RE: Hyphenated OpCode Names

["Holden Tommy" <Tommy.Holden@xxxxxxxxxxxxxxxxx>]

so that's the reasoning?  Then would it not be better to implement as a BIF 
%corr(this_stuff:that_stuff) as someone else suggested...sorry i forgot who 
suggested the bif approach..)

Thanks,
Tommy Holden


-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx]On Behalf Of Barbara Morris
Sent: Tuesday, June 07, 2005 2:43 PM
To: rpg400-l@xxxxxxxxxxxx
Subject: Re: Hyphenated OpCode Names


Bob Cozzi wrote:
> ...I mean, one can not
> advocate free format and then use Z-ADD as justification when Z-ADD is not
> part of free format syntax to begin with.
> 

Bob, Z-ADD is not the justification.  The reason we need to have a
hyphen (or some other character that is not valid in a variable name) is
that we can't add an opcode that might be the same as a variable in an
existing program, because of upward compatibility.  When the compiler
begins parsing a free-form statement, it looks at the first name, and if
it's not the name of an opcode allowed in free-format, then it parses
the statement as an eval or callp statement.  If it is the name of an
opcode allowed in free-format, it parses the statement for that opcode. 
If we take names that would formerly have been parsed as eval or callp
statements and suddenly start parsing them as an evalmap statement, the
compiler would not be upward compatible.