On 2013-04-18, Edward K. Ream wrote:
thanks for the report.
The strange thing is that in both sets we see a lot of characters that
are definitely no delimiters (and not in the hard-coded set). There seems
to be a problem with encoding for printing. It seems your terminal does not
support utf8. This makes it hard to spot the difference.
Differences can be safely ignored for production. They might be a false
positive or a hint that it is time to update the pre-generated strings.
However, I cannot reproduce the problem here (Debian/testing, Python 2.7,
Docutils 0.11svn, locale de_DE.utf8.