On a semilog plot, this is not correct, which can be seen by doing some interpolation:

The red line is not interpolated at all and is completely wrong; the dashed blue line is howax.plot works, which also uses no interpolation but keeps the right endpoints only; the green line is howaxline would work with a bit of extra points (instead of just 0,0->1,1). As can be seen in the figure, this requires a bit of finesse with the sampling as moving the view around will change the smoothness of the curve depending on how close it is to the asymptotes.

I guess it might be possible to tweak the sampling to get this right, though I don't want to spend too much time trying until we're sure what the intended API/effect is here.

loglog plots look okay since lines are still lines on them.

This is pretty neat! Givenplot is technically broken on semi-log plots (as shown above), I think adding the currentaxline version that works on log-log and linear-linear would be fine.

A tentative suggestion on arguments, how about following forms:

def axline(a, b, *args, **kwargs):    """    a, b: scalar or tuple        Acceptable forms are        y0, b:            y = y0 + b * x        (x0, y0), b:            y = y0 + b * (x - x0)        (x0, y0), (x1, y1):            y = y0 + (y1 - y0) / (x1 - x0) * (x - x0)    ...

Accept following combinations:

• intercept, slope
• point, slope
• point, point

This may cover most forms as suggeted by#5253 .

Surely, we may assign default value toa, b as you do.

@QuLogic do you have any plans to work on this? If not let me know and I'm happy to pick it up. I think it would be a cool and useful feature to add.

I don't really have time at the moment, so feel free to prepare something based on it.

Closing in favour of#9321