Re: updates

On Mon, 2003-10-27 at 01:12, seth vidal wrote:
> > I would suggest option 3.  Just assume any character > 127 is evil if
> > the string isn't UTF-8, and replace it with '?'.  Technically you can do
> > better, but it probably isn't worth the time.

Actually the unicode() function can already do this for you:

unistr = unicode(str, 'iso-8859-1', 'replace')

> The question that Joe raised, correctly, is - what do you do for a file
> name, where it may actually matter.
> 
> I've already hit this in the fedora core rpms.
> 
> What are the standards and rules for this, in general. Should we be
> encouraging some sort of policy of 'only utf-8 filenames' or somesuch?

I think encouraging a policy of UTF-8 filenames is the only thing we can
do, short of additional encoding data per-file in the RPM header.

Joe