PWG PSI, SM, and IPP reflectors.
A) Print-by-reference authentication
(IPP "Send-URI" <==> PSI "AddDocumentByReference")
PWG PSI's design center is print-by-reference (a hand-held
telling a PSI Print Service to print some document available
on a Web or FTP server). But the necessary authentication
credentials to _access_ the referenced file's URL must also
be available (or no security exists). PSI actually sends
these credentials (inside a TLS-secured session). But, as
Michael Sweet (CUPS) has pointed out:
a1) Simple end-user impersonation (username/password) is
fragile and the print server may accidentally expose
the user's security info - a liability for the vendor.
a2) Stronger certificate-based public key authentication
(usually used in TLS-secured sessions) may fail because
some certificates are only valid if used _from_ the end
user's home system (as identified by an FQDN stored in
the certificate and validated by DNS lookup for source
IP address for the transaction).
a3) Without Kerberos-style single-use "tickets" the delegation
of end user credentials to an intermediate server is an
unsolved software problem (existing solutions only work in
certain not-widely-deployed middleware).
B) REQUIRED versus OPTIONAL operations and attributes
PWG PSI/1.0 (now in working group 'last call') has only
REQUIRED operations (including print-by-reference), but
the IPP Document Object spec defines many operations and
almost all Document Description attributes as OPTIONAL.
This impacts any adoption of PSI by FSG PAPI, because
interworking with IPP-based intermediate systems and
printers will be degraded or will fail in some cases.
C) Flat-file registries needed for key Document attributes
IPP Document object (based on input from PWG PSI and from
CIP4 JDF people) adds to the base "document-format" (MIME
type values) the parallel qualifier "document-format-versions"
(with values like "PDF/1.4" and "PDF/is-1.0"). The operation
of the PWG IPPFAX protocol (soon to enter 'last call') and
of IPP-based Document operations _depends_ on the presence
of "document-format-versions".
However, it remains to be settled whether the PWG will rent
space on a commercial server (to avoid burdening our current
host Lexmark with a file that might be downloaded by many
clients) or the CIP4 will archive the file on their Web site.
Several other attributes that can be present in the new
"document-format-details" compound attribute also require
registration of standard values, such as
"document-source-application-version" ("MS Word 2000").
D) Breaking IPP Document object into two (or more) specs
Dennis Carney (IBM) recently suggested a compromise solution
of making _two_ specs: one with only REQUIRED operations
and attributes; and one with only OPTIONAL ones. It turns
out (per Tom Hastings) that the first spec would be _very_
skinny. Also, a whole bunch of important (but OPTIONAL)
Document attributes would be delayed in the second spec
(expected to move more slowly through the adoption process).
There really should be a third spec (again, per Tom H) that
contains the Job-level operation extensions and attributes.
E) IPP/1.2
Recently, Dennis Carney (IBM) observed that IPP Document
object was starting to look a lot like "IPP/1.2".
Michael Sweet suggested yesterday that perhaps we should
be _writing_ an IPP/1.2 spec, and gathering up the numerous
IPP extensions (several dozen) into one place (by reference,
I hope) with a new set of conformance requirements.
(I would collaborate on such a project, but I wouldn't
take it on alone.)
Predications:
I predict that some part of IPP Document object will go to
working group 'last call' pretty soon because it's holding up
the release of both the PWG Semantic Model 1.0 (and schemas)
and PWG PSI/1.0 (and WSDL headers), both of which want full
Document object semantics in their content.
I also predict that some/most of the IPP Job-level extensions
and some/most of the "document-format-details" attributes will
be delayed much longer (months at least) by the process and also
by the bandwidth of the editors (basically Tom Hastings, with
some help from Dennis Carney and a few others).
Cheers,
- Ira McDonald
High North Inc
_______________________________________________
Printing-architecture mailing list
Printing-architecture@freestandards.org
http://freestandards.org/mailman/listinfo/printing-architecture
This archive was generated by hypermail 2b29 : Fri Apr 25 2003 - 14:44:03 EDT