std-unix@longway.TIC.COM (Moderator, John S. Quarterman) (08/24/89)
An Update on UNIX* and C Standards Activities
August 1989
Jeffrey S. Haemer, Report Editor
USENIX Standards Watchdog Committee
IEEE 1003.5 Ada Language Update
Ted Baker (tbaker@ajpo.sei.cmu.edu) reports of the April 1989 meeting:
The Minneapolis meeting started off poorly. The chairman, co-
chairman, and technical editor were absent, though each for good
reasons. ("Co-chairman" is POSIX for vice-chairman.) Only one of the
members present had received a copy of the latest draft (2.0). Many
of the changes agreed upon at the last meeting (Fort Lauderdale) were
not yet reflected in this draft. There was no agenda.
Despite these handicaps, the group made considerable progress. Steve
Deller acted as chair, working up an agenda and holding the group
fairly closely to it. (Indeed, Steve Deller has now become an
official co-chair, but is still doing a good job.)
By the second day copies of Draft 2.0 had been made. This draft was
reviewed completely, and several changes were approved. The hottest
issue was how signals would be mapped to Ada task entries. Several
semantic gaps in the P1003.1 C-language binding were discovered, and
passed on to the P1003.1 working group.
Most major semantic issues were, at this point, resolved.
1. Each Ada program consists of a single POSIX process, or at least
appears to be so through the POSIX/Ada interface.
2. POSIX signals are handled by Ada tasks via the same mechanism as
hardware interrupts, as logical entry calls.
3. POSIX character and string types are distinct from the standard
Ada character and string types.
4. The C-binding's "errno" values are translated into distinct Ada
exceptions.
__________
* UNIX is a registered trademark of AT&T in the U.S. and other
countries.
Jeffrey S. Haemer, Editor USENIX Standards Watchdog Committee
August 1989 Standards Update - 2 - IEEE 1003.5 Ada Language
5. The Ada-binding need not follow the organizational and naming
conventions of the C-binding, especially where they violate
principles of data abstraction.
What remains is filling in a lot of details, including most of the
text of the document, and making it stylistically consistent.
Group members volunteered to edit the agreed-upon changes into the
draft document, while filling in missing text. This work was to be
completed before May 10-12, at which time a subset of the working
group would meet in Bedford Mass. for a "writing party". The goal of
this party would be to catch up, completing all missing portions of
the draft, so that it could be submitted for mock ballot before the
July P1003 meeting. There was some question whether this goal would
be met. (The mock ballot date was missed, so it appears 1003.5 won't
have an official Ada language binding that corresponds to 1003.1 by
end-of-year 1989.)
There were also coordination meetings (BOFs) with the groups working
on language-independent specifications (P1003.1) and threads
(P1003.4). The Ada group seemed generally pleased with progress on
the language-independent specification, and hopes that the draft Ada-
binding will provide some guidance to that activity. The group is
less pleased with the tendency of other groups (e.g. P1003.2 and
P1003.4) to aggravate the problem of C-dependencies in their draft
documents.
The Ada group is very interested in having the 1003.4 standard include
multi-threaded processes, but is very concerned that any such standard
be compatible with the semantics of Ada tasks. Some of the preliminary
proposals coming out of the threads working group do not seem to be
compatible with this goal.
Jeffrey S. Haemer, Editor USENIX Standards Watchdog Committee