[Nipy-devel] nipy without pynifti

Gael Varoquaux gael.varoquaux@normalesup....
Mon Nov 23 11:48:07 CST 2009

On Mon, Nov 23, 2009 at 08:27:36AM -0800, Matthew Brett wrote:
> >> Great suggestion. Indeed, the code is stable, eventhough many of us may
> >> not be satisfied by its quality (I am not blaming anyone, just talking
> >> about my own work).

> Well, belatedly, let's try the release-early release-often scheme!

Yuo, I like that scheme.

> >> I think that the NeuroSpin branch is indeed ready for merging. It has
> >> seen a lot of code improvements. You know, the devil is in the details,
> >> and these details have been worked on.

> That's good to know.  Will there be a candidate for merge soon?

We have to fix a failure in the design matrices on recent sympy versions
(actually, the failure is probably not due to neurospin code, but the
neurospin tests expose it), and then we will be good for a merge. I
suspect we will need until wednesday.

> Ah - users.  My guess is that, at present, we have the developers as
> the main users, and others download and experiment.   I've a suspicion
> that some of those who download NIPY are using just the core image
> reading and writing, hence my desire for a release, to give the new
> nifti code a place to sit until the recoding is done.

> >> Also, I can help for the release, whether it is mid december or end of
> >> december/january, regardless of whether CanICA goes in or not.

> That would be very good indeed, I'd appreciate the help.

OK, just coordinate with me on tasks when the time comes.

> I'm also not decided on whether to wait a little.  If we waited, I
> would spend the time on pynifti - and that might be the best way.
> I'm open to thoughts.

I think that the feeling of the people I talked to here was to get on
release fast, and to try and get a new one not too far afterward. That
would mean that CanICA would go in in between those releases. Feedback


More information about the Nipy-devel mailing list