Mark J. Reed wrote:
> Following up to a bug report, Mr Lenz wrote:
> Another detail: if i do this in the REPL, it works fine, until I close
>> the REPL - then I get the same backtrace as before.
> ...except that this is a Rakudo bug, and Rakudo doesn't have a REPL.
BTW there's a patch for that in RT, waiting for review and application:
> Pugs, of course, does have a P-ful REPL. This led me to wonder if there's
> any intent to include the interactive behavior of the interpreter as part of
> the spec. I can see arguments for leaving it up to the individual
> implementation, but it'd also be nice to have some consistency there.
> Maybe not require that there be a REPL, but supply some guidelines for how
> it should behave if it exists.
I think it's only worth being specified if there's really innovative
going on. The standard thing (readline lib + evaluating + printing
result or .perl of result) should be fairly common sense.
> Since Perl 5 has no REPL, I'm not sure where such a spec would go. S20,
> maybe, since the debugger is the closest thing?
or maybe S19, because it defines the console interface to the rest of
the world. Or just pick a not-yet-used number, S34, and have fun with it ;-)