[Python-Dev] Request for Opinions

[Guido van Rossum]

> Even though I'm one of the 4XPath developers, I've largely stayed on the
> sidelines because I've never been on python-dev, and I don't know how
> such matters are usually considered.  Guido's response gives good
> guidance, and my ideas are expressed in-line.

Thanks for responding!

> Guido van Rossum wrote:
> > 
> > > By the way, one of the major reasons there is no clear consensus from
> > > the XML SIG yet is because we are hung up on an issue of fact. Is it
> > > technically and politically feasible to put 4XPath into the Python
> > > distribution. Our discussions over there will be radically different if
> > > everybody over here says "sure, that's no problem."
> > 
> > I have no idea what 4XPath is.  Assuming it is a separately maintained
> > and distributed 3rd party library, the answer is no for the source
> > distribution and yes for binary distributions.
> 
> Paul did suggest a couple of other options.  One was to make 4XPath
> easier to integrate into Python (by moving to SRE or otherwise removing
> the lex/yacc dependencuy), and this is a possibility.  However, I don't
> know how you gauge what a "third-party" module is.  My guess would be
> that it's a matter of copyright.  If so, that might be an obstacle, but
> maybe not an insurmountable one.  The other solution was to write an
> XPath-lite, probably all in Python to go with minidom in the Python
> distro.  This would be the path of least resistance, but we are still
> discussion on xml-sig the useful subset of XPath.
> 
> Is there any clear statement of policiy for including modules?  Even a
> mailing-list archive entry?  I don't expect one because this is
> open-source/BDFL world where the policy usually lies in a
> highly-respectable head, but it could certainly guide this conversation
> either here or on xml-sig.

My policy for including modules is that Python becomes the only
distribution point of the code, and the code is maintained through the
Python CVS tree (possibly by the original author).  This places
restrictions on the copyright notice.  There are exceptions
(e.g. cPickle.c has a life of its own in the Zope world, although the
modules frequently get re-sync'ed), but they are rare and must serve
very special purposes.

Unless I'm mistaken, 4XPath leads a separate life so can't be
included.

--Guido van Rossum (home page: http://dinsdale.python.org/~guido/)