From: Russell O. <ro...@uw...> - 2012-09-18 23:48:16
|
On Sep 18, 2012, at 4:05 PM, Paul Hobson wrote: > On Tue, Sep 18, 2012 at 3:57 PM, Russell E. Owen <ro...@uw...> wrote: >> In article <505...@st...>, >> Michael Droettboom <md...@st...> >> wrote: >> >>> I have tagged and created a tarball for 1.2.0rc1. The githash is >>> bda6dd9feab8. The tarball is on the github download page here: >>> >>> https://github.com/matplotlib/matplotlib/downloads >>> >>> I have created a new branch, v1.2.x, for continuing 1.2.x development. >>> The feature freeze on master is now lifted and big experimental changes >>> can be merged into master. Any bugfixes that need to go into 1.2.x >>> should be merged into both places. Please mark any PRs for the 1.2.x >>> branch with the 1.2.x milestone so we can verify that things are merged >>> in both places. >> >> It appears that >> import matplotlib >> no longer imports matplotlib.dates -- that I must do that explicitly: >> import matplotlib.dates >> >> Is that an intentional change? It breaks existing code of mine, which is >> easily fixed and perhaps was making unwarranted assumptions. But I >> wonder what else will break. > > Russel, > > Which version were you on? with MPL v1.1 i get: > In [28]: import matplotlib > > In [29]: matplotlib.dates > --------------------------------------------------------------------------- > AttributeError Traceback (most recent call last) > <ipython-input-29-a13aa8cf36d8> in <module>() > ----> 1 matplotlib.dates > > AttributeError: 'module' object has no attribute 'dates' > > In [30]: matplotlib.__version__ > Out[30]: '1.1.0' > > In [31]: import matplotlib.dates > > In [32]: matplotlib.dates > Out[32]: <module 'matplotlib.dates' from > 'C:\Python27\lib\site-packages\matplotlib\dates.pyc'> I was using 1.1.1 most recently. I'm not sure when I started using matplotlib.dates without explicitly importing it. In any case, it sounds as if it's not meant to work, so there's no need to change anything in matplotlib. Regards, -- Russell |