Re: [xml] PATCH: XPOINTER and friends

[Daniel Veillard <veillard redhat com>]

On Fri, Sep 29, 2006 at 05:39:08PM -0500, Ross J. Reedstrom wrote:
> Daniel and all:

  Hi,

 I finally had time to work on it.

> I'm writing on behalf of Brian West, Stefan Anca and myself.  By merging
> all our brains, we've come up with a candidate implementation of python
> wrapper code for returning XPATH_POINT, XPATH_RANGE, XPATH_LOCATIONSET.
> We've choosen to represent a POINT as a tuple of a node and an integer
> offset into the content of the node. A RANGE then is a list of POINTs,
> and a LOCATIONSET is a list of RANGEs.
>
> I've attached a diff against CVS head (from repository at
> anoncvs.gnome.org:/cvs/gnome/libxml2 ), with new files as separate
> attachements.

  The patch looks just fine to me, I applied it and commited it to CVS,
thanks a lot !

> There's a bit of impedance mis-match between xpointer's 1 based offsets
> to points between characters, and python 0 based indexes into character
> array-like strings. We chose not to attempt to adjust those.
>
> The final hurdle was to fixup xpathObjectRet() to not assume any list
> was a list of nodes (since we've got ints in there as well). This did
> cause one change in the API from the python side: previously, tuples of 
> nodes were returned as a list. We preserve the tupleness of those now.
> Looks like all other uses of python tuples are in error handlers or as
> arguments to PyEval: none were passed to xpathObjectRet.
>
> A student working with us even coded up some tests, which I've attached
> as well: they need to be dropped in python/tests. They may need fixing
> up to match your test machinery. Right now, they're a bit chatty.

  Right the test as is can't really be commited since it breaks the no output
assumption of 'make tests' in python. Basically the python code would have
to embbed the logic of checking the output, instead of relying on the user
to garantee it's correct (and honnestly it's really not trivial). Can you
please fix tstxpointer.py to work that way and send it back ?
I also had to roll back the python/test/Makefile.am change to not reference
tstxpointer.py since I can't commit it as is.

  thanks !

  thanks !

Daniel

-- 
Red Hat Virtualization group http://redhat.com/virtualization/
Daniel Veillard      | virtualization library  http://libvirt.org/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine  http://rpmfind.net/