Re: UTF-8 encoding for plain text

Stefan Behnel wrote:

> >    if isinstance(value, unicode):
> >       value = value.encode('utf-8')
> >    # pass value into libxml2
> >     
> Would you really want to pass the above string into libxml2, then see that it
> didn't work and finally try to make up a good error message to throw back an
> exception? It's much easier now, where Python generates that exception for you.
>   
Hmm.. so you're saying that the string -> unicode -> string round trip 
serves for detecting non-ascii characters?
Maybe you're right in that we should not accept non-ascii regular 
strings, as that is just too easy to get wrong. Your example above would 
work in my proposal, but only if the encoding of your Python source is 
set to UTF-8.
A check for any bytes >= 128 would be more efficient than to do a 
unicode round trip.
Regards,
Geert