does xmllib support non-UTF-8 charsets?

[Skip Montanaro]

I'm using xmlrpclib 0.9.9 (similar to what will ship with 2.2 I think).  The
database on the server contains a fair bit of Latin-1 stuff, mostly in the
form of European surnames, cities and addresses.  After updating to 2.1 on
my development machine I was left with only the xmllib-based
xmlrpclib.SlowParser class because I hadn't yet installed sgmlop in my 2.1
tree.

I was getting tracebacks like this

    Traceback (most recent call last):
      File "<stdin>", line 1, in ?
      File "/usr/local/lib/python2.1/site-packages/xmlrpclib.py", line 569, in __call__
        return self.__send(self.__name, args)
      File "/usr/local/lib/python2.1/site-packages/xmlrpclib.py", line 683, in __request
        if transport is None:
      File "/usr/local/lib/python2.1/site-packages/xmlrpclib.py", line 607, in request
        headers
      File "/usr/local/lib/python2.1/site-packages/xmlrpclib.py", line 641, in parse_response_gzip

      File "/usr/local/lib/python2.1/xmllib.py", line 168, in feed
        self.goahead(0)
      File "/usr/local/lib/python2.1/xmllib.py", line 264, in goahead
        self.syntax_error('illegal character in content')
      File "/usr/local/lib/python2.1/xmllib.py", line 794, in syntax_error
        raise Error('Syntax error at line %d: %s' % (self.lineno, message))
    xmllib.Error: Syntax error at line 1472: illegal character in content

However, on the server end, it was generating xml that begins with

    <?xml version='1.0' encoding='ISO-8859-1' ?>
    <methodResponse>
    <params>
    <param>
    <value><array><data>
    <value><int>20</int></value>
    <value><array><data>
    <value><array><data>
    <value><string>MusicEntry</string></value>
    <value><struct>
    <member>

It appears that xmllib is not paying attention to the encoding attribute.  I
realize this is a deprecated module.  Still, is this behavior expected?

-- 
Skip Montanaro (skip at pobox.com)
http://www.mojam.com/
http://www.musi-cal.com/