Mailman 3 May 2016 - Matplotlib-devel

matplotlib v2.0.0b1
by Thomas Caswell June 12, 2016

June 12, 2016

Folks, We tagged the first beta for v2.0.0 tonight. Please check out the new defaults! This is tagged as a beta because we anticipate a longer than normal release cycle. The style changes are substantial and we want to make sure that we have not crippled any common use cases. The target for the final release around scipy. A build of the docs is available at http://matplotlib.org/2.0.0b1 Preview conda packages are available via conda install -c conda-forge/label/rc -c conda-forge matplotlib pre-release packages should be on pypi tomorrow. Brief release notes: This previews the new default style and many bug-fixes. A full list of the style changes will be collected for the final release. In addition to the style change this release includes: - overhaul of font handling/text rendering to be faster and clearer - many new rcParams - Agg based OSX backend - optionally deterministic SVGs - complete re-write of image handling code - simplified color conversion - specify colors in the global property cycle via `'C0'`, `'C1'`... `'C9'` - use the global property cycle more places (bar, stem, scatter) There is a 'classic' style sheet which reproduces the 1.Y defaults: import matplotlib.style as mstyle mstyle.use('classic') A big thank you to everyone who worked on this release. Tom

4 5

Should we add manylinux wheels for previous versions of matplotlib?
by Matthew Brett June 11, 2016

June 11, 2016

Hi, As y'all probably know, we can now build reasonable manylinux wheels for matplotlib. For the upcoming 1.5.2 there is a firm plan to ship manylinux wheels. I also proposed putting up wheels for the current (1.5.1) and older versions - see [1]. My argument for doing this, is that it's relatively common for people to want to check that their code is compatible with older versions of matplotlib, especially using continuous integration testing on services like travis - and that putting wheels up on pypi would make this much easier to do. For example, there are now historical manylinux wheels up of numpy, scipy and cython up on pypi. Thomas C worried that this might cause problems. Numpy did get one complaint about the historical wheels (see [2]). We numpiers decided to continue providing the old numpy wheels on the basis that the person's use-case was unusual, and it was easy for him to solve the problem in a clean way by forcing a source install with the `--no-binary` flag to pip. So - the benefits of providing matplotlib wheels for old versions are easier installation of old versions for testing, or from pip requirements files that name specific versions. The risks are that someone with a more complex CI setup might run into problems we haven't yet thought of. Any thoughts / votes on this? Best, Matthew [1] https://github.com/matplotlib/matplotlib/issues/6473 [2] https://github.com/numpy/numpy/issues/7570

2 4

REL: v1.5.2rc2
by Thomas Caswell June 1, 2016

June 1, 2016

Hey folks, I just tagged rc2. This includes a some-what aggressive backport of the reworking of how we link into Tk to decouple from the exact version of tk we are compiled against (https://github.com/matplotlib/matplotlib/pull/6442) As such, please try to build this on your your systems and verify that mpl builds and works correctly with the tkagg backend. A major motivation for including this in the 1.5.2 release is that we can now release wheels for all of the major platforms. Much of the credit for this should go to the tireless Matthew Brett. The new rc (and the old one) should be available (assuming the CI passes) for conda via conda install -c conda-forge/channel/rc matplotlib Tom

3 5

Setting the mathtext font via the object oriented API?
by Nathan Goldbaum June 1, 2016

June 1, 2016

Hi all, This is in reference to an issue I opened today: https://github.com/matplotlib/matplotlib/issues/6514#issuecomment-222833315 Is there a way to tell the mathtext renderer to use the computer modern font that used to be the default? I know I can do this via the mathtext.fontset rcparam, but this is in the context of a library that depends on matplotlib, so I'd prefer a way to do this without overriding any rcparams a user has set already. For non-mathtext text, this is easy, I just need to set the FontProperties instance associated with the Text object. However, there doesn't seem to be a way to do something similar for mathtext. Thanks for your help, Nathan

2 2

Re: [Matplotlib-devel] [Matplotlib-users] matplotlib v2.0.0b1
by Thomas Caswell May 31, 2016

May 31, 2016

python -m pip install --upgrade pip # upgrade pip to latest pip install --pre matplotlib Should now work on all platforms. Tom On Tue, May 31, 2016 at 6:31 PM Skip Montanaro <skip(a)pobox.com> wrote: > > A build of the docs is available at http://matplotlib.org/2.0.0b1 > > Thanks. Is there a "What's New" page in the docs? Or maybe an upgrade > guide which identifies incompatibilities with 1.5? > > Skip Montanaro > who still remembers John Hunter's interview presentation at TradeLink, > lo these many years ago... >

1 0

final 2.0b1 work
by Thomas Caswell May 28, 2016

May 28, 2016

Folks, If you have anytime over the (US) holiday weekend, please look at anything in https://github.com/matplotlib/matplotlib/issues?q=is%3Aopen+milestone%3A%22… I would like to tag a beta Sunday or Monday. Tom

3 8

markers disappear using 'seaborn-notebook' style in v1.5
by Pierre Haessig May 27, 2016

May 27, 2016

Hi, I just noticed an issue with markers using the 'seaborn-notebook' style in v1.5. Usual markers like '+' and 'x' are invisible. It comes from the line 13 in the stylesheet: lines.markeredgewidth: 0 https://github.com/matplotlib/matplotlib/blob/v1.5.1/lib/matplotlib/mpl-dat… Is this intentional or just a typo ? Should I send a (one line) PR ? I also made a notebook with plots while investigating the issue: https://gist.github.com/pierre-haessig/7bcceb99257da94a04525011cb2e81c3 It seems that other seeborn styles are not affected. best, Pierre

3 4

Manylinux external library test case - matplotlib, tk
by Matthew Brett May 24, 2016

May 24, 2016

Hi, I am building matplotlib manylinux wheels: https://github.com/matthew-brett/manylinux-builds/blob/master/build_matplot… The problem is, what to do about the default matplotlib tk optional dependency? Normally, matplotlib builds against the installed system tk libs : https://github.com/matplotlib/matplotlib/blob/master/setupext.py#L1522 When building a manylinux wheel, we can only build against the tk libs / headers on the manylinux docker image, by doing `yum install -y tk-devel` before building. This means that the tk that matplotlib is built against, is different from the one that Python uses with `import Tkinter`. I suppose that is what causes the following, when building matplotlib with tk-devel installed: >>> import matplotlib >>> matplotlib.get_backend() u'TkAgg' >>> import matplotlib.pyplot as plt >>> plt.plot(range(10)) [<matplotlib.lines.Line2D object at 0x7f2a8e4b4390>] >>> plt.show() Segmentation fault I don't immediately see a good way to deal with this. Any thoughts? Cheers, Matthew

8 34

Offer of additions to subplots
by Christopher Field May 23, 2016

May 23, 2016

I love pyplot.subplots and use it to create almost all of my graphs. I have implemented code for two subplots enhancements. Because I have never contributed to an open source project, I wanted to clear the additions before I put in the effort to learn how to make a contribution and implement all of the supporting documentation and samples. I found that was often writing Fig, ax = plt.subplots( N, M, squeeze=False ) ax = ax.flatten() or ax=ax.ravel() My addition provides a squeeze=‘flatten’ option, which always returns a 1D numpy array. This is useful under three situations. First, there is a 2D arrangement of plots that are to be populated by indexing in a single loop. Second, there is a 1D arrangement of plots of variable number which, if squeeze=True, might return a single axis or a numpy array. Third, there is a 2D arrangement of plots that you wish to access with a single subscript. Does providing squeeze with bool or string values violate any rules? What is the clearest string to use? I have considered ‘flatten’, ‘flat’, ‘ravel’, and ‘1d’. The second enhancement permits sharing axis between different figures. It extends the sharex and sharey arguments to accept either an axis instance (in which case all of the new plots share the same x or y limits), or 1D or 2D numpy arrays of axis instances (in which case rows/columns are shared or each plot shares with the corresponding plot passed in. Should I invest the effort required to submit these? Is there a time frame in which these additions would be more or less welcome? Thank you, Christopher Field

4 3

v1.5.2rc1 tagged
by Thomas Caswell May 22, 2016

May 22, 2016

Folks, I have tagged a rc for 1.5.2, please check it out and kick the tires. Once the conda-forge PR is merged I will make a wider announcement to the user list. If things go well the plan is to next weekend tag a 1.5.2 final and a 2.0.0b1 Tom

1 0