Clark Morris wrote:

<begin extract>
. . . there should be an IBM supplied service that maps all PLS and/or
Assembler DSCECT to the appropriate mappings in other languages
including COBOL, PL1, C/C++, Rexx and others . . .
</end extract>

and  we should all, I suspect, agree that something of this sort would
be highly desirable.   Unfortunately, the obstacles to constructing
such a facility are not, or not chiefly, bureaucratic or financial.
Not all of the data types supported by one or more of these target
languages are supported by all of the others.

Notoriously---It is under current discussion here---COBOL does not
support explicit bit/boolean variables;  nor does it support
extended-precision HFP floating-point values or BFP asnd DFP
floating-point values of any precision.  REXX does not support proper
pointers.  C knows nothing of current-length-halfword prefixed varying
strings.   Etc., etc., ad nauseam.

Something can probably be done.  PL/I, for example, permits a
data-type slot in a structure declaration to be replaced by an
asterisk placeholder, and in COBOL 'FILLER' could be [mis]used for
this purpose.

The design problem is that of doing what is possible in a target
language without fatally compromising the usefulness of the templates
one produces, and it is a daunting one.

Moreover, statement-level procedural languages are not
interchangeable.  Some are much more powerful, in the sense of being
closer to assembly language in the facilities they make available,
than others.  Templates constructed in some target SLPLs are thus
likely to be very much more useful than those constructed in others.

None of these comments is intended to discourage worthy efforts of
this sort, but the difficulties do need to be appreciated.

John Gilmore, Ashland, MA 01721 - USA

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to