Standards Update, 1003.9: POSIX Fortran-77 Bindings

[Peter Collinson]

Submitted-by: pc at hillside.co.uk (Peter Collinson)

USENIX Standards Watchdog Committee
Stephen R. Walli <stephe at usenix.org>, Report Editor
Report on 1003.9: POSIX Fortran-77 Bindings


E. Loren Buhle, Jr. Ph.D. <BUHLE at XRT.UPENN.EDU> reports on the April
15-19, 1991 meeting in Chicago, IL:

POSIX.9 met to resolve objections and comments raised to the first
ballot of the FORTRAN binding to ISO/IEC 9945-1 Standard (also known
as POSIX.1).  The ballot began in late December 1990 and ended on
February 20, 1991.  This first proposal did not obtain the necessary
75% acceptance of the balloters.  There were 73 people in the total
balloting group, of which 56 were eligible to vote on the standard.
The others were parties of interest.  Of the official balloting group,
there were 23 affirmative votes, 15 negative votes, and 8
abstentions.  This 82% response was only 60% affirmative.  Thus the
first ballot failed to make the existing draft a standard.

At the Chicago meeting, objections and comments from all voters (both
official and unofficial) were reviewed and acted upon.  Many valid
points were made by the voters, resulting in changes to the draft.
Some revisions included changing the F77 prefixes to PXF (e.g.
F77WAIT became PXFWAIT).  Joseph King's request for a ``fast exit'' was
also added.

Fast exit was added back to the draft to gain the _exit()
functionality contained in POSIX.1. It is required to allow proper
recovery from failed calls to any of the PXFEXEC() functions within a
child process.  It seems that recovery means that the child process
must be able to exit without flushing buffers. The file buffers of a
child process are copies of the parent's.  The current draft says that
on failure when PXFEXIT(), STOP and END are executed, the data in the
buffers will be written to the file and the child will terminate. So
when the parent writes or closes the file, the output buffers will be
flushed and data will be duplicated (once from the failed child and
once from the parent) in the file.

Most of the objections and comments were resolved in a positive
fashion, providing for the possibility of a successful second ballot.
With some fast work from the 8 attendees to the POSIX.9 meeting, the
revised draft may be recirculated in June for a 30 day period.  If all
goes well, the results of the recirculation ballot can be ready for
resolution during the July meeting.

The next meeting of the POSIX.9 working group will be July 8-12, 1991
at the Doubletree in Santa Clara, California.  The subsequent meeting
will be October 21-25, 1991 in Parsippany, NJ.


Volume-Number: Volume 23, Number 98