> On Tue, Jun 24, gordon inscribed on the eternal scroll:
> (well, no-one seems to have offered an answer, so I suppose I might
> > The weird thing is that with a 5.8.0 compiled from source with default
> > settings, both matches work, which is what we'd expect.
> I don't see any logical reason why it would not work, so I'd rate it
> prima facie as a bug in the particular implementation that was
> giving the problem.
> Sorry, I'm not in a position to reproduce your error, so I'm neither
> confirming nor denying your report - just saying that on the basis of
> what you reported, it does seem like a bug.
> > Here's a tiny script to show the problem:
> [works for me, on several different platforms, but I didn't have the
> specific one you mentioned]
> (I think you'd need to report the release details of the RPM,
> the output of perl -V and so forth, to make it a proper bug report.)
> As you rightly say: with all of the characters involved being
> us-ascii, there shouldn't be any difference. Could it be that for
> some bizarre reason one of them got "upgraded" to unicode, and the
> other didn't, and they were then reported as not matching? But I
> might be talking rowlocks - it needs someone who understands the