pyplot(ticks=:native) doesn't use native ticks

[jonniedie]

Details

I'd like to use pyplot(ticks=:native) so my ticks and grid lines will update when I zoom. It seems to do some floating point label weirdness though:

using Plots
pyplot(ticks=:native)
plot(rand(100))

When I use PyPlot directly, this doesn't happen:

using PyPlot
plot(rand(100))

Backends

This bug occurs on ( insert x below )

Backend	yes	no	untested
gr (default)			x
pyplot	x
plotly			x
plotlyjs			x
pgfplotsx			x
inspectdr			x

Versions

Plots.jl version: v1.6.8
Backend version (]st -m): v2.9.0
Output of versioninfo():

julia> versioninfo()
Julia Version 1.5.3
Commit 788b2c77c1 (2020-11-09 13:37 UTC)
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Xeon(R) E-2276M  CPU @ 2.80GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-9.0.1 (ORCJIT, skylake)

[ma-laforge]

Yes. I've noticed that too. That's very strange. It shouldn't technically be an issue with setting the "major formatter" to display all digits (because 1 is precisely represented using floating point values).

As a temporary workaround, you could use Plotly(JS) or InspectDR.jl which also are interactive backends.

I haven't tried Plotly, but I know InspectDR.jl (my own backend) doesn't suffer from this issue when ticks:=native.

Please be aware that in InspectDR:

• panning is done with shift+click+mouse drag, and
• zooming is done with a right-mouse-button box zoom.

(NOTE: I am not really part of Plots.jl's core team, so I don't really have much more to contribute, sorry.)

[isentropic]

@jonniedie I'm able to reproduce this with commit head half a year ago, probably something else has changed, not really the pyplot.jl backend code

[isentropic]

However the zooming updates the numbers on the ticks right?

[isentropic]

I believe the issue is due to Showoff.jl as it gives the tick locations. Would someone be able to test out JuliaGraphics/Showoff.jl#31 with Ryu enabled to see if this persists?

[isentropic]

This is very curious and I have no idea what is happening. Also, the digit problem is so much more apparent now. I'm a little lost why this started happening recently, as I remember native ticks to be working fine. Perhaps matplotlib itself changed or something? I'm running

v"3.3.2"

[isentropic]

Perhaps @daschw has some ideas, I'm lost with this issue

[isentropic]

@jonniedie
82f31c4