python 3.0, pywin32 and scipy

Terry Reedy tjreedy at udel.edu
Thu Aug 2 14:24:22 EDT 2007


"vml" <victor.lebrun at gmail.com> wrote in message 
news:1186070475.659511.24130 at q3g2000prf.googlegroups.com...
| I am trying to promote python in my job, my collegue only see matlab
| and microsoft scripting language.
| I understood that there willl be no backward compatibility between
| python 2.x and 3.0, does it means that:

Most of the basic language and syntax will be unchanged.  Many to most of 
the new features will be introduced in 2.6 also so you can add their usage 
gradually, as needed, without converting everything.  Most of the 
incompatibilities between 2.6+ and 3.0+ will be handled by 2to3.py.

| - my script using pywin32 for the COM layer and scipy for the maths
| won't work under 3.0
| - will we have the equivalent of pywin32 and scipy in python ?

As with all 3rd party addons, that will depend on the authors or author 
groups.  The same is true today for every new release, especially in 
regards to Windows binaries.

| I will be incharge of designing a python module which will be a
| 'matrix calculator' in our current software. Will it be compatible
| with python 3.0 ? I guess no.

But you will be able to use it *and distribute it* with Python2.5 
indefinitely.  Some people still run 1.5.2 code with the nearly decade old 
1.5.2 interpreter.  And I expect that there will be people running 2.5 code 
with Py2.5 a decade or more from now.

The PSF still distributes versions back to about the first.  Compare this 
with Microsoft pulling from distrubution the free version of VC2003 early 
in 2006, while maintaining the prohibition on anyone else distributing it. 
This has been a MAJOR nuisance for new Python-C-Windows developers who did 
not download it in time.

| What can I answer to my collegue who will say 'Python is changing and
| the stuff you are doing now is useless'?

Short answer: stuff that is useful now will continue to be useful until 
requirements change.

A possible longer answer: Google continues to make major use of Python 
without handcuffing themselves with such misunderstandings.  They hired 
Python's main developer both to write Python code for internal use and to 
continue Python's development as he saw fit.  They will be a test site for 
both the conversion process and the usability of the new version, giving 
Guido face-to-face feedback.  When both are good enough for Google's 
developers, they should be good enough for most of the rest of us too.

Terry Jan Reedy






More information about the Python-list mailing list