statsmodels-0.9.0dev0

93 views
Skip to first unread message

Big Stone

unread,
Sep 24, 2017, 5:43:08 AM9/24/17
to pystatsmodels
Wouldn't it be nice to tag a fresh statsmodels-0.9.0dev0 release on github ? (a "dev" series of releases, like github/bokeh team does)

It would not take a minute to the release manager, and help propagate recent codebase, and bug fixes

josef...@gmail.com

unread,
Sep 28, 2017, 11:21:49 PM9/28/17
to pystatsmodels
On Sun, Sep 24, 2017 at 5:43 AM, Big Stone <stone...@gmail.com> wrote:
Wouldn't it be nice to tag a fresh statsmodels-0.9.0dev0 release on github ? (a "dev" series of releases, like github/bokeh team does)

It would not take a minute to the release manager, and help propagate recent codebase, and bug fixes

What's the purpose of this? Is it for a pre-release or for a reasonably stable state of the repo during the development cycle?

Who is picking this up? Is this just for users that explicitly download that version, similar to directly installing master or is this also picked up without explicit version request?


Current state: master is all green,
No testing yet with scipy 1.0 and one incompatibility with it fixed in master. No **known** incompatibilities with numpy, scipy and pandas right now, AFAIK.

If my patience with maintenance keeps up, then we have one more week of reducing the bug-fix and merge backlog including one serious change (GLM weights), a few more promised enhancements and a few more easier bug fix and maintenance merges.

That is master is something like statsmodels-0.9.0a1 right now.

Right now we are at 160 open PRs, down from around 130 after 0.8 was released.

Josef




Big Stone

unread,
Oct 4, 2017, 2:43:17 PM10/4/17
to pystatsmodels
Hi Joseph,

Thanks for the lonesome work.

I downloaded the October 1rst master, to get something compatible with pandas-0.20.0+

I hope NumFocus/Anaconda/BloomSomething will come to the rescue before you give up entirely.

josef...@gmail.com

unread,
Oct 4, 2017, 3:54:04 PM10/4/17
to pystatsmodels
On Wed, Oct 4, 2017 at 2:43 PM, Big Stone <stone...@gmail.com> wrote:
Hi Joseph,

Thanks for the lonesome work.

To avoid giving a wrong impression:

Only parts of reviewing PRs is lonesome work.
In general, we have a good number of contributors and several maintainers.
Especially Kevin is doing a lot of maintenance work, and without his work it would be a lot more difficult to keep statsmodels running.

Josef
Reply all
Reply to author
Forward
0 new messages