Python 2.2 seriously crippled for numerical computation?

[Fernando Pérez]

Huaiyu Zhu wrote:

> There appears to be a serious bug in Python 2.2 that severely limits its
> usefulness for numerical computation:
> 
[snipped]

Just to give some data points on this. I confirmed the problem on a Mandrake 
8.1 box with python 2.2 built by hand, gcc 2.96. But the python 2.2 windows 
binary from python.org does NOT exhibit it in my WindowsXP box. The reuslt 
there is 1e-200**2 -> 0.0, as it should.

I posted this report to the scipy-dev list to see if someone there (many 
linux users, all directly involved in numerical computing) can find out what 
the problem is.

Cheers,

f.