I'm most of the way through rewriting the URL loading for Emacs/W3. In the
process, I'm making the URL library much more standalone. This means that
it would be much easier to include just it in the base XEmacs tarball and
use it for package retrieval, etc. This way a user would be able to stick
FTP and HTTP URLs in the package directory list, and all would be well
with the world.
The new HTTP loader is 100% asynchronous, and does all the work of waiting
for headers, decoding 'chunked' transfer-encodings, parsing headers, etc,
with after-change-functions. It will be up to the calling application to
deal with making it a synchronous operation.
I'm probably going to be hacking on this late tonight, so I hope to have
something I can ship out to the w3-(beta|dev) and xemacs-beta lists