develooper Front page | perl.perl5.porters | Postings from August 2013

Re: experimental: \s in regexp matches \cK

Thread Previous
Karl Williamson
August 16, 2013 15:08
Re: experimental: \s in regexp matches \cK
Message ID:
On 08/15/2013 07:37 PM, Ricardo Signes wrote:
> According to perlexperiment, \s matching \cK is experimental as of 5.18.0.
> I don't see it framed that way in perl5180delta.

I thought I had written something to that effect for a 5.17 perldelta; 
either I misremember or it got dropped in the 5.18 delta merge.

  It is, though, in
> perlrecharclass.  There, it has this comment:
>    it could be backed out in v5.20 or v5.22 if experience indicates that it
>    breaks too much existing code
> So far, I feel like we're A-OK to call it gold in 5.20.  If we'd rather wait
> until 5.22, though, I will make a ticket hung on "blocking 5.22" to remind us
> when we get there.
> Thoughts?

Our general plan for deprecations is to have two major release cycles 
before removing things.  This is motivated by the likelihood that a 
vendor will have annual release cycles like we do, and the timing could 
be off so that our latest will not get into their latest for nearly 
another year, so that an end-user wouldn't have any time to react -- 
unless we have a 2 year window.

The \cK issue isn't technically a deprecation, but it seems to me that 
we should follow this plan anyway in the absence of a good reason not 
to.  The only reason that I can think of to do it is that it avoids 
uncertainty on the part of those who are keeping up with our annual 
releases as they happen.  And I don't see that as a real issue.

I left the different code paths in, so that internally [:space:] and \s 
are handled separately, and just happen to match identically.  Thus it 
is easy to back out this change, unless code in the meantime is written 
by someone unaware of this possibility.  The amount of code involved is 
quite small.

In any event, I think it's too soon after 5.18.0 to make an informed 
decision to make it non-experimental..  I think we should either decide 
now to wait for 5.22, or wait until much closer to 5.20 before making 
the decision.

I personally doubt that there will be problems; but I've been wrong 
before -- lots of times.  It was rjbs' idea to treat this as a bug fix 
and not a switchable feature.  I do expect that history will prove that 
he made the right call, but it's still too early to be sure, imo.

Thread Previous Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About