Add style for xkcd

[souravsingh]

Added a style file for xkcd
Fixes #5992

[mdboom]

Shouldn't these be numbers, rather than the placeholders "scale, length, randomness"?

[mdboom]

This is missing the patheffect that puts a white border behind all lines.

[afvincent]

Ping @souravsingh : are you still interested working on this style sheet? With @mdboom's remarks, I guess one should get something that starts to resemble xkcd plots :).

[souravsingh]

@afvincent I am interested in working on the style sheet. I am confused about the parameter for patheffect

[WeatherGod]

The patheffect part is probably going to be the hardest part at the moment,
since it doesn't exist as a style parameter yet.

On Mon, Jul 18, 2016 at 9:22 AM, Sourav Singh notifications@github.com
wrote:

@afvincent https://github.com/afvincent I am interested in working on
the style sheet. I am confused about the parameter for patheffect

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#6157 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AARy-Kj819aprk-HzdGTI4qYEXge__T_ks5qW34ogaJpZM4HwCQT
.

[afvincent]

@WeatherGod You're right: the patheffect part might be a bit more difficult to setup than I expected :/. I was fooled by the example of rcParams['path.effects'] = [patheffects.withStroke(linewidth=4, foreground="w")] where the relevant rcParams key exists. But patheffects is actually loaded elsewhere before…

[afvincent]

@souravsingh Apparently, there is an official xkcd font, so it could be nice to add it as the default choice (another PR is open to add it to plt.xkcd, see #6811 for details). I don't have a solution for the patheffect part yet, but I'll try to have a look at this problem during the weekend.

[souravsingh]

@afvincent Is the problem fixed?

[afvincent]

Wohh I totally forgot this one : apologies @souravsingh! The xkcd font seems to have been added to the list so this part should be OK.

Unfortunately, I cannot see an easy solution to the patheffect part 😞. The validator of the corresponding rcparam (rcParams['path.effects']) would require to eval a string like "patheffects.withStroke(linewidth=4, foreground='w')", which may raise security issues if I am correct. But maybe a core dev. will prove me wrong ^^.

[dopplershift]

@afvincent Well, we control that string, and patheffects, so it's certainly possible to create some kind of string syntax that we can then parse ourselves without using eval....

[jklymak]

If we parsed path.effect as a string like "withStroke(linewidth=4, foreground='w')" and then did something like:

patheffect = getattr(patheffects.modules[__name__], str)
rc['path.effect'] = patheffect(**kwargs)

would that be safe enough? ping @anntzer

[anntzer]

Please no more parsing craziness.

[souravsingh]

Is there a resolution for the PR?

[anntzer]

I hope the resolution will be #9528.

[jklymak]

I'll close in favour of #14943