[snip]
>>Your quest is: And NOW forcing back the notation for tables to "->" ?
>>I see no direct need though. The function anylib(), as built now, assumes that the cursor is open. If it has reason to suspect that it could be non-existing, then it should first test: IF USED('mycurs')
>>That would be a better security measure than using ->, IMO.
>Peter, you wished for a subroutine coder to be certain, that only memvars are accessed, and not a field in a table which is open by circumstance. That would entail accessing memvars first, possibly switchable by a set command. ***If*** this future Set command was active, how could you prevent reading from the (not planned for) memvar object if not by using ALLTRIM(myCurs->cMyField) ?

Another reason to be skeptical about SET MDOT indeed.

Do you see similar traps for the other proposals?
- USE (table) LOCAL | PRIVATE | PUBLIC
- HIDE TABLES ALL | EXCEPT (skeleton) | EXCEPT (list) && always only local
- LPARAMETERS( TABLE LocalAlias, VARIABLE LocalVar )

Your anylib() could look like:

FUNCTION anylib( TABLE ttMyCurs )
   ? ALLTRIM( ttMyCurs.cMyField )
   RETURN

or like:

FUNCTION anylib()
   HIDE TABLES EXCEPT MyCurs
   ? ALLTRIM( MyCurs.cMyField )
   RETURN

My own objection against both functions is that they would still have the MDot problem. In the first function there could still be an object variable ttMyCurs.
In the second function only HIDE TABLES ALL would work, but would be useless for anylib() since we're interested in the table's fieldvalue.