Re: 5.13.1 localized tied X no longer tied breaks File::chdir
[email protected] (Dave Mitchell)
| Newsgroups | perl.perl5.porters |
|---|---|
| Message-ID | <[email protected]> |
On Sun, May 23, 2010 at 03:19:14PM +0100, Dave Mitchell wrote:
> On Fri, May 21, 2010 at 10:30:39PM -0400, David Golden wrote:
> > I'm sorry I missed this thread the first time around or didn't realize
> > the implications:
> >
> > http://www.xray.mpe.mpg.de/cgi-bin/w3glimpse2html/perl5-porters/2010-04/msg00561.html
> >
> > Ironically, I think the original bug was reported by Schwern because
> > "local @CWD" didn't work in File::chdir. The resulting change means
> > that remaining use of local with File::chdir is now broken:
> >
> > use File::chdir;
> > $CWD = "/foo"; # it's /foo out here.
> > {
> > local $CWD = "/bar"; # /bar in here
> > }
> > # still /foo out here!
> >
> > While I co-maintain File::chdir, I don't have a strong opinion about
> > whether the changes should be reverted, but if not, I'll need to have
> > better documentation of the 5.13+ limitations.
> >
> > For reference, the relevant commits are:
> >
> > commit 191ad7eff570fc96c93993e4358f83e2033365d6
> > commit 5afa72af5aa99c40932771ad390abf5ba229611b
>
> The first commit of mine fixed @tied and %tied, upon which there was a
> consensus, especially as the existing behaviour didn't do anything useful.
> The second commit, which did the same for $tied, was a bit of an
> after-thought, with no real discussion. Given that it breaks existing
> "reasonable" behaviour, I'm now strongly tempted to revert it (but to
> leave the @tied/%tied change).
Now reverted with 89adf4e8e0de17fedf78328b32920f47635850d0,
documented with 5f3e44b689cab8bd7cec81aa6fc21c2ec5b3489f.
If people want to work to a consensus on what a LOCALISE method might do,
I'll implement it; in the meantime, I think its best to stick with the old
behaviour.
--
Justice is when you get what you deserve.
Law is when you get what you pay for.