Thanks to visit codestin.com
Credit goes to github.com

Skip to content

IGNORE: Lockout new axis converters #9736

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
wants to merge 2 commits into from
Closed

IGNORE: Lockout new axis converters #9736

wants to merge 2 commits into from

Conversation

jklymak
Copy link
Member

@jklymak jklymak commented Nov 9, 2017 •
edited
Loading

Keeping here in case we want to use instead of #9776, but #9776 is my preferred solution.

PR Summary

A script like:

    base = datetime.datetime(2017, 1, 1, 0, 10, 0)
    time = [base - datetime.timedelta(days=x) for x in range(0, 3)]
    data = [0., 2., 4.]
    fig, ax = plt.subplots()
    ax.plot(time, data)
    ax.plot(['a', 'b'], [1., 2.])

would run and make a very non-sensical and mislabeled plot. This now returns a TypeError if this is done.

Note that the following will not be caught, and will result in an ugly error that pops up all the time:

    dt = datetime.datetime.fromordinal(ix).replace(tzinfo=UTC)
ValueError: ordinal must be >= 1

    base = datetime.datetime(2017, 1, 1, 0, 10, 0)
    time = [base - datetime.timedelta(days=x) for x in range(0, 3)]
    data = [0., 2., 4.]
    fig, ax = plt.subplots()
    ax.plot(time, data)
    ax.plot([0., 3.], [1., 2.])

The problem here is that the update_units function also gets called by set_xlim and set_ylim, and if these are on auto it passes bare floats. Maybe that can be fixed, but is a bigger project...

See #9713

PR Checklist

  • Has Pytest style unit tests
  • Code is PEP 8 compliant
  • New features are documented, with examples if plot related
  • Documentation is sphinx and numpydoc compliant
  • Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
  • Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way

@jklymak jklymak changed the title Lockoutnewconverters Lockout new axis converters Nov 9, 2017
story645
story645 reviewed Nov 9, 2017
View reviewed changes
lib/matplotlib/axis.py
@@ -1434,6 +1434,13 @@ def update_units(self, data):
if converter is None:
return False

if (self.converter is not None) and (self.converter != converter):
raise TypeError('Attempting to plot data that is '
'registered to be converted with %s but '
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should the language be something like "data with units %s which is incompatible with existing axis units %s" - kinda wondering if registered convertors is too weedy for users.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Open to this - I find the word "units" very confusing in this context. Looking at the jpl stuff I can see there was an attempt to let the data have different units and then get converted by what the axis units were (i.e. if xunits='centimetres' but the x-axis units were 'meters' 1.0 would be plotted as 0.01) but that seems to have been lost along the way. Or I'm misunderstanding (more likely)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that's sort of what the basic units example might be about too-same numbers, but we'll do the conversion internally. Maybe just stick to type? "attempting to plot data of type %s that is incompatiable with existing axis data of type %s"

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oooh, except it returns fun things like matplotlib.dates.DateConverter not the data type. I could state what the new type is (i.e. type blah cannot be processed by axis unit converter matplotlib.dates.DateConverter)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like that last suggestions @jklymak

FWIW a converter that converts meters should also be able convert centimeters fine (they're both lengths!)

@dstansby dstansby added this to the v2.2 milestone Nov 9, 2017
@jklymak
Copy link
Member Author

jklymak commented Nov 9, 2017 •
edited
Loading

OK, so I have a new version that actually locks out everything, but still passes most tests. I've not uploaded yet...

It fails tests where people try to override the units of the axis:

def test_fill_units():
        def test_single_date():
        time1 = [721964.0]
        data1 = [-65.54]

        fig = plt.figure()
        plt.subplot(211)
        plt.plot_date(time1, data1, 'o', color='r')

Fails because they try to pass a float time after they have set the axis to the "datetime" converter.

My tendency would be to just rewrite these (very few) tests so they behave and update this PR, but thought I'd seek input... I can't tell what level of folks require this mixing of modes. Ironically the jpl folks have such an example:

def test_fill_units():
        from datetime import datetime
        import matplotlib.testing.jpl_units as units
        units.register()

        # generate some data
        t = units.Epoch("ET", dt=datetime(2009, 4, 27))
        value = 10.0 * units.deg
        day = units.Duration("ET", 24.0 * 60.0 * 60.0)

        fig = plt.figure()

        # Top-Left
        ax1 = fig.add_subplot(221)
        ax1.plot([t], [value], yunits='deg', color='red')
        ax1.fill([733525.0, 733525.0, 733526.0, 733526.0],
                 [0.0, 0.0, 90.0, 0.0], 'b')

This was referenced Nov 13, 2017
Closed
Closed
dstansby
dstansby requested changes Dec 27, 2017
View reviewed changes
Copy link
Member

@dstansby dstansby left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am very in favor of this, since a single axis should only have a single unit being plotted on it!

lib/matplotlib/axis.py
'is incompatible with existing axis data '
'converter %s' % (converter.__class__,
self.converter.__class__))

neednew = self.converter != converter
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This line can change to neednew = self.converter == None, and maybe change the variable name neednew to needupdate too since that's really what it means now.

lib/matplotlib/axis.py
@@ -1434,6 +1434,13 @@ def update_units(self, data):
if converter is None:
return False

if (self.converter is not None) and (self.converter != converter):
raise TypeError('Attempting to plot data that is '
'registered to be converted with %s but '
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like that last suggestions @jklymak

FWIW a converter that converts meters should also be able convert centimeters fine (they're both lengths!)

@dstansby
Copy link
Member

Oh, should definitely get an API change note too.

@jklymak
Copy link
Member Author

jklymak commented Dec 28, 2017

@dstansby Did you have a chance to look at #9776 which is probably what I think we should consider using? I think this approach of locking an axis to a converter should be thought about seriously - and maybe by someone more experienced w/ the units code than me. This all came up wrt datetime code and categorical code. So, @tacaswell, @anntzer, @story645 should chime in on if this is a good protection to add or not.

@anntzer
Copy link
Contributor

anntzer commented Dec 28, 2017

Haven't thought about this recentely but IIRC making the implicit (pass-thru) converter explicit (i.e. #9776) seemed a better approach.

@jklymak jklymak changed the title Lockout new axis converters IGNORE: Lockout new axis converters Jan 3, 2018
@jklymak jklymak closed this Jan 3, 2018
@QuLogic QuLogic modified the milestones: needs sorting, v2.2.0 Feb 12, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@story645 story645 story645 left review comments

@dstansby dstansby dstansby requested changes

Assignees
No one assigned
Labels
topic: units and array ducktypes
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@jklymak @dstansby @anntzer @story645 @QuLogic