Re: Opnqryf with Many Selections

Richard,

It can be, I did say -can be-, messy keeping track of all of the quote
marks.

And it really does look ugly.  Am I quoting the value or am I quoting
the string?  How many quotes do I -REALLY- need?

IMHO, building the string can and should be black boxed so a particular
CL program (and the programmer) does not need to go through the "count
the quotes" exercise.

If the purpose of a program is to provide a list/display with selection
capability, the -raw how- to build the selection string should not be in
the picture.  If it is, the -raw how- must be replicated in each and
every program that provides selection capability.  Now in our shop
that's just about every one of them.

Since you are going to do the same thing in every one of them how about
we make the exercise a clean function to reduce programmer effort?  All
it amounts to is a simplification of the -how-.

Let's compare your case to the one that our shop uses:

Here's yours:

if (&fld1 *ne ' ') do
chgvar &qryslt ('fld1 = ')
chgvar &qryslt (&qryslt || &q || &fld1 || &q)
enddo

if (&fld2 *ne ' ') do
if (&qryslt = ' ') chgvar &qryslt ('fld2 = ')
else chgvar &qryslt (&qryslt |> '*and fld2 = ')
chgvar &qryslt (&qryslt || &q || &fld2 || &q || ' ')
enddo

Here's ours:

if (&fld1 *ne ' ') do
chgvar &value (&dblquote *cat &fld1 *tcat &dblquote)
call $qryslt2 parm(&andor 'fld1' 'eq' &value &qryslt)
chgvar &andor ('and')
enddo

if (&fld2 *ne ' ') do
chgvar &value (&dblquote *cat &fld2 *tcat &dblquote)
call $qryslt2 parm(&andor 'fld2' 'eq' &value &qryslt)
chgvar &andor ('and')
enddo

Now there isn't a whole lot of difference, except that in your case for
&fld3 to determine if you should place the *and, you must test &fld1 and
&fld2 and as you add more fields to must add more compares.

In my case, each test does not need to know the existence or absence of
any other value.  New selection criteria can be added without having to
retrofit any other blocks of code.

There is a trivial program called ($qryslt2) that builds the string.
BTW, I couldn't figure out from your code where you put the '(' ')' from
your code, if both values are entered you would result in the following
string:
fld1 *eq 'whatever' *and fld2 *eq 'somethingelse'

In my case I have the following string:
(fld1 *eq 'whatever') *and (fld2 *eq 'somethingelse')

I've never used a string of your construct and I'm sure that it is
valid, since you said so, but IMHO, the second is easier to debug. 
Especially if the string is constructing many more variables.

It's not that I disagree with you in concept, building a string is a
minor exercise.  I would just like to add that it can be made easier and
cleaner for the programmer that must do the job.

There was a thread on this list some time ago where I vehemently stated
that if a programmer doesn't know how to do the job they can either keep
up or eat dust.  Since then I've made it easier to keep up.  I'm getting
lazier in my old age :-)

> Richard Jackson wrote:
> >
> > Buck:
> >
> > Why is writing a CL program messy?
> >
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---