Perl and UTF-8

[Scot L. Harris]

On Mon, 2004-06-21 at 14:27, J. Erik Hemdal wrote:
> Hi Scot:
> 
> > I seem to recall some limited discussion previously regarding 
> > the UTF-8 support that was introduced as far back as RH8.  I 
> > recently ran into this problem again on FC2 when installing a 
> > package that uses perl.  
> > 
> 
> What is the problem you have with Perl?  Or is this a problem with
> Fedora?  Sorry if I seem dense, but I don't understand the issue at
> hand....and other posts seem to be talking past the problem.
> 
> > To resolve the problem I needed to edit the 
> > /etc/sysconfig/i18n file and remove the .UTF-8 from the LANG variable.
> 
> This sounds more like a limitation in what you were trying to install
> rather than a Perl problem.
> 
> > 
> 
> Erik

It appears that perl's regex stuff does not play nice with UTF-8.  I
don't know exactly why but the results are that some (can't say all)
packages that use perl makefiles and some perl scripts that use regex
fail to operate as expected if UTF-8 is used.  If UTF-8 is removed or
LANG=C is exported then the perl scripts work correctly.  There does
appear to be patches for the Red Hat perl packages but they apparently
have not been released/back ported yet.

The bugzilla cases 112339 and 87682 never really come out and explain
what the actual problem is.  There are a couple of different workarounds
that have worked for me.  It is just getting old to have to apply the
workaround on every system that gets built.  Plus I don't know what
turning off UTF-8 support might break going forward.

-- 
Scot L. Harris
webid at cfl.rr.com

I would like to electrocute everyone who uses the word 'fair' in connection
with income tax policies.
		-- William F. Buckley