This and other RFCs are available on the web at
http://dev.perl.org/rfc/
=head1 TITLE
caller->eval BLOCK
=head1 VERSION
Maintainer: David Nicol <[EMAIL PROTECTED]>
Date: 28 Sep 2000
Mailing List: [EMAIL PROTECTED]
Number: 339
Version: 1
Status: Developing
=head1 ABSTRACT
C<caller> is extended to allow complete access to the "call frame" of
the current subroutine call.
=head1 DESCRIPTION
Some new syntax is introduced, involving the C<caller> keyword,
that allows evaluation of expressions from the point of view of the
situation from which the current method was called. After thinking
about it for a while I think
caller->eval EXPRESSION;
would work well, with the prototype and semantics of this construction being
identical to regular C<eval>, except that resolution of names is done as
if the eval was happening where the call happened.
Neat tricks become possible by directly accessing the immediate symbol
table of the calling context.
=item caller->eval and the hastily destructing return
Also, if we adopt
<a href=http://www.mail-archive.com/perl6-internals@perl.org/msg01442.html>
a hastily destructing C<return>
</a> (or even if we don't, but easier if we do) we might be able to
break out of loops in the calling context, for instance
return caller->eval{return undef} if $WeAreDone;
would, in the case of C<$WeAreDone> being true, make a first pass over
the expression to be returned, incrementing the reference counts therein,
in this case evaluating "caller" for a value, but not running its eval method,
then destroy the subroutine context (except for rescued values, which there
are none of here) and then evaluate the returned expression.
In this case, the returned expression is a flow control directive, which
will be followed, returning an undefined value from the calling routine.
In this case, prefixing C<caller->eval> with C<return> might be optional.
=head1 IMPLEMENTATION
C<caller> is altered to return a tied scalar that stringifies to
the "caller's package name" if defined, to match perl5. This object
has some other methods though, in particular an "eval" method which
is described above.
we will still need to keep track of the names of lexical variables that
have been lost to packed offset optimizations, in case they get referred
to by name from within a called function.
If C<caller->eval> contains flow control directives which take us out
of the calling context, the current context must be destroyed
=head1 REFERENCES
perldoc -f caller
How To Implement Tail Recursion:
http://www.mail-archive.com/perl6-internals@perl.org/msg01442.html
