Re: Using a D-spec to define an action in the eval

[Barbara Morris <bmorris@xxxxxxxxxx>]

Brian Johnson wrote:
> I left out the bits not relevent to the question.  Here's another sample
> that does compile (with guess for parameter and return types and sizes):
>
>      d RawReading      c                   'RawReading'
>      d f               pr             1a   extproc('XMLRtnAttr')
>      d  func                         10a   const
>      d wRawReading     s              1a
>       /free
>        wRawReading = f( RawReading ) ;
>        return ;

Using the name "f" does save space, but I find that it significantly 
decreases readability, especially if this coding pattern is used for 
other similar problems.  If you don't know what RawReading is connected 
with, there is no hint that this line has something to do with XML 
attributes.

For maximum maintainability, I would take this in the opposite 
direction.  It is easy to find everything related to XMLRtnAttr, and it 
is difficult to make an error by passing an irrelevant parameter to the 
procedure.

(I just struggled with a similar problem in my own code yesterday, and 
after trying a version with briefer names, I decided it was better to 
use the same name everywhere, to avoid the same problems that Booth 
mentioned.)

     d XMLRtnAttr_RawReading...
     d                 c                   'RawReading'
     d XMLRtnAttr      pr             1a   extproc('XMLRtnAttr')
     d  func                         10a   const
     d wRawReading     s              1a
      /free
       wRawReading = XMLRtnAttr( XMLRtnAttr_RawReading ) ;
       return ;

To reduce the typing effort, if you use an editor like the WDSC editor, 
and type say XMLR then CTRL-space, it will bring up a list of all the 
possible names for you to choose from.