[PYTHON MATRIX-SIG] naming conventions

[Jean-Bernard ADDOR]

So, it seems that intcode(8) would continue to be supported, it would be 
great to save space.

How to detect overflows (and correct by setting to the minimum or maximum 
value)?

	Jean-Bernard


On Wed, 14 Aug 1996, Jim Hugunin wrote:

> Jean-Bernard ADDOR wrote:
> > 
> > On Wed, 14 Aug 1996, Jim Hugunin wrote:
> > >
> > > complexcode(32)
> > > floatcode(32)
> > > intcode(16)
> > > ...
> > >
> > What about floatcode(16), I have arrays of more than 1 000 000 elements,
> > I need to save memory.
> 
> Float16 will not be suported.  Float32 and Float64 are the only sizes of
> float that I know are supported natively on architectures that I'm
> familiar with.  I would guess that Float128 will work on Crays (and
> others?)
> 
> I have no desire to support any data types other than the native C ones
> (and complex numbers which are too important to ignore).
> 
> Besides, if you really want to store only 16 bits of precision you
> should probably be using fixed precision numbers anyway (and Int16 will
> work just fine).
> 
> -Jim
> 
> =================
> MATRIX-SIG  - SIG on Matrix Math for Python
> 
> send messages to: matrix-sig@python.org
> administrivia to: matrix-sig-request@python.org
> =================
> 

=================
MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
=================