[Numpy-discussion] Release of NumPy

[Sebastian Haase]

On Tue, Apr 15, 2008 at 10:59 AM, Jarrod Millman <millman at berkeley.edu> wrote:
> On Tue, Apr 15, 2008 at 1:35 AM, David Cournapeau
>
> <david at ar.media.kyoto-u.ac.jp> wrote:
>
> > Sebastian Haase wrote:
>  >  > How about releasing 1.0.5 without the new masked array,
>  >  > i.e.
>  >  > 1. put back the old masked array and release as 1.0.5
>  >
>  >  Now that it is done, going back will be a lot of work. It would be great
>  >  if someone were willing to do it, but who ?
>
>  I am not convinced that having a 1.0.5 (sans the new masked array
>  implementation) is very important.  So I am personally not inclined to
>  put any effort into making this happen.  Even if someone else were to
>  make the branch without the new ma, it would still require work on my
>  part.  I would need to do some testing, ask everyone else to try out
>  the releases, get the binaries built, put together release notes,
>  upload everything to sourceforge (if you haven't used sourceforge,
>  trust me it is no fun), update the scipy wiki, update the PyPI site,
>  and then send out the announcement.  This process will divert my
>  effort as well as whoever actually does the majority of the work.  It
>  will also divert the community as I ask several people to test and
>  help me prepare the release.
>
>  If there is some unknown regression in the current code, then we will
>  need to make a 1.0.6 and a 1.1.1 release.  Having to support the 1.0.x
>  and 1.1.x series as we attempt to start developing 1.2 may be to
>  onerous.  We also need to start concentrating on getting SciPy 0.7
>  released.  Moreover, I intend to put in more effort on this year's
>  conference and I intend to ask others to help.  And, of course, we all
>  have our day jobs.
>
>  So given the added effort that a 1.0.5 release would require, we
>  should think very carefully about what having a 1.0.5 release would
>  gain us before starting on that path.
>

Also *I* would never install 1.0.5 if there was already a 1.1 release.

I presume that the masked array user base is a sub-group anyway --
furthermore I think that those people would *benefit* from the new ma.
So, let's just call the numpy 1.0.5 release the one that was never
released, because we (you guys) were too enthusiastic getting numpy
into what will now be known as numpy 1.1

((It's getting late in California -- even though I'm actually in
Berlin,Germany these days .... ))

Cheers,
-Sebastian Haase