Re: Chain doesn't set %EOF -- RPG400-L

Why doesn't a failed CHAIN (which sets off %FOUND) also set on %EOF?

Because CHAIN jumps straight to a particular record. Either that record is there or it isn't. It has no concept of whether it's at the end of the file or not. Wait... that's not entirely true. If it succeeds in reading a record it knows it's not at the end of the file :) But, when it fails, it doesn't know.

Rather that doing a SetLL/ReadE combo to start the loop it used to be
shorthand to code a Chain with:

Chain ( SomeKey ) File;
DoW Not %Eof( File );
  // do something
  ReadE ( SomeKey ) File;
EndDo;

Don't think of %EOF or %FOUND as a replacement for the indicators declared in your program. Think of them as a replacement for the HI-LO-EQ positions of the source record.

CHAIN always set the "HI" indicator, which meant "%FOUND". READ and READE always set the "EQ" indicator which meant "%EOF". When they designed the BIFs they followed the same philosophy. Granted, they probably should've looked at how RPG programmers used this capability instead of just copying the way it's always worked -- BUT that's what they did.

Anyway, assuming that you followed that logic :) you can do what you always did:

   chain (SomeKey) File;
   *IN67 = not %found;

   dow not *in67;
     // do something
     ReadE ( SomeKey ) File;
     *IN67 = %eof;
   EndDo;

Even better, using the BIFs you can also use a named indicator!

 D success         s              1N

 . . .

   chain (SomeKey) File;
   success = %found;

   dow success;
     // do something
     ReadE ( SomeKey ) File;
     success = not %eof;
   EndDo;

Always better to use meaningful names!

Any possibility of getting Chain to set the %EOF indicator as well as the %FOUND indicator? Or would that break existing code?

If you're going to "fix" it, I'd suggest that it'd be better if they actually worked like functions and RETURNED whether they were successful or not. That way you could code your loop as follows:

   setll (SomeKey) File;

   dow (%READE (SomeKey) File);
     // do something
   EndDo;

That makes the loop far more elegant, and eliminates the need for the awkward "priming read" and "read again at the end of the loop". If they're going to change it anyway, wouldn't that be better?