On Tue, Sep 8, 2009 at 9:23 AM, Bruce Southey<bsouthey@gmail.com> wrote:
On 09/07/2009 10:44 PM, Robert Kern wrote:
On Mon, Sep 7, 2009 at 11:35,<josef.pktd@gmail.com> wrote:
After the initial release of scikits.statsmodels, we started to have email discussions about the design, implementation and extension of statsmodels, as well as about datahandling and how to use statsmodels for different types of data. Since this discussion seemed to specialized to fill up scipy-dev with it, we decided to start a dedicated google group.
Maybe it's too late, but this is not at all too specialized for scipy-dev. We explicitly welcome all discussions about any scikit here, especially those aimed at inclusion into scipy. Keeping the discussion here may help people searching for information at a later date, particularly with respect to design decisions made early on.
I tend to agree that scipy-users and scipy-dev should be able to handle a large part of the context proposed. I find that design and implementation discussions help others develop their own software and datahandling almost is required by all areas. But I do recognize that certain aspects are very specific and that could be a basis for a separate list, perhaps along the lines of the separation between the scipy-users and scipy-dev lists.
At the end of the gsoc, in preparation for the first scikits release and to a slightly lesser extend throughout the summer, Skipper and I had a huge mail traffic. After the release, this has slowed down since we are both busy with other things, but we started to have detailed discussion about improvements to the design and extensions. We wanted to move the 2 or 3 way mailing threads to a public forum, but I felt, that we have more freedom in producing long, "boring" mail threads on a dedicated list than on the general purpose scipy lists. The second reason was, that we started to discuss pandas, which looks like a good package for data handling, used mainly in finance, and which will have its first release in the near future. I'm also keeping an eye on some other packages that might become useful for econometrics (also BSD or compatible). So, we preferred a theme oriented mailing list that gives as more freedom to also discuss packages that are not directly related to the use and development of scipy, the package. These were our initial intentions, how it will turn out, we will see. For me, it's not always easy to tell which mailing list is the best location for a discussion (which of course might not improve if there is one more). Josef
Bruce _______________________________________________ Scipy-dev mailing list Scipy-dev@scipy.org http://mail.scipy.org/mailman/listinfo/scipy-dev