Re: [Pytables-users] how to 'cast' a pytable row

[Ivan V. i B. <iv...@ca...>]

David Worrall (el 2007-12-03 a les 00:28:15 +1100) va dir::

> Excellent - thank you both -
> I had forgotten the python  'row[:] ' technique since I migrated a few=20
> years ago to using tuple() and list().
>
> And the second way is probably faster for computation, no? ,  - they're=
=20
> going in and out of a couple of dicts

Well, the implementation of both is little more than one ``.item()``
call away. :)  However, I'd be carefull when using NumPy records as dict
keys, since they are hashable but also mutable::

    In [37]:a =3D numpy.rec.array([(1,'a'), (2,'b')])
    In [38]:r =3D a[0]
    In [39]:rr =3D r.copy()
    In [40]:d =3D {r: 'x'}
    In [41]:d
    Out[41]:{(1, 'a'): 'x'}
    In [42]:d[r]
    Out[42]:'x'
    In [43]:d[rr]
    -----------------------------------------------------------------------=
----
    <type 'exceptions.KeyError'>              Traceback (most recent call l=
ast)
   =20
    /home/ivan/<ipython console> in <module>()
   =20
    <type 'exceptions.KeyError'>: (1, 'a')
    In [44]:r['f1'] =3D 'z'
    In [45]:d
    Out[45]:{(1, 'z'): 'x'}

> Beautiful tool BTW, thanks!

Glad to hear that! :)

::

	Ivan Vilata i Balaguer   >qo<   http://www.carabos.com/
	       C=E1rabos Coop. V.  V  V   Enjoy Data
	                          ""