Re: [open-cobol-list] Re: IS "PIC -.99999999E-99" invalid picture string for OpenCOBOL?

[John R. Culleton]

On Tuesday 08 November 2005 07:07 am, Roger While wrote:
> This is a floating-point edited field.
> It consists of 3 parts; the significand (AKA mantissa), the
> symbol 'E' and the exponent.
>
> It was not defined in the 85 standard.
> It is defined in the 2002 standard.
>
> Question to Bill :
> Why does the 2002 standard not allow a minus '-'
> following the 'E' ??
>
> It is not currently supported by OC.
>
> Roger

So now I have a question: does OC offer support for the 2002
standard? 

If so perhaps we need a description of what features have been
included and what excluded. 

My instinct is to write to the earliest standard that will get
the job done, since any given compiler is likely to have
implemented those features first. Any standard is an amalgam of
features old and new. The older features are more likely to be
included and debugged in any given compiler. In particular I
wouldn't want to use unusual features from the latest standard
in a job to be compiled by a free compiler. That is asking for
difficulty. 

I repeat my earlier advice to Mr Wang. Statistical analysis
usually doesn't need to be done as part of a workflow. It is a
separable function. So the easy way home is to write the data
from the COBOL program to a file, the data fields in display
format, the file itself defined as LINE SEQUENTIAL.

This file should be readable by a C program without difficulty,
and the fields placed in appropriate form for statistical
manipulation. This involves Mr. Wang in minimal manipulation of
COBOL. It ain't elegant, but it is the surest way home. 

Grace Hopper said more than thirty years ago "Don't ask FORTRAN
to the work of COBOL, or COBOL to do the work of FORTRAN." It is
still good advice. 
-- 
John Culleton
Books with answers to marketing and publishing questions:
http://wexfordpress.com/tex/shortlist.pdf

Book coaches, consultants and packagers:
http://wexfordpress.com/tex/packagers.pdf