home *** CD-ROM | disk | FTP | other *** search
- From: <jsh@usenix.org>
-
-
- An Update on UNIX*-Related Standards Activities
-
- June 1990
-
- USENIX Standards Watchdog Committee
-
- Jeffrey S. Haemer, Report Editor
-
- IEEE 1003.5: Ada bindings
-
- Jayne Baker <cgb@d74sun.mitre.org> reports on the April 23-27 meeting
- in Salt Lake City, UT:
-
- Overview
-
- The Utah meeting was the group's first since our October meeting in
- Brussels. In the interim, we had completed a mock ballot of Draft
- 4.0. Jim Lonjers of Unisys, one of our two co-chairs, managed the
- effort. The document was mailed out to reviewers on December 1, 1989
- and comments were due January 19, 1990. Although only 16% of the
- ballots were returned, the high quality of the comments received made
- the mock ballot a success. Ted Baker, of Florida State University,
- hosted a special meeting in Tallahassee, March 19 - 23, to resolve
- issues and comments; the result was draft 4.1. We did not attend the
- January, New Orleans meeting because balloters lacked sufficient time
- to review and return comments prior to the meeting, though some
- members came to attend other groups' meetings.
-
- Our main goal in Utah was to prepare the Ada Language Binding Document
- for IEEE and ISO Ballot. We addressed the few, unresolved technical
- issues from mock ballot; read draft 4.1 cover-to-cover, for accuracy
- (of text and Ada code), content, and consistency; established a plan
- for addressing the ISO formatting issues; adopted an optimistic
- schedule for IEEE and ISO ballots; and tried to establish a position
- on threads.
-
- Unresolved Technical Issues from Mock Ballot
-
- Most unresolved technical issues from the mock ballot were trivial,
- and quickly resolved. They included the details of iterations (e.g.,
- through a directory), string lower bounds with respect to a string
- returned by a function, the behavior of a file that is opened non-
- blocking when the I/O operation cannot complete, static initialization
- versus ``easy implementation'' of constants, and Text I/O page
- terminators.
-
- __________
-
- * UNIX is a registered trademark of AT&T in the U.S. and other
- countries.
-
- June 1990 Standards Update IEEE 1003.5: Ada bindings
-
-
- - 2 -
-
- The most detailed discussion involved whether or not files should be
- closed on an Exec. The Ada binding provides a Start_Process function,
- which is a primitive that safely creates a new process. In the face
- of Ada tasking, Fork and Exec are unsafe and cannot be used to
- accomplish the results of a Start_Process call. Once one of these
- unsafe primitives is issued, an application program is no longer under
- the control of the Ada run time system; the operating system is
- involved. Therefore, the integrity of the child process is
- jeopardized, and the state of the process's I/O (i.e., which files are
- open/closed) is not guaranteed. Application programs that must be
- safe with Ada tasking and must have files closed and buffers flushed,
- should call Start_Process to create a new process.
-
- Global Issues Effecting the Document
-
- We solved several global, editorial issues. We agreed to use a terse
- wording style except where a more lengthy, explanatory style is needed
- for clarity. We accepted the current packaging of the Ada code
- (multiple packages) and the non-Ada-Language-Reference-Manual coding
- style. Chapter authors were assigned action items to complete their
- respective references and rationale sections.
-
- We spent a large portion of the meeting going through the document,
- chapter-by-chapter, noting very specific changes. Changes recorded in
- a ``master red-lined'' copy were forwarded to appropriate chapter
- authors at the close of the meeting. These changes will be made
- before the June delivery of the document to WG 15.
-
- ISO Format Issues
-
- We need to make several minor modifications, additions, and deletions
- before the June WG 15 meeting, to put the document in ISO standard
- format. After the March, Tallahassee meeting, Jim Moore, of IBM,
- investigated the possibility of hiring a consulting technical editor
- to do this work. IBM volunteered to fund this effort at a level
- sufficient to translate the document into ISO format, maintain that
- format, and make one major edit and two to three minor editorial
- revisions. We accepted IBM's offer, and hired Hal Jespersen.
-
- Threads Issues
-
- As in New Orleans, several group members met with P1003.4 for threads
- discussions. Most group members feel we should establish a position
- on threads, but we remain firmly divided on what it should be.
- Several members believe the currently defined primitives (i.e., the
- most basic system functions) are insufficient, and think that any
- thread model and primitives proposed should be sufficient to support
- Ada tasking, and implement an Ada Run-Time. In contrast, at least one
- group member believes we are unrealistic to require a threads proposal
- in C to meet Ada requirements -- we should, instead, require that C
- and Ada be able to play together in some reasonable fashion, and have
-
- June 1990 Standards Update IEEE 1003.5: Ada bindings
-
-
- - 3 -
-
- a fair understanding of how it will be accomplished. We decided to
- admit our dissension to P1003.4. Interested P1003.5 members are
- acting as liaisons to represent their own views, but these liaisons do
- not represent any consolidated P1003.5. view.
-
- The IEEE and ISO ballots
-
- Steve Deller, our chair, asked the Sponsor's Executive Committee (SEC)
- to approve our entry into the IEEE ballot process. Jim Isaak, SEC
- Chair, met with us early in the week to discuss the IEEE and ISO
- ballot processes and help us establish a schedule to reach IEEE and
- ISO ballots simultaneously. Since the ballot process seems to be of
- general interest, here is a brief overview.
-
- A hierarchy of organizations is responsible for producing
- international operating-system standards and managing the ISO ballot
- process. Two independent, international standards organizations, the
- International Standards Organization (ISO) and the International
- Electrotechnical Committee (IEC), sit on top. Joint Technical
- Committee 1 (JTC 1), a combined effort of these two organizations
- designed to coordinate their efforts in areas of overlap, is at the
- second level; Subcommittee 22 (SC 22), Languages, at the third; and
- Working Group 15 (WG 15), Portable Operating Systems for Computer
- Environments, at the fourth. National organizations, such as the
- American National Standards Institute (ANSI), manage ISO balloting
- within each country. Each participating country has one or more
- representatives in WG 15. The United States has a Technical Advisory
- Group (TAG), which meets with and advises the United States' WG 15
- representatives on the US's position on important issues.
-
- This bureaucracy requires quite a bit of coordination and planning to
- coordinate IEEE and ISO ballots. Most documents require about one
- year to complete the IEEE ballot cycle. Historically, POSIX documents
- have begun with the IEEE ballot process; three to four months later,
- either the original draft, or a newer version incorporating IEEE
- -ballot-process comments, enters the ISO process, and is delivered to
- both WG 15 and SC 22 for approval. Typically, the IEEE ballot is held
- open until all comments from both IEEE and ISO processes are received,
- reviewed, and incorporated. The result is returned to both the IEEE
- and ISO ballot groups for their approval. If the IEEE comments are
- substantive, they enter into the ISO process by the submission of a
- United States position. For example, P1003.1a is the U.S. position on
- P1003.1..
-
- Our group also initiated formation of a formal ballot group -- is the
- group that will actually vote on the current draft. We will deliver
- Draft 5.0, in ISO format, to WG 15 at the Ada Europe meeting this
- June. Draft 6.0 will go to IEEE ballot on August 6. If we receive
- the required 75% response by September 21, the ballot will close
- immediately; if not, we must reconsider the ballot group membership,
- and revise our schedule. In early October, draft 6.0 will be delivered
-
- June 1990 Standards Update IEEE 1003.5: Ada bindings
-
-
- - 4 -
-
- to SC22. At the October meeting, in Seattle, we will resolve the IEEE
- ballot comments and produce Draft 7.0, which will enter the ISO Ballot
- process. At the January '91, New Orleans Meeting, we will determine
- whether a second IEEE Ballot is needed. Any changes to Draft 7.0
- resulting from a second IEEE Ballot will be entered into the ISO
- process through a pro forma objection. There are no guarantees, but
- P1003.5 could reach Draft International Standard (DIS) status by late
- second quarter of 1991.
-
- Conclusion
-
- The April '90/Salt Lake City Meeting was a success. We addressed the
- issues we hoped to address and attained our goal for the meeting. We
- also established a schedule for reaching IEEE and ISO ballot; although
- this schedule is optimistic, we think we can meet it.
-
- June 1990 Standards Update IEEE 1003.5: Ada bindings
-
-
- Volume-Number: Volume 20, Number 38
-
-