I applied the patch for PyFile locally and I'm in favor of it at this
time. It provides the necessary mechanics for instantiating a PyFile
without too much overhead of keywords (as I previously suggested) by
deferring to the underlying constructors. The biggest issue I have
with it is it's difficult to document as well as promoting the use of
an internal class. Perhaps it would be better if only 'file' or 'open'
supported this feature?
In any event, it helps pass a number of the failing regrtests. I'm
going to run the whole suite and then likely commit it.
On Jun 12, 2005, at 04:46 PM, Frank Wierzbicki wrote:
> Yes I'm going to try to get the PyString problems fixed tonight in the
> manner that Samuele and I discussed. The PyFile stuff is still under
> discussion so I'll work on that next (not likely tonight).
> On 6/12/05, Brian Zimmer <bzimmer@...> wrote:
>> Are you working on implementing these changes this weekend? It looks
>> like Samuele and you have come to agreement on the implementation,
>> On Jun 12, 2005, at 11:13 AM, Frank Wierzbicki wrote:
>>>> In this case I think it is easier to introduce a type specifier that
>>>> means None or string, such that for None it produces null, .split()
>>>> already by passing null to the 2 arg version.
>>> Let me make sure I understand. I would need to add a method to
>>> PyObject like "asStringOrNull" (which from PyObject should just
>>> asString()) and subclass that in PyNone so that it returns null, and
>>> then add a new type (probably a symbol that will replace the ? in
>>> "s?", maybe "s!") to gexpose.py that will use asStringOrNull.