Re: [Jython-dev] python test targets

[Samuele P. <ped...@st...>]

Frank Wierzbicki wrote:
> Brian,
> 
> The developer guide suggests using Python 2.2 as the base, but since
> (I think) the ultimate goal over the next year is to try to reach
> parity with CPython, I'm wondering if, when we are looking at fixing
> up a failing test case, we should take a look at the CPython in CVS
> and see what can be done with the latest test case.
> 
> To take a specific case, test_types.py is pretty close to functioning
> at against CVS CPython.  There are a few problems with Longs and
> "buffer" appears to be completely unimplemented in Jython (I'm
> guessing this is deliberate since buffer has been in CPython for a
> long time but is doomed in Python 3000).  If I could manage to get
> that test to pass, which would include conditionally ignoring the
> buffer tests, I could try to get the upstream CPython developers to
> accept the conditional ignore of buffer in their test_types.py, which
> would be nicer than maintaining our own.

I have commit access to the CPython CVS, so that's not the problem.
But I would not bother with that until we have progressed more.
Targetting a moving target like the CPython HEAD is not a good idea,
it would be nice to target 2.3/2.4. PyPy itself is targeting 2.3 right 
now and will move at some point to 2.4.


> 
> Thoughts?

Catching up with CPython is not that hard once the foundation is in 
place. And for some single features it may make sense to target feature
parity with 2.4 directly. But if we cannot pass a substantial part of 
2.2 or 2.3 test_descr we have even less chances with the head. The 
biggest differences between 2.2/2.3/2.4 are about pickling, at 
test_descr level, and for example sre changes. Until we have a strategy 
for bool, we pass a big chunk of test_descr and pickling stuff 2.3 or 
the HEAD does not make a big difference.

To target 2.3 we need to clean up the Py(Java)Class/Py(Java)Instance 
cluster, because to implement proper dispatch on bool we need to modify 
those and it is far better to start from a clean base. But there's still
work to finish new-style classes to do.


> 
> P.S.  Is CPython still the accepted way to refer to the original Python?
> 
> Thanks,
> Frank
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Jython-dev mailing list
> Jyt...@li...
> https://lists.sourceforge.net/lists/listinfo/jython-dev