Re: CLEAR Question for Input Files

[Dan <dbcerpg@xxxxxxxxx>]

Did your external DS idea fail because you had already defined one or
more of the file's fields in another DS?

It's been awhile, but wasn't there a discussion awhile back that
suggested there might be an enhancement in the future to allow a CLEAR
on an input-only record?

Question for Barbara, Hans:  Why does attempting to clear an input-only
record generate only a severity 00 warning (RNF7114)?  It would seem to
me to be more appropriate to generate a severity 30 to cause the
compile to abend.  After all, the intention to clear the record format
is rejected.  Are "save the programmer from himself" type errors not
used as a basis for severity 30 errors?  

Carrying this a little bit further.  If I change the RNF7114 message
description's Severity Code to 30, will that generate the compile abend
I'd like to see?  Are there any negative effects of doing this?  (I
know that release upgrades and perhaps PTF applys will likely reset
this parameter.)

- Dan

--- Michael Naughton <mnaughton@juddwire.com> wrote:
> Hi Everyone,
> 
> I've run into a problem that I wonder how other people solve. I've
> got a
> data file that I use for input only in my program, and I use the
> values to
> populate a screen (where they may get changed and then saved off
> someplace
> else). Typically, I CHAIN to the file and then fill the screen
> fields.
> 
> My problem is that if the CHAIN fails (which it will, from time to
> time),
> I get decimal data errors on the screen fields. I thought I could
> solve
> this by using CLEAR on the data file format, but CLEAR doesn't work
> since
> the file is input only.
> 
> I know I could just zero out all the fields individually, but that
> seems a
> little crude :-). I tried using an externally-defined data structure
> (I
> thought maybe I could clear that), but I got compile errors telling
> me I
> hadn't renamed the fields. I finally solved the problem by adding an
> A to
> the F-spec and putting in a dummy subroutine that writes the record
> format
> (which the program never calls). That worked like a charm, since now
> the
> CLEAR works after the failed chain, but I wonder -- is there a better
> way
> to do this?


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com