Repository navigation
scatter drops points if xscale is activated TOO LATE #3059
Description
I'm using matplotlib 1.3.1 with GTKagg backend. I found that if I call ax.set_xscale before calling ax.scatter everything works fine. However if I change the order of the commands, most points are not displayed. Here is some sample code:
fig, ax = plt.subplots(1, 2)
n = 1000
x = np.abs(np.random.laplace(size=n))
y = np.arange(n)
ax[0].scatter(x, y)
ax[0].set_xscale('log')
ax[1].set_xscale('log')
ax[1].scatter(x, y)
This leads to the following output: http://imgur.com/xqsR9fr
I'm not sure if this is the same as #2872 or not, maybe someone who has the current master installed can test that.
Activity
tacaswell commentedon May 13, 2014
This is an issue of auto-scale not working as well as it should. Calling
makes the two axes identical.
tacaswell commentedon Jun 4, 2014
Punting to 1.4.x.
I am no longer sure this is a bug as I am not sure if scatter should call auto-scale or not.
cimarronm commentedon Sep 6, 2014
PR #3481 should resolve this issue
tacaswell commentedon Aug 11, 2020
On current master and with v3.3.0 this produces
Close (and not chasing back to when we actually fixed this...)
QuLogic commentedon Aug 11, 2020
It's fixed in 3.2.2 at least, so I'd wager it's the lazy autoscaling work in 3.2.0.