On May 16 00:48 94, Michael Stillwell of 3:633/262.1 wrote:

 MS> An interpreter changes one line of source code into machine 
 MS> language, runs it, then turns the next line of source code 
 MS> into machine language, runs it, and so on.

Well, not quite.  I've yet to see an interpreter that changes each line of
source code into machine code and then runs it.

Instead, a typical interpreter is likely to call a different machine
language routine for each different interpreted statement, and that machine
language routine will handle the rest of the statement, perhaps by calling
other machine language routines to access variables, or evaluate
expressions, or whatever.

As well as the straight interpreter and compiler, there's an
"inbetween" form that works like both - the source code gets
compiled into an intermediate code, and that intermediate code is then
interpreted.  QuickBASIC (when used as an interpreter) and QBASIC are
examples of such programs.

If there's any interest, I've written an interpreter for a basic-like
language, in QuickBASIC, that I could do a bit of "clean up" work
on, and then post here.  My language doesn't use line numbers, allows
program labels of any length, and has both string and numeric variables.

On the "down side", to make it easy for me to write, the
interpreter allows only a fixed number of variables with simple names,
which was sufficient for my purposes at the time.  Labels *start* with a
colon instead of ending with one like "regular" languages, and
variable names *start* with a $ or # instead of ending with one, again
simply to make it easier to write the interpreter.

Regards, Peter

--- Msgedsq 2.2e