Front page | perl.perl5.porters |
Postings from August 2000
We could consider having 'undef' mean to break the current branch,
while other types of false merely set $^R as they do now.
But before we do anything, perhaps we should find out the reasoning that
went into making it the way it is. Who added this feature? Ilya?
Jeffrey
Dominic Dunlop <domo@computer.org> wrote:
|> At 05:01 -0600 2000-08-04, Tom Christiansen wrote:
|> >Agreed. This has been previously observed, at least by some of us.
|> >In fact, to some of us, that [a zero-width assertion] was not
|> >[capable of being unsuccessful] came as quite a shock
|> >(myself included). It belies the term "assertion". Abortable blocks
|> >should be on the table for perl6 -- at least, and maybe earlier, too.
|>
|> Amplifying (part of) what Hugo said, as the pod (still) says:
|>
|> `(?{ code })'
|> WARNING: This extended regular expression
|> feature is considered highly experimental, and
|> may be changed or deleted without notice.
|>
|> the feature could be fixed to work 'right' as soon as anybody has a
|> tuit. ASAP would be best: for now, AFAICT, nothing in the
|> distributed lib tree uses it (except, unsurprisingly, re.pm). I
|> can't trawl through the whole of CPAN to see if it's used much or at
|> all there, but I suspect not. But, as soon as MRE II hits the
|> streets, my bet is that it'll see more use, however it works and
|> whatever its status. The 5.6.0 distribution also has the fine
|> perlretut.pod, which could also encourage its use as currently
|> implemented, even though it's careful to warn of its experimental
|> status. (As it happens, I _think_ all its examples would still work
|> even after the proposed change, because the code blocks all look as
|> if they would to return a perlish 'true'.) So I say (without
|> volunteering) "Do it now -- ideally in the maintenance track
|> (provided we're damned sure it works."
|> --
|> Dominic Dunlop
|>
|>