On Friday 08 April à 00:56, Mike Brown wrote: > Sylvain Thénault wrote: > > after some times working on other stuff, I've remembered that I had > > still the prepare_input_source patch pending. I've joined to this > > message the patch as it is now, and it's well working for me. I'll > > check it in in the next few days if no one objects. > > I still have not had time to review and test your changes, and don't > anticipate being able to do it anytime soon. :/
maybe just a quick look at test_saxutils.py I've joined in my previous post ? > I would just be happy to know that in PyXML, > > (1) OS-specific file system paths or relative URI references > are never used as base URIs; only absolute URIs are > (and they can be derived from what is given, in some cases) > > and > > (2) resolution of a relative URI reference to absolute form > is carried out in accordance with RFC 3986 -- e.g., > via Absolutize() in Uri.py. So be happy :) I've modified (actually simplified) the patch so that a ValueError is raised if a non empty but non absolute path is given to the absolute_system_id function, and actual absolutization is done with the Absolutize function. > Anything else, like the details of OS path <-> URI conversion, improvements > to > percent-encoding/decoding, better str vs unicode handling, IDN support, etc. > is lower priority. +1 > One thing you should do is compare your Uri.py against the current one. We've > made a number of changes since February, and some of them are important: > http://cvs.4suite.org/viewcvs/4Suite/Ft/Lib/Uri.py?r1=1.98&r2=1.89 I've checked this and updated my imported version of Uri.py -- Sylvain Thénault LOGILAB, Paris (France). http://www.logilab.com http://www.logilab.fr http://www.logilab.org _______________________________________________ XML-SIG maillist - XML-SIG@python.org http://mail.python.org/mailman/listinfo/xml-sig
