Posted: 11 years 3 months 7 days 20 hours 38 minutes ago
Of course it would. by definition an "empty" VARYING field is '' (zero length).
So CLEAR myVARFld would set its length to 0 and may even clear out the existing data.
FYI, I normally do this:
myVarField = '';
Because that's often more clear to others.
Posted: 11 years 3 months 7 days 20 hours 14 minutes ago
I presumed it would but had to check anyway. You know, Murphy's Law.
Posted: 11 years 3 months 7 days 19 hours 32 minutes ago
Note the following regarding VARLEN/VARYING fields:
Increasing the length causes RPG to clear out the new locations before moving in the new data. So if a 50-byte VARYING field is currently 10 bytes long, and you move a 20 position value into it, positions 11 through 20 are cleared before the move is performed.
If you then move an 8 position value into that same field (which now has a "current length" of 20) the compiler simply moves the data into the field, left-justified and changes the current length to 8. This leaves the old data in position 9 to 20 and is what causes VARYING to perform better than fixed-length fields.
Posted: 11 years 3 months 7 days 2 hours 10 minutes ago
Thanks for the confirmation Bob.
Posted: 11 years 3 months 7 days 19 minutes ago
Never thought about the performance aspect of VARYING vs. fixed; will have to keep that in mind. For the record, I've always explicitly zeroed the length:
%LEN(myVarField) = 0;
Sounds like I should switch to CLEAR, which does exactly the same thing with less typing. 
Posted: 11 years 3 months 2 days 4 hours 9 minutes ago
"I don't see this behavior documented in the RPG reference PDF."
Actually, it is documented. In Ch. 22 (Op Codes), CLEAR says it sets elements in a structure or a variable to their default initialization value depending on field type. In Ch. 9 (Data Types and Data Formats), Character Data Type, Variable-Length Character..., it says, "A variable-length field is initialized by default to have a current length of zero." QED