numerical packages for 1.5.2?

[eric jones]

Hey Mark,

You might be able to get SciPy working for 1.52, but I haven't tried.  I
expect it would require some effort and multiple changes.  It should be
possible though.

The array comparison stuff like a < b, etc.  was added to Numeric when
Python got rich comparisons (either 2.0 or 2.1...), so they will give
exceptions in 1.52.

I don't think there is an easy solution here.  All the SciPy development is
going on with versions 2.1 and on.  Are there a ton of people needing 1.52
support?  If it's extremely popular, we could accept patches to the current
version of SciPy that fix the incapatibilities.  I think it might be easier
to just upgrade the 10 machines though... :)

eric



"Mark Fardal" <fardal at coral.phys.uvic.ca> wrote in message
news:yj1vgfi3hwd.fsf at coral.phys.uvic.ca...
>
> Hi,
>
> I'd like to do some scientific/numerical work with Python.  On the
> computers accessible to me, Python 1.5.2 is quite common but Python
> 2.x is not to be found anywhere.  I don't have root password to most
> of these computers.  I'd also like to share my code with
> collaborators, and don't want to make them upgrade Python itself
> (which is significantly harder than upgrading Python packages).  So,
> it looks like Python 1.5.2 is it for now.  These are mostly Linux/x86
> systems with a few SGIs.
>
> Given that restriction, what version of Numeric should I be using?  Is
> there a version of SciPy that will work on 1.5.2?  Are there known bugs
> in any of these versions I should watch out for?
>
> thanks
> Mark Fardal
> University of Victoria
>
>