RE: SHARE Comments Regarding Revision to IEEE 754-1985

["Bill Klein" <wmklein@xxxxxxxxxxxxx>]

Dr. Kahan and Mr. Michael, (et al)
   I am sending this note as an individual - not representing any group or
company.  However, I wanted to reply to one specific part of Dr. Kahan's
note where he stated,


________________________________
                
    DECIMAL   How will a decimal arithmetic standard be used,  and by whom?

________________________________

As the original author and presenter of the SHARE session dealing with the
ISO 2002 COBOL Standard (and IBM's current direction related to and
implementation of it), as well as a former member of and current follower of
the INCITS J4 Technical Committee (developer of past, current, and future
COBOL specifications),

I can assure you that *IF* the 754r revision is completed in time and
provides the facilities already specified in the existing draft, that the
decimal floating point standard will be heavily used by future conforming
COBOL compilers and those environments in which COBOL continues to be used
and extended.  This certainly includes many IBM environments (z/OS, VSE,
OS/400, etc) but also a number of Unix, Linux, and other "more open"
platforms.  Furthermore, it seems likely to me (but slightly less certain)
that many of the programming languages that interface with COBOL in such
operating system will ALSO have strong impetus to provide support for and
usage of this revised standard.  This will include languages such as PL/I,
REXX, RPG - not to mention Java, C++, etc.

Similarly (and in response to one of Dr. Kahan's other comments), the
current draft work on the ISO 2008 COBOL Standard also provides enhancements
to the "Common Exception Handling" facility introduced in the ISO 2002 COBOL
Standard that should provide both development-time debugging facilities and
production-time detection and recovery facilities. I would expect other ISO
standardized programming languages to provide comparable functionality.

The bottom-line (to me personally) is that there is a need for the speedy
completion and approval of the revision work - including a specific
"encoding" requirement for each type of (newly introduced) decimal floating
point item.  If provided to the computing community in a timely manner, I do
expect it to be widely used and implemented (whether in software, hardware,
or both).