[Pkg-exppsy-pynifti] [Nipy-devel] Image, design, usecases
Michael Hanke
michael.hanke at gmail.com
Wed May 13 07:52:43 UTC 2009
HI,
[just answering to topic which are not yet covered in my eralier
response]
On Wed, May 13, 2009 at 12:19:55AM -0700, Matthew Brett wrote:
> > Any scientist who is just
> > starting to learn Python, hoping that he will one day be able to use
> > it in a productive manner, might be discouraged when he sees that he
> > has to deal with a package of such complexity as Nipy. One day, he
> > will learn that Nipy is the right choice, but his first start, just
> > reading his precious data out of his Niftis, should be as simple as
> > possible. So why not put some basic functionality into scipy.io?
>
> That would be a good aim, but I think we should work out the design
> and get the testing done well - and then, when it's ready, propose for
> addition to scipy.io. I have no strong feelings though - we could try
> going for it straightaway...
I would do it separately first till it is perfect. I'd also continue to
ship it separately (maybe in addition to a version already included in
SciPy) till SciPy starts building recent binary packages for all major
platforms, or any other solution that allows us to post things like
'Just go and _simply_ install a more recent SciPy version to make it
work'.
> > scipy.io.vimage.load/save-Level
> > scipy.io.vimage.VolumeImage-Level
> > ?pynifti?-Level (Which would then need another name as it should
> > implement more formats)
> > Nipy-Level (The ultimate level)
> >
> > In each step, the documentation, e.g. docstring, should give a hint
> > for the next level, so whenever a user might feel the need for more
> > control, he should know where to go.
>
> Yes, but it's tricky isn't it, because, at the moment, all the heavy
> lifting is in in fact in scipy.io.vimage.VolumeImage stuff, whereas
> the scipy.io.vimage.load/save-Level, and the NIPY level, are fairly
> thin wrappers.
>
> Michael - are we getting towards the point you were making?
Yeah, progress is there, but see my other post regarding the datatype of
'meta'.
Michael
--
GPG key: 1024D/3144BE0F Michael Hanke
http://apsy.gse.uni-magdeburg.de/hanke
ICQ: 48230050
More information about the Pkg-exppsy-pynifti
mailing list