Minor example updates #7559
Minor example updates #7559
Conversation
| self.collection = RegularPolyCollection( | ||
| fig.dpi, 6, sizes=(100,), | ||
| 6, sizes=(100,), |
There was a problem hiding this comment.
6 -> rotation angle; not sure if the fig.dpi was intended to produce perfect circles and the 6 should have been deleted instead.
|
| @@ -12,7 +12,7 @@ | |||
| datafile = cbook.get_sample_data('goog.npy') | |||
| try: | |||
| # Python3 cannot load python2 .npy files with datetime(object) arrays | |||
| # unless the encoding is set to bytes. Hovever this option was | |||
| # unless the encoding is set to bytes. However this option was | |||
| # not added until numpy 1.10 so this example will only work with | |||
| # python 2 or with numpy 1.10 and later | |||
| price_data = np.load(datafile, encoding='bytes').view(np.recarray) | |||
There was a problem hiding this comment.
If you regenerate the npy under a recent py3, can you cover all cases?
There was a problem hiding this comment.
Not having luck with it:
$ ipython3
In [1]: import numpy as np
In [2]: np.__version__
Out[2]: '1.11.2'
In [3]: i = np.load('lib/matplotlib/mpl-data/sample_data/goog.npy', encoding='bytes')
In [4]: np.save('goog.npy', i)
$ python2
>>> import numpy as np
>>> np.__version__
'1.11.2'
>>> i = np.load('goog.npy')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib64/python2.7/site-packages/numpy/lib/npyio.py", line 406, in load
pickle_kwargs=pickle_kwargs)
File "/usr/lib64/python2.7/site-packages/numpy/lib/format.py", line 637, in read_array
array = pickle.load(fp, **pickle_kwargs)
ValueError: non-string names in Numpy dtype unpickling
>>> i = np.load('goog.npy', encoding='bytes')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib64/python2.7/site-packages/numpy/lib/npyio.py", line 406, in load
pickle_kwargs=pickle_kwargs)
File "/usr/lib64/python2.7/site-packages/numpy/lib/format.py", line 637, in read_array
array = pickle.load(fp, **pickle_kwargs)
ValueError: non-string names in Numpy dtype unpicklingWhat would probably be the most portable is np.datetime64, but unfortunately, it appears Matplotlib doesn't handle it as well as datetime.datetime. Probably because we didn't depend on a new enough NumPy to carry the converter in-tree.
There was a problem hiding this comment.
np.datetime64 has been a mess from the start. It is marginally usable now that we require numpy 1.7, so we do need to add support as best we can. Maybe someone will want to take that on for 2.1.
There was a problem hiding this comment.
np.datetime64 can be converted back to datetime.datetime with arr.astype('O'); would that look better than this current try/except stuff?
There was a problem hiding this comment.
This example doesn't use dates or times in any form! It is just plotting random-looking numbers that happen to come from some stock statistics. As a plotting example, it doesn't need to read data in from a file at all.
There was a problem hiding this comment.
Yes, but there are a couple others (just search for this same typo message in this PR) that do use the file and do use the dates, so I was mainly referring to them.
96e6e50 to
d348baf
Compare
|
PEP8 issues should be fixed. |
|
Can you please change the authorship of the last commit to @sirlittle. He's the one who put in all the effort into fixing a ticket opened by one of our own contributor which is not reflected in that cherry-pick. I think that is very unfair considering the screw up comes from our end. |
598dce8 to
c7b1b3c
Compare
|
Thanks! |
|
Updated attribution and I cleaned up the sample data for stocks. No deprecation for those as @tacaswell suggested. |
428230e to
8ce6cdd
Compare
|
Oh, forgot to comment that I undid the stock data removal on @tacaswell's request as that probably won't be going into 2.0.x. |
It just shells out to mencoder, which is less available, and we've got better demos that automatically call ffmpeg or whatever with the right arguments.
svg.embed_char_paths -> svg.fonttype axisbg -> facecolor Skip 'spectral' colormap in a demo.
input() does not return until you press Enter, so be explicit about it.
Move the large label closer to the bottom spine so it appears related; remove extra whitespace around the figure.
This keeps the location in a relatively normal place. Also, fix the text to actually be different between the two objects.
8ce6cdd to
94f47da
Compare
| @@ -25,6 +25,6 @@ | |||
| ax.set_title('only every 5th errorbar') | |||
|
|
|||
|
|
|||
| fig.suptitle('Errorbar subsampling for better visualibility') | |||
| fig.suptitle('Errorbar subsampling for better appearance') | |||
There was a problem hiding this comment.
how about just "Errorbar subsampling"?
|
This PR contains no bug fixes and large amounts of changes. These may look minor, but no change should be considered minor. This should not target 2.0 but 2.1. |
|
These are bug fixes for the examples if you consider that the style changes have made them appear inconsistent. The rest of the changes are documentation, which have traditionally been backported. The real large changes were removed and will go into 2.1 (but I haven't opened that PR yet.) |
|
The files that are not examples / doc which are touched by this are: lib/matplotlib/axes/_axes.py : correcting a docstring I missed as part of #7172 lib/matplotlib/tests/test_coding_standards.py : because more files now pass pep8 I see no reason to not put this in for 2.0 |
|
Thanks @tacaswell! |
I ran all the examples, and fixed any deprecations from matplotlib or NumPy (~the first half) and then fixed a few things that look wrong with the new 2.0 defaults or might be confusing (~the second half). Lastly, ran some spellcheck using PyCharm.