home *** CD-ROM | disk | FTP | other *** search
- Submitted-by: stephe@usenix.org (Stephen R. Walli)
-
- Report on POSIX.17 - Directory Services API
-
-
- Mark Hazzard <markh@rsvl.unisys.com> reports on the April 6 - 10, 1992
- meeting in Dallas, Texas:
-
- Summary
-
- Draft 3.0 of POSIX.17 began IEEE ballot on April 7th and finished the
- first round of balloting May 19th with 84% of the ballot group
- responding. We completed sending responses to all who participated in
- the Mock Ballot of Draft 2.0.
-
- The group formed a ballot resolution team, and dealt with the "Which
- track to ISO?" issue. Splitting/re-casting our Project Authorization
- Request (PAR) was a hot topic. We're following a PMC recommendation
- to separate the Directory Services API work (which is in ballot) from
- the POSIX name space issue which hasn't received much attention.
-
- Introduction
-
- The POSIX.17 group has defined and is balloting a user to directory
- API (e.g. API to an X.500 DUA - Directory User Agent). We used APIA -
- X/Open's XDS specification as a basis for work. XDS is an object
- oriented interface and requires a companion specification (XOM) for
- object management.
-
- XOM is a stand-alone specification with general applicability beyond
- the API to directory services. It is used by IEEE P1224.1 (X.400 API)
- and is being standardized by the P1224 working group.
-
- The current POSIX.17 PAR has a two part scope. The first authorizes
- the group to work on an API to directory services. The second (and
- more contentious) part addresses the POSIX name space issue. The
- working group has discussed name space, but decided to focus on the
- API to directory.
-
- POSIX.17 is one of five "networking" groups under TCOS, and comes
- under the purview of the Distributed Services Steering Committee
- (DSSC).
-
- Status
-
- The group finally completed all the written responses to the comments
- received from the Mock Ballot of Draft 2.0 of our document. If you
- responded, you should have a reply by now.
-
- Draft 3.0 of POSIX.17 was distributed for IEEE ballot just prior to
- the Dallas meeting, and included all test methods and the language
- independent specification (LIS). The document grew from 234 pages in
- Draft 2.3 to 478 pgs in Draft 3.0 with the inclusion of all remaining
- test methods.
-
- As of this writing, the 1st ballot is now officially closed, with 84%
- of the Ballot Group returning ballots.
-
- Once again, we met with POSIX.12 (Protocol Independent Interfaces) in
- joint session and discussed their requirements on directory services.
- The white paper produced by POSIX.17 was used as a basis for moving
- ahead on requirements. (The white paper was the result of an action
- taken in Irvine to document agreements, assumptions, issues, options
- and proposed actions.)
-
- The meeting was quite productive and resulted in an understanding on
- how to progress the work. POSIX.17 took an action to assist the
- POSIX.12 group with writing an annex mapping a simplified, more
- focused interface to the POSIX.17 API.
-
- Some POSIX.17 members met with P1224 to process the
- comments/objections raised during the initial round of balloting of
- the object management specification.
-
- The PMC recommended in January that the POSIX.17 project request (PAR)
- be split into two separate projects, one for the Directory Services
- API work (which is in ballot) and the other for the POSIX name space
- issue which hasn't received much attention.
-
- Name space conjures up many different things for different audiences.
- Some folks see the issue as a language issue, dealing with function
- prefixes and the like. The working group sees the issue as one in
- which objects are uniquely named in a global context, i.e. beyond a
- single kernel. If we use the process id as an example, we find that
- the 5- digit positive integer used as the name for a process within
- most kernels doesn't scale too well globally. If I want to have a
- utility that determines the status of all my processes, even those on
- other kernels, I have to somehow extend the name space.
-
- There was a spirited debate as to whether or not a second PAR was
- needed for name space work, in that the issues could be resolved by
- some other mechanism in the TCOS realm. Neither POSIX.17 nor the
- System Interface Coordination Committee (SICC) believe that POSIX.17
- owns the ``C'' name space issue. A white paper will be produced
- summarizing the name space issue and the work to date. Stay tuned ...
-
- The road to ISO
-
- The group spent much time debating how to progress the POSIX.17 API
- work for ISO standardization. The central point of contention was a
- proposal to remove the POSIX.17 API from ISO 9945-1 to join
- P1224/P1224.1 in a to-be- determined track in ISO. [Ed. - ISO/IEC
- 9945-1 is the ISO name for IEEE 1003.1, or POSIX.1. All other system
- interfaces, such as POSIX.4 real-time and POSIX.6 security, are
- supposed to be integrated to 9945-1 in future ammendments.]
-
- The rationale given was that since the POSIX.17 work is dependent on
- P1224 and all three documents share the same style of interface and
- roots, they should all be progressed to ISO within the same Working
- Group. Since P1224 and P1224.1 aren't part of (and won't be part of)
- 9945-1, POSIX.17 should be pulled out of 9945-1 and progressed with
- the other two documents.
-
- There is a risk that ISO SC22/WG15 (the ISO POSIX SubCommittee 22
- Working Group) will not accept a work item for an API to directory
- services outside of 9945-1. The implication is that a new SC22 working
- group (or one from SC21 or SC18) may be required for this work, with
- all the associated start-up overhead. All this could delay the work
- and subsequently jeopardize its completion as an ISO standard.
-
- Taking the work from 9945-1 also breaks the link requiring a
- distributed POSIX system to include an API to directory services. At
- least one other distributed services working group (POSIX.12) was
- concerned about this as well.
-
- Arguments against the non-9945-1 track to ISO resulted in a compromise
- that will (hopefully) allow us to retain the reference to the POSIX.17
- work in a work item for 9945-1. The work item could revert to a
- pointer to the work being done outside of 9945-1 (if that comes about)
- and also serve as a place holder for our work within SC22 WG15 if
- another track couldn't be found.
-
- A resolution was prepared for the SEC, proposing that the SEC
- authorize POSIX.17 to take several actions relating to the mechanics
- of progressing our document through the IEEE ballot process and on to
- ISO. After some initial tough sledding late Thursday night, (my
- Minnesota roots showing), the SEC accepted all the time critical
- aspects of the resolution, deferring the rest until Chicago.
-
- In Closing ...
-
- Once again, there are quite a few homework assignments between
- meetings. The ballot resolution process begins. Look for a white
- paper rationalizing the directory services API work with the name
- space issue. We also need to submit a New Project proposal for
- progressing the POSIX.17 to ISO within SC22.
-
- The group will meet next time in Chicago, concentrating on Ballot
- resolution and name space issues. We plan to meet in Utrecht and
- possibly for a few days in Reading, UK, to complete the work for our
- first (and hopefully final) ballot recirculation.
-
- Volume-Number: Volume 28, Number 37
-
-