Re: Two reade loops: which one preferred ?

A reason why style #1 might be preferred is that 'Do forever' is 
misinformational unless you have a veeery large file.

But I actually use style #3:
----- Style #3 -------------------------------------------

mykey setll record
dou %eof
mykey reade record
if not %eof
... process ...
endif
enddo

because I think it's cleaner.
Why?
1. First of all: Read can return either a) a record or b) an eof flag, and we 
(you) only want 'process' in case 'record'
   You could also inside the loop have coded for both situations (select or 
if..else)
2. It's clean: only a single 'read' to control a 'read-loop'.  Both style #1 
and #2 has two or three different routes through the code: zero records, a 
record read and no more records
3. No 'Do forever' that definately is not 'forever'
4. Same construction (the fewer the better) for 'read entire file' (except the 
missing setll of cause).

It might mean a few more cykles, but they are cheaper than programmer time

Henrik
http://hkrebs.dk

> date: Fri, 21 Nov 2003 05:58:13 +0100
> from: afvaiv <afvaiv@xxxxxxxxxx>
> subject: Two  reade loops: which one preferred ?
> 
> Two different but functionally equivalent "read equal" loops follow.
> 
.
.
> 
> So, applying the C/PASCAL approach to RPG forces to include one 
> additional "read" before the loop as in Style#1.
> 
> Just for curiosity, some people will code it the #2 way, but WHY do MOST 
> people seem to prefer #1 ?
> 
> ----- Style #1 -------------------------------------------
> mykey setll record
> if %equal
> mykey reade record
> dow not %eof
> ... process ...
> mykey reade record
> enddo
> endif
> 
> ----- Style #2 -------------------------------------------
> 
> mykey setll record
> if %equal
> do *HIVAL
> mykey reade record
> if %eof leave
> ... process ...
> enddo
> endif
> 
> -- 
> Antonio Fernandez-Vicenti