Document Number WG14 N856

Title: Liaison report on WG15 activities

Date: 1998-09-16

Source: Keld Simonsen, liaison from WG14 to WG15 - POSIX

Action: for discussion at WG14 Santa Cruz meeting

This liaison report updates WG14 on issues pertaining to C in POSIX.

There are no formal messages that WG15 has requested 
to be relayed to WG14.

The POSIX work is planned to be restructured. There is a proposed
policy to do joint development of POSIX standards between ISO, IEEE and
The Open Group (TOG), at least on the core API and utilities
standards. The work is to be done via joint meetings, much like what
has been done for some years now between WG14 and J11.
The procedures previously has been that POSIX standards be
developed in IEEE or TOG (independently, but with quite some
overlap in personnel) and then processed for completion in ISO
when the standards had been adopted.  The intention by the joint
work is to avoid duplication of work, and also to possibly
have just one set of standards in the field. One added benefit
is to gather all experts in the field for the work and thus get
the optimum level of expertise focused to work on a common standard.

One problem with this is that the common standard will need to
be adopted by each of the standardization pocedures of the
3 groups involved, but it is hopet that this will be a solvable issue.

One of the main standards produced in this way will be
the common base standard, which is planned to be based on the TOG
Single UNIX specification Issue 2 (SUS2), comprising of the documents
XBD5, XCU5, XSH5 and XNS5.  This is roughly equivalent to ISO/IEC 
9945-1 and 9945-2 plus the C standard and networking APIs.

One problem I can see here for WG14 is that the SUS2 incorporates
the C standard and have some enhancements, instead of referencing
it normatively. I would like WG14's opinion on what should be done here.

The POSIX groups are concerned on how to use
C9X in the new common POSIX base standard.
Some of the goals that has been discussed in the POSIX groups,
and where it was thought that C9X may facilitate a solution were 
the following:

n-bit neutrality. The common POSIX standard should not make
any implicit assumptions about the size of datatypes.
All sizes sould be explicit. Previously some 32-bit
implicit assumptions had crept into the standard.

Architecture neutrality. This would include binary
representation neutrality and neutrality in endianness.

Are there facilites in C9X that WG14 could recommend to
the POSIX groups for this purpose?

The POSIX groups would like to encourage participation from 
C experts in the POSIX work. Are there any special measures
that WG14 could make to do this?

End of document.