>Can you prove that it's faster? It is certenly isn't cleaner.

yes. within a INSERT the difference is minimal, but difference exist.

VFP eval "(ALIAS())" with 10% extra time respect to "ALIAS()"

CREATE CURSOR my (x L)
t1=SECONDS()
FOR i=1 TO 3000000
NEXT
t2=SECONDS()-t1
t1=SECONDS()
FOR i=1 TO 3000000
 =ALIAS()
NEXT
? SECONDS()-t1-t2
t1=SECONDS()
FOR i=1 TO 3000000
 =(ALIAS())
NEXT
? SECONDS()-t1-t2
t1=SECONDS()
FOR i=1 TO 3000000
 =((((((((((((((((((((((ALIAS()))))))))))))))))))))))
NEXT
? SECONDS()-t1-t2

whichever character in a VFP variable expression demands time to runtime.
VFP have a expression optimizer only for constant expression.

t1=SECONDS()
FOR i=1 TO 3000000
NEXT
t2=SECONDS()-t1
t1=SECONDS()
FOR i=1 TO 3000000
 =4
NEXT
? SECONDS()-t1-t2
t1=SECONDS()
FOR i=1 TO 3000000
 =(4)
NEXT
? SECONDS()-t1-t2
t1=SECONDS()
FOR i=1 TO 3000000
 =((((((((((((((((((((((4))))))))))))))))))))))
NEXT
? SECONDS()-t1-t2

>From VFP help:
>"INSERT INTO dbf_name
>Specifies the name of the table to which the new record is appended. dbf_name can include a path and can be a name expression".

With this logic, UPDATE,SELECT AND DELETE don't support name expression.

VFP Help is a correlate HELP, not a exact HELP.

>I didn't say that. It's your own conclusion.
> I don't see how is it relevant.

Is relevant because nobody know the true VFP specific design ( for me this document don't exist ), and whichever conjecture is possible.

However I hope that also this message comes read from the VFPT, and that they draw their conclusions.