Thank you for spotting this. I fixed it in CVS.
Great! Thanks.
> John: my fix contains the fancy expression
> re.sub(r"[^ -~\n]", lambda x: r"\%03o"%ord(x.group()),
> s)
> to quote all non-ASCII characters. Is this safe with all
> supported Python versions or do I need to be more portable
> here?
It looks valid to me across 2.2 - 2.4. The best solution though, is
to test it. How about a unit test called "whacky_chars" which makes
titles, xlabels and ylabels as if they came from a foul-mouthed comic
strip character. We could add this to unit test dir and test it
across backends.
Thanks,
JDH