Plotly backend: surface(X,Y,Z) has x-axis in wrong direction

[dlfivefifty]

pyplot()
X=linspace(0.,1.,40).*ones(40)'
T=ones(40)*linspace(0.,1.,40)'
Y=T.*(1.-X)
Z = X
surface(X,Y,Z)

looks fine, while

plotly()
surface(X,Y,Z)

has the x-axis in the wrong direction, see attachment.

Untitled2.pdf

[mkborregaard]

The input args X and Y in Plots are not mesh-grids / matrices; just Vectors. Do you mean something like

X=linspace(0.,1.,40)
Y = X .* (1-X)
Z = X .* ones(40)'
surface(X, Y, Z)

?

[dlfivefifty]

No since the plotting grid is a triangle. It works fine in PyPlot, as seen in the attachment, so it's clearly not the case that they have to be vectors

…

On 29 May 2017, at 15:56, Michael Krabbe Borregaard ***@***.***> wrote: The input args X and Y in Plots are not mesh-grids / matrices; just Vectors. Do you mean something like X=linspace(0.,1.,40) Y = X .* (1-X) Z = X .* ones(40)' surface(X, Y, Z) ? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[mkborregaard]

Interesting. AFAIK this is undefined behaviour that just happens to work in pyplot, @tbreloff most likely knows this. If it's a small fix to get it working in plotly too that's definitely worth looking into.

[dlfivefifty]

@tbreloff added this behaviour after I requested it: #96

[mkborregaard]

Ah! Thanks for pointer, and this

It will likely only work with PyPlot

also explains the current issue. I'll add it to the list of desirable enhancements for the other backends as well.

[dlfivefifty]

Except it does work in Plotly almost perfectly, which suggests it was added at some point

…

Sent from my iPhone

On 29 May 2017, at 16:26, Michael Krabbe Borregaard ***@***.***> wrote: Ah! Thanks for pointer, and this "It will likely only work with PyPlot" also explains the current issue. I'll add it to the list of desirable enhancements for the other backends as well. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[mkborregaard]

Perhaps, I'm not sure. The pyplot functionality was added in ed7cc9b , it doesn't look awfully complicated.

[mkborregaard]

Looks like it was added to plotly in response to #479 , there is a nice example of plotting on a sphere. I guess on a sphere noone'd notice if the x axis was off. I wonder if this is the cause of this issue as well: #643

[mkborregaard]

This turned out to be a very long goose chase. You're just not viewing the planes from the same angle. Using your code:

Your first screenshot

Your second screenshot

I've had it with Plots issues for a few days now.

[mkborregaard]

I have opened a new issue with an appropriate name and am closing this #875.

[dlfivefifty]

OK, I think there is an issue here but it's not what I said: it's the plotly is putting the y-coordinate axis in the "wrong" direction:

x = linspace(-3.,3.,100)
y = linspace(0,3.,101)
surface(x,y,x' .+ y)  # axes are not consistent with gr()

[dlfivefifty]

Ah I'm being stupid: it's just a different vantage point, like you said.