This issue seems to be more complicated. When I try to explicitly set the axis limits, axis limits revert to the default (i.e. the opposite of what I want). This is independent of whether I run a plot command afterward. Also the default range is too large and default number of ticks is too high.

import matplotlib.pyplot as plt
import numpy as np
f, ax = plt.subplots()
N = 50
ax.set_yscale('logit')
# ax.set_ylim((0.1,0.9))
ax.plot(np.linspace(0,1,N), np.linspace(0.1,0.9,N))

import matplotlib.pyplot as plt
import numpy as np
f, ax = plt.subplots()
N = 50
ax.set_yscale('logit')
ax.set_ylim((0.1,0.9))
ax.plot(np.linspace(0,1,N), np.linspace(0.1,0.9,N))

As of master, the results are

So at least the number of ticks issue is solved (cf. #14512).

The original issue was fixed in #14928, but there's a small followup: when doing gca().set(yscale="logit", ylim=(.9, .999)), there is no tick at x=0.999, but it shows up if the upper limit is 0.999001 for example. Probably just a matter of < vs <=...

Is anyone else currently working on this? If not I'll give it a try.

Go for it.

Is anyone else currently working on this? If not I'll give it a try.

RiazCharania and I will not be working on this issue.

The remaining example gca().set(yscale="logit", ylim=(.9, .999)) now gives:

which seems different from the description, but hardly correct.

I think that's what I said: a tick is missing at the top (1-1e-3).