Use xkcd: prefix to avoid color name clashes. #6390

anntzer · 2016-05-09T17:52:56Z

Split out of #6382.

Without the prefix, merging CSS4_COLORS and XKCD_COLORS becomes order-dependent.

It's also unclear why space-less names are needed, but I can put them back if someone feels strongly about it...

Otherwise, merging CSS4_COLORS and XKCD_COLORS becomes order-dependent. It's also unclear why space-less names are needed, but I can put them back if someone feels strongly about it...

tacaswell · 2016-05-09T19:55:50Z

doc/users/colors.rst


 All string specifications of color are case-insensitive.

 Internally, mpl is moving to storing all colors as RGBA float quadruples.
-
-Name clash between CSS4/X11 and XKCD


Please leave this plot.

Even if we are explicitly name-spacing the colors, I think this chart is still interesting to include in the docs.

Are you OK for explicitly namespacing everything then? It'll always be easier to add un-namespaced versions later than removing them...

tacaswell · 2016-05-09T19:57:36Z

I am still in favor of leaving both the prefixed and un-prefixed versions. For names that are only in the xkcd name space there is no reason to include the extra 5 characters.

We should also document the names as 'XKCD:...' as I think that is his 'preferred' branding.

anntzer · 2016-05-09T20:10:34Z

I can change the prefix to XKCD, no problem there.
The whole point of the PR is so that {**CSS4_COLORS, **XKCD_COLORS} and {**XKCD_COLORS, **CSS4_COLORS} give the same results, and I'm sort of hoping that this can become an "official" namespacing mechanism for named colors (perhaps seaborn can do its color-remapping by putting its colors in the seaborn-{deep,muted,...}:<...> namespace and we can even have a mechanism to change the named colors resolution order, e.g. plt.colors.set_name_resolution_order(["seaborn-deep", "css4", "xkcd"])).

tacaswell · 2016-05-09T20:35:43Z

One of the things we decided in the discussion on the pr where the XKCD
colors went in was that we should not change the meaning of the css colors.

On Mon, May 9, 2016, 16:10 Antony Lee notifications@github.com wrote:

I can change the prefix to XKCD, no problem there.
The whole point of the PR is so that {**CSS4_COLORS, **XKCD_COLORS} and {**XKCD_COLORS,
**CSS4_COLORS} give the same results, and I'm sort of hoping that this
can become an "official" namespacing mechanism for named colors (perhaps
seaborn can do its color-remapping by putting its colors in the
seaborn-{deep,muted,...}:<...> namespace and we can even have a mechanism
to change the named colors resolution order, e.g. plt.colors.set_name_resolution_order(["seaborn-deep",
"css4", "xkcd"])).

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#6390 (comment)

anntzer · 2016-05-09T20:37:05Z

Then the "base" namespace could cover both the one-letter codes and the CSS4 names?

tacaswell · 2016-05-09T20:43:20Z

I am hesitant to combine the single letter colors (which are home grown)
and the css4 colors (which is a published spec) for the sake of keeping the
code tidy (where tidy here is clarity of what things are) rather than a
minimum number of variables.

The other concern is that we can not encode the single letter colors as hex
due to float rounding issues where as the CSS4 colors are defined as RGB
hex values.

On Mon, May 9, 2016 at 4:37 PM Antony Lee notifications@github.com wrote:

Then "base" could cover both the one-letter codes and the CSS4 names?

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#6390 (comment)

anntzer · 2016-05-09T20:47:24Z

TBH I don't feel strongly about anything here, except that having a name clash by default looks... unappetizing. A namespacing/priority mechanism (say, "base", "css4", "xkcd", "seaborn-<...>") would not, by default, change the meaning of the CSS colors (but it would be possible for the user the change them -- in a more "ordered" manner than seaborn currently changes the base colors.

anntzer · 2016-05-11T04:04:06Z

By the way, regarding xkcd vs XKCD: http://xkcd.com/about/

For those of us pedantic enough to want a rule, here it is: The preferred form is "xkcd", all lower-case. In formal contexts where a lowercase word shouldn't start a sentence, "XKCD" is an okay alternative. "Xkcd" is frowned upon.

tacaswell · 2016-05-11T04:13:45Z

huh, fair enough re xkcd. Could have sworn I read the opposite....

tacaswell · 2016-05-11T04:14:24Z

lib/matplotlib/tests/test_colors.py

@@ -563,7 +563,7 @@ def test_xkcd():
        mcolors.colorConverter.to_rgb('blue'))
    assert x11_blue == '#0000ff'
    XKCD_blue = mcolors.rgb2hex(
-        mcolors.colorConverter.to_rgb('XKCDblue'))
+        mcolors.colorConverter.to_rgb('xkcd:blue'))


This does read much better.

tacaswell · 2016-05-11T04:17:03Z

I am also 👍 on this modulo a few comments.

anntzer · 2016-05-11T05:19:38Z

I have no idea what's wrong with this build...

tacaswell · 2016-05-12T03:06:02Z

doc/users/colors.rst

-   j = 0
-
-   for n in sorted(overlap, reverse=True):
+   for j, n in enumerate(sorted(overlap, reverse=True)):


This means you can also drop the j += 1 below.

tacaswell · 2016-05-12T03:14:45Z

That failure was a bug in pyparsing 2.1.2 which just died on import on 3.4, it is fixed multiple ways already.

@efiring Thoughts on this naming change?

I would be fine if this were merged and it was my job to fix that plot (but would not be opposed to @anntzer fixing it 😉 )

efiring · 2016-05-12T03:48:07Z

I like the use of 'xkcd:' as a prefix for all xkcd colors.
I'm confused by the references to "space-stripped" names. The html names are all stripped, but this ugliness is somewhat reduced in http://www.w3schools.com/colors/colors_names.asp by the use of CamelCase. (Leaving out spaces has its advantages, so I'm not complaining about the html names.) The xkcd colors are a mess, with "dark blue" (00035b) and "darkblue" (030764). I suspect whoever thought that was a good idea was high at the time. Overall, the color matching in xkcd is good, but there are too many of them, and too many with revolting names. But we are stuck with that, and I have digressed.

tacaswell · 2016-05-12T04:05:46Z

I completely missed that there is both 'dark blue' and 'darkblue'. The downsides of machine learning + crowd sourcing....

anntzer · 2016-05-12T04:09:01Z

Here's the offenders:

In [4]: sorted([c for c in XKCD_COLORS if " " in c and c.replace(" ", "") in XKCD_COLORS])
Out[4]: 
['xkcd:aqua marine',
 'xkcd:blue green',
 'xkcd:blue grey',
 'xkcd:dark blue',
 'xkcd:dark green',
 'xkcd:egg shell',
 'xkcd:golden rod',
 'xkcd:green blue',
 'xkcd:grey blue',
 'xkcd:light blue',
 'xkcd:light green',
 'xkcd:orange red',
 'xkcd:terra cotta',
 'xkcd:yellow green']

Anyways, I guess this answers the question...

tacaswell · 2016-05-12T21:32:38Z

I will fix that figure my self in a follow on PR

API: Use xkcd: prefix to avoid color name clashes.

tacaswell · 2016-05-12T21:37:50Z

backported to 2.x as c2b6769

Use xkcd: prefix to avoid color name clashes.

3c1d4e6

Otherwise, merging CSS4_COLORS and XKCD_COLORS becomes order-dependent. It's also unclear why space-less names are needed, but I can put them back if someone feels strongly about it...

mdboom added the status: needs review label May 9, 2016

tacaswell added this to the 2.0 (style change major release) milestone May 9, 2016

tacaswell reviewed May 9, 2016
View reviewed changes

tacaswell reviewed May 11, 2016
View reviewed changes

tacaswell reviewed May 12, 2016
View reviewed changes

Restore CSS/xkcd comparison plot.

30ffa63

anntzer force-pushed the xkcd-colors-namespace branch from aa9822a to 30ffa63 Compare May 12, 2016 03:43

tacaswell merged commit d03e798 into matplotlib:master May 12, 2016

tacaswell removed the status: needs review label May 12, 2016

anntzer deleted the xkcd-colors-namespace branch May 12, 2016 21:34

tacaswell added a commit that referenced this pull request May 12, 2016

Merge pull request #6390 from anntzer/xkcd-colors-namespace

c2b6769

API: Use xkcd: prefix to avoid color name clashes.

anntzer mentioned this pull request Apr 5, 2017

[feature request] move default colors into styles #8430

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Use xkcd: prefix to avoid color name clashes. #6390

Use xkcd: prefix to avoid color name clashes. #6390

anntzer commented May 9, 2016

tacaswell May 9, 2016

tacaswell May 11, 2016

anntzer May 11, 2016

tacaswell commented May 9, 2016

anntzer commented May 9, 2016

tacaswell commented May 9, 2016

anntzer commented May 9, 2016 •

edited

tacaswell commented May 9, 2016

anntzer commented May 9, 2016

anntzer commented May 11, 2016

tacaswell commented May 11, 2016

tacaswell May 11, 2016

tacaswell commented May 11, 2016

anntzer commented May 11, 2016

tacaswell May 12, 2016

tacaswell commented May 12, 2016

efiring commented May 12, 2016

tacaswell commented May 12, 2016

anntzer commented May 12, 2016

tacaswell commented May 12, 2016

tacaswell commented May 12, 2016

Use xkcd: prefix to avoid color name clashes. #6390

Use xkcd: prefix to avoid color name clashes. #6390

Conversation

anntzer commented May 9, 2016

tacaswell May 9, 2016

Choose a reason for hiding this comment

tacaswell May 11, 2016

Choose a reason for hiding this comment

anntzer May 11, 2016

Choose a reason for hiding this comment

tacaswell commented May 9, 2016

anntzer commented May 9, 2016

tacaswell commented May 9, 2016

anntzer commented May 9, 2016 • edited

tacaswell commented May 9, 2016

anntzer commented May 9, 2016

anntzer commented May 11, 2016

tacaswell commented May 11, 2016

tacaswell May 11, 2016

Choose a reason for hiding this comment

tacaswell commented May 11, 2016

anntzer commented May 11, 2016

tacaswell May 12, 2016

Choose a reason for hiding this comment

tacaswell commented May 12, 2016

efiring commented May 12, 2016

tacaswell commented May 12, 2016

anntzer commented May 12, 2016

tacaswell commented May 12, 2016

tacaswell commented May 12, 2016

anntzer commented May 9, 2016 •

edited