matplotlib / matplotlib.github.com Goto Github PK

matplotlib.org should redirect to matplotlib.org/latest/, which should redirect to matplotlib.org/3.0.2/ or whatever "latest" is. Then people making links can link the versions link and it won't die if we rearrange doc pages (and it will clearly say what the link is).

The proper way to do this would be with an .htaccess file, but GitHub pages doesn't support that.

So far, looks like we can't do this with soft links.

We could do with http-equiv redirect. But that has to be in the header of each file which kind of means having two versions of the docs lying around doing the same thing, but with one version pointing to the canonical version. https://www.w3.org/TR/WCAG20-TECHS/H76.html

Another option is to migrate to readthedocs https://readthedocs.org

In #49 we added banner at the top of all the old html pages that is

/* "Go to released version" message. */
#unreleased-message {
  background: #d62728;
  box-sizing: border-box;
  color: #fff;
  font-weight: bold;
  left: 0;
  min-height: 3em;
  padding: 0.7em;
  position: fixed;
  top: 0;
  width: 100%;
  z-index: 10000;
}

#unreleased-message + div {
  margin-top: 3em;
}

#unreleased-message a {
  color: #fff;
  text-decoration:underline;

}

For the 3.5.0 push /index.html is currently redirected to /stable/index.html

I think the plan was to move mpl-brochure-site/index.html to here. Maybe the redirect script needs to be updated? Maybe the release notes need to be updated?

The really really great xkcd homepage of matplotlib somehow misses the respective xkcd calls in the gallery codes. I pressume the whole page is more a nice gimmick instead of a real source, however at least some comment may be appropriate that the listed code will not give you the xkcd styled image.

I'm really looking forward to 1.3.0 cause I love the xkcd style 😄

(logging this issue here since things get lost on gitter very soon)

The sphinxsidebar in some pages lets the text overflow, ref:
https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.pyplot.tight_layout.html

Possible Solution:

@brunobeltran: adding overflow_wrap to the appropriate element in the custom mpl css (https://developer.mozilla.org/en-US/docs/Web/CSS/overflow-wrap)
if we increase sidebar width, I assume it would have to be responsive, and that's a pretty big rabbit-hole to go down. on the other hand specifying overflow-wrap would probably be an immediate improvement (and we would probably want it to be set regardless of if we eventually increase the width)

/cc @timhoffm

I was expecting info on where the code is hosted and how one can contribute in the 'Open Source' section of the index page of the project, but there's none listed there. I eventually found my way here via Google.

I saw now there is one link to 'the github tracker' (i.e. the issues of the project's repo on GH) included in the Documentation > Need help? section of the index page, but 'help' was obviously not what I was looking for.

I'd suggest linking to the two most relevant repos – the one for the project and the other for the website documenting it, i.e. this repo – in the 'Open Source' section of index.html. I could make a pull request if you're accepting contributions? (There's, unfortunately, no README or CONTRIBUTING file with info for potential contributors and the two open issues are rather old...)

Hi,

Is there any plan for matplotlib binaries for Python 3.4 ? :)

Thanks by advance for the answer !

As far as I can see

matplotlib.org is still 1.5.1
furthermore http://matplotlib.org/1.5.2 is missing

For instance, http://matplotlib.org/style_changes.html

https://matplotlib.org/cmocean/

links to

http://gcoos2.tamu.edu/gandalf_data/deployments/tamu/unit_540/plots/sci_water_temp.png
http://gcoos2.tamu.edu/gandalf_data/deployments/tamu/unit_541/plots/sci_water_temp.png

etc but they don't exist

We want the header of the current docs to point to the stable docs as canonical so search engines find those: <link rel="canonical" href="https://matplotlib.org/stable/index.html" />. This didn't happen for 3.5.0 and 3.5.1. Not sure if the fix is a doc-release python script or a fix to the doc build. Note that we have make_redirects_links.py that added the link to 3.4 and earlier, but I guess we don't want to run that whole hassle each time.

The what's new page has gotten huge. A toc at the top would be hugely valuable with maybe a sublevel.

See: https://matplotlib.org/api/axis_api.html#ticks-tick-labels-and-offset-text

This issue is to discuss possible readthedocs migration. I've set up a bare cloned repo for these tests: https://github.com/jklymak/mpldoc

Test on readthedocs. After I got the requirements file to build things in the correct order (had to add ./ to the top of requirements/doc/doc-requirements.txt) I got a real build. Looks terrible, but seems feasible once one figures out how to use our own styling....

https://mpltest2.readthedocs.io/en/latest/tutorials/index.html

The link

# https://www.ibm.com/developerworks/community/blogs/jfp/entry/My_Christmas_Gift

in matplotlib.github.com/3.2.1/gallery/showcase/mandelbrot.py does not work any more

When you try to access the linked content, IBM tells you

Where is the IBM Developer Answers (formerly developerWorks Answers) forum?
The platform was sunset on 30 April 2020. Much of the content was migrated to the IBM Support forum. Links to specific forums will automatically redirect to the IBM Support forum. You can find migrated questions by using the forum’s search mechanism or by choosing the product or topic tag.

I tried to find the linked content using IBM Support forum's search mechanism to no avail.

Thank you ፨ gb

It says: "Sorry, could not import Basemap.".

http://matplotlib.org/_images/plotmap.png
Copy: http://i.imgur.com/1DMGFHF.png

On http://matplotlib.org/1.2.0/index.html the changelog link is dead.

Hello,

Could you please provide an exact copy of the matplotlib.org because an offline version of the matplotlib docs simply does not work properly.

ADRIAN

When I want to find the matplotlib gallery, the first google result is https://matplotlib.org/gallery.html

This redirects to https://matplotlib.org/2.0.2/gallery.html , which prompts me to click through to the stable docs:

However that leads me to https://matplotlib.org/stable/ and I have to hunt for the link to the gallery on that page. As a familiar user this chain of several clicks is annoying, and I imagine that to a new user this is extremely confusing. Or they would miss the banner and stay on the old documentation site.

So, three points here:

1. Is there a reason the gallery redirect goes to the old docs?
2. This may be an issue with other top-level urls and could use an audit, I haven't checked others.
3. Is there a way of redirecting to the corresponding stable documentation page from the old docs (if it exists), rather than to the documentation main page?

In 3.4.0, the docs for some Axes methods (pick, pickable, get_picker, set_picker, get_contains, set_contains) were removed. This is because the docs are inherited from the Artist methods.

Sphinx automatically switches all references when generated the docs, but our redirect/canonicalization script here doesn't recognize it. Instead, it thinks these methods were last documented in 3.3.4 and points there.

We should decide if we want to point to the new location (and figure out if it's possible), and do so.

image link is broken on matplotlib.dates
http://matplotlib.org/api/dates_api.html

I want to load and install matplotlib in NVIDIA TX2 with archiconda, but if I install it with "pip3 install matplotlib==3.3.4", the package only can be download, can‘t be installed, if I install with "conda install matplotlib==3.3.4", the package can install successfully, but I can't import packages,there is a example:

python3
import matplotlib
illegal instruction(core dump)

Note: I tried add "export OPENBLAS_CORETYPE=ARMV8" into ~/.bashrc, but there is no help.
Can you help me to solve this problems?

In 3.4.0, we added a redirect for /tutorials/introductory/customizing.html from /users/customizing.html. The latter file was previously deleted after 2.0.2. When 3.3.4 was latest, we said the canonical link was 2.0.2/users/customizing.html, but now that a file has returned in that location, the canonical link is to /stable/users/customizing.html, which is a only meta-refresh to the right file.

I have not committed these changes. We should decide on what to do here, but if those old files will stop pointing to 2.0.2, then they should point to the final location instead of a meta-refresh.

In the Installing section of the documentation, Canopy is listed as a scientific Python distribution for Windows.
However, Canopy seems to have reached end-of-life (see its download page and an announcement by Enthought). There seems to be an alternative called "EDM", though the Matplotlib documentation should probably not refer to a product that is no longer available.

I hope this is the correct repository to report this.

3.5.0 somehow does not have the "canonical" links in the header, so it is not often trending above stable in google search. Probably worth running the script to fix this?

New pages (3.5.1, and those on 3.5.2) have the correct canonical, as do all the old ones....

3.5.1 and 3.5.0 do not have a "old version of Matplotlib" banner, however. Do we want that?

Hi,

I have a problem using matplotlib and am struggling to get to the bottom this. The problem is there are 2 seperate Figure objects displayed on the screen. Each Figure object contains an Axes object. How can I overwrite / replace the Axes object in Figure 1, with the Axes object from Figure 2?

In other words:

1. We have some data called A.
2. We plot the data called A.
3. The plot is returned in a new Figure and displayed.
4. We have some data called B.
5. We plot the data called B.
6. Now we want to copy and overwrite the Axes containing data B into the Axes containing data A

import matplotlib.pyplot as plt
import numpy as np

plt.ion()

# Draw the first figure and plot
x_1 = np.linspace(0, 10*np.pi, 100)
y_1 = np.sin(x_1)

figure_1 = plt.figure()
axes_1 = figure_1.add_subplot(111)
line_1, = axes_1.plot(x_1, y_1, 'b-')
figure_1.canvas.draw()

# Draw the second figure and plot
x_2 = np.linspace(0, 9*np.pi, 100)
y_2 = np.sin(x_2)

figure_2 = plt.figure()
axes_2 = figure_2.add_subplot(111)
line_2 = axes_2.plot(x_2, y_2, 'r-')
figure_2.canvas.draw()

# Now clear figure_1's axes and replace it with figure_2's axis
figure_1.delaxes(axes_1)
figure_1.add_axes(axes_2)
figure_1.canvas.draw()

Using the add_axes() method is raising ValueError "The Axes must have been created in the present figure".

Is there a way to detatch the Axes from the current figure so that it can be added to the other one?