Re: How do I clip a vary length field?

[Vernon Hamberg <vhamberg@xxxxxxxxxxxxxxx>]

Hi Barb

I did wonder about that - it might be a MEMCPY kind of thing - whatever 
that function is. It just felt kind of like a "cute" technique, not 
well-commented on, hence not easily understood or modified later. Gary 
Cornell, who wrote several books on programming in Windows, spoke 
strongly against "cute" approaches - 3 rules of optimization - Don't! 
Don't! Don't!

Still , it will be interesting either to see the value in debug or to 
see the output of a dump after the error.

Cheers
Vern

On 1/16/2020 9:44 AM, Barbara Morris wrote:
> Vern, in the earlier part of this thread from last year, this technique was
> discussed (mapping a varying field across an integer followed by a string).
>
>     dcl-ds ds;
>        whatever char(10);
>        len int(10);
>        data char(100);
>        more stuff char(100);
>     end-ds;
>
>     dcl-s varfld varchar(100:4) based(p);
>     dcl-s p pointer inz(%addr(len));
>
> Chris is doing something like that. Whether it's a good idea or not, I
> think it does work.
>
> At least, it works if the length (subfield EZDQLSTL, in Chris's case) is
> less than or equal to the defined length of the based VARCHAR field.
>
> Chris, if you can debug this, what is the value of EZDQLSTL when you do the
> assignment to Upper?
>
> -- Barbara
>
> -----------------------------------------------------
> message: 1
> date: Wed, 15 Jan 2020 19:06:10 -0600
> from: Vernon Hamberg <vhamberg@xxxxxxxxxxxxxxx>
> subject: Re: How do I clip a vary length field?
>
> I think I see the problem, Chris - EZDQLSTL is defined as an integer -
> it is the length of the SQL statement, not the statement itself.
>
> There are a couple elements that are described as "varying" - IIRC, that
> doesn't mean they are a varying-length field, just that the length is
> not fixed in the data structure, and the length is in the above-noted
> element. In working with APIs like this, you have to use the offsets and
> lengths provided. The SQL statement doesn't have an offset in the DS,
> since it is always at a fixed point, but the "extended cursor name" DOES
> have an offset, because it is located at different places in the DS,
> depending on the length of each instance, where the statements are not
> the same length.
>
> you'll need to modify things - you can use a pointer, perhaps, to the
> element that follows EZDQLSTL, then a substring on the variable there
> and use the length for that.
>
> Have fun!! It's API heaven!
>
> Vern