[HT2HTML-devel] class-private considered harmful

[David G. <go...@py...>]

I'm thinking about reimplementing the Docutils website with ht2html.
Having looked at ht2html a bit, I think I see what to do.  There are a
few places where I'd like to subclass ht2html's default behavior
though.  For instance, I'd like to put a Google search box in the
banner.  I still want to use Banner.Banner.get_banner to do it, but
much of that class is double-underscore-class-private.  Which makes it
a royal pain to subclass.

What's reason for all this __privacy?  I can see that with so many
mixin classes, there's a risk of namespace collision.  But that's
pretty easy to avoid.  Any other reason?

Any objection to refactoring the guts of ht2html to remove all this
__privacy?  My first thought is to replace class-private method and
attribute names with single-underscore names, qualified with the class
name.  So Banner.__links becomes Banner._banner_links.  Basically what
class-privacy does, but in a non-magical form.

-- 
David Goodger <http://python.net/~goodger>