Merge lp://staging/~facelessuser/beautifulsoup/lxml-fix into lp://staging/beautifulsoup

I'm not sure if we want to force " obj.name = name" to be None in this case? Or just assign the empty string. Let me know what you think. In that case, it may make more sense to do something like this:

    def __new__(cls, prefix, name, namespace=None):
        if not name:
            name = None

        if name is None:
            obj = unicode.__new__(cls, prefix)
        elif prefix is None:
            # Not really namespaced.
            obj = unicode.__new__(cls, name)
        else:
            obj = unicode.__new__(cls, prefix + ":" + name)
        obj.prefix = prefix
        obj.name = name
        obj.namespace = namespace
        return obj

It's a small thing, but I think it's better to make sure name is set to None. https://www.w3.org/TR/xml-names/#defaulting says: If there is no default namespace declaration in scope, the namespace name has no value.

"No value" says None to me, not the empty string. The lxml change seems to be a side effect of a change made for convenience, rather than a change made for better compliance with a spec.

> It's a small thing, but I think it's better to make sure name is set to None.

I kind of figured that would be the preference :). The more I had thought about it the more I thought it would make sense to force the convention to remain the same as it always was. I can update accordingly.

Never mind, it sounds like you committed a fix already :).