Sven Schreiber wrote:
> belinda thom schrieb:
>> Also, since numpy borrows from matlab,
not really -- pylab is specifically designed to be similar to matlab,
numpy is not -- and the matlib is left over from Numeric, and I don't
think it was all that well maintained there, either.
> Maybe you know that already, but in scipy there is something like
> scipy.io.read_array and write_array which is very similiar to mpl's
> load/save (IIRC).
>
> IMHO something like that would be a welcome addition to numpy, but I
> have learned that adding features to numpy is quite controversial...
> (oh, we're on the mpl list right now, ok)
It's an obvious thing to have in a comprehensive package -- maybe not so
much in numpy. numpy does have the core tools to build that sort of
thing, however.
> In the end I wrote my own csv read and write functions (see post in
> other thread), because I didn't see why my code should depend on having
> scipy or mpl installed just because of 20 or 30 lines of code.
Again, that is another example of why SciPy should be better modularized
-- what if it just depended on having scipy.io installed?
> (AFAIK numpy fromfile and tofile are for binary data, not text files.
> Don't know if you want that.)
As mentioned, fromfile and tofile do basic text file support -- but not
preserving the shape of the array. However, what it does provide is very
good performance for reading/writing lots of numbers to/from text files
-- they are great tools for building your own text file
parser/generator. In fact, your csv reader/writer and much of scipy.io
could probably benefit from using them (I don't think it does now, as
the text file support is new to numpy).
I don't know if someone has written it yet, but a load/save pair that
put a small header with type and shape info in a file, then dumped the
contents with tofile(...sep=something) could be pretty handy, and fast.
> So if you never use numpy standalone, I'd say go for mpl's load/save.
What about pickle? I'm pretty sure pickle works, doesn't require any
additional packages, and preserves the format of the array:
import numpy as N
import cPickle
>>> a = N.arange(24, dtype=N.float)
>>> a.shape = (2,3,4)
>>> a
array([[[ 0., 1., 2., 3.],
[ 4., 5., 6., 7.],
[ 8., 9., 10., 11.]],
[[ 12., 13., 14., 15.],
[ 16., 17., 18., 19.],
[ 20., 21., 22., 23.]]])
>>> cPickle.dump(a, file("test.pickle",'wb'))
>>> b = cPickle.load(file("test.pickle",'rb'))
>>> b
array([[[ 0., 1., 2., 3.],
[ 4., 5., 6., 7.],
[ 8., 9., 10., 11.]],
[[ 12., 13., 14., 15.],
[ 16., 17., 18., 19.],
[ 20., 21., 22., 23.]]])
This is one of those cases where "There should be one obvious way to do
it" is failing :-(
-Chris
--
Christopher Barker, Ph.D.
Oceanographer
Emergency Response Division
NOAA/NOS/OR&R (206) 526-6959 voice
7600 Sand Point Way NE (206) 526-6329 fax
Seattle, WA 98115 (206) 526-6317 main reception
[EMAIL PROTECTED]
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/matplotlib-users
