Charles Lane wrote:


> Well, *I* certainly prefer the :: form, but by a tiny margin.  Beats
> the hell out of me why it was changed (or maybe it was always that way
> and us VMSers were always incompatible, but we just now noticed).  The
> earliest version that I have sitting around with $Config{extensions}
> is 5.005.56, and it has the "::" form.

With I perl 5.005_02 built for Solaris I do see '/' as in:

% perl -V:extensions | fold
extensions='B Data/Dumper Fcntl IO IPC/SysV NDBM_File ODBM_File Opcode POSIX 
SDBM_File Socket attrs re Errno';

note the 'Data/Dumper' and 'IPC/SysV'.


> That said, *whatever* the separator is I'd think we should stick with
> a single platform-independant standard...it's simple enough to do
> s#::#/#g; (or s#/#::#g) when you're building Perl and writing the
> $Config{extensions} string; it's much more of a pain to get a zillion
> module authors on the same wavelength.
> 
> I don't recall any discussion on P5P about this...makes me wonder if there
> really was a deliberate decision.

Beats me.

> > I am not too attached to either approach and I do realize that my reply to you
> > might have inspired you to take the approach you did.
> 
> Well, it was either a choice of "patch these module tests" or "patch
> VMS build code"...making the module tests incompatible with the rest
> of the Perl world wasn't going to work, so so it seemed best to get
> with the standard.
> 
> > What do you think about changing the regular expression?
> 
> It just puts off the day of reckoning ;)

Until we switch it to /File::Glob/i on unix even :-)

> From what I can see, $Config{extensions} is relatively new, but I would
> be willing to bet that it will be used more and more heavily as people
> try and get better integration of modules in Perl.  So maybe we patch
> the [.lib]glob-*.t tests today, but we'll be patching ad infinitum
> unless we can get the rest of the Perl community to switch to "::".
> 
> I think it's unlikely to change, but if you want to argue it on p5p you've
> got my support.

<sarcasm>
Another argument for p5p.  Hmmm that's just what they need.  :-)
</sarcasm>

This would require a tr or sed patch to the Bourne shell rendition of 
Configure.  I'll see if I can ___ er  ___ let's just see.

Peter Prymmer

Reply via email to