Error: Multiple colormaps found for plot

[ffevotte]

Hi, it looks like the latest release (GLMakie v0.8.10) broke something

• are you running newest version (version from docs) ?
  • the issue arises when updating to GLMakie v0.8.10 or WGLMakie v0.8.14. Older versions seem to work fine.
• can you reproduce the bug with a fresh environment ? (]activate --temp; add Makie)
  • yes, a fresh environment containing only GLMakie seems to be the easiest way to reproduce the issue
• What platform + GPU are you on?
  • I'm on Ubuntu with an NVIDIA GeForce RTX 3050 card.

GPU info

$ nvidia-smi 
Mon Sep 25 08:21:46 2023       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.86.05              Driver Version: 535.86.05    CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA GeForce RTX 3050 ...    Off | 00000000:01:00.0 Off |                  N/A |
| N/A   43C    P0             ERR! /  60W |      8MiB /  4096MiB |      0%      Default |
|                                         |                      |                  N/A |
+-----------------------------------------+----------------------+----------------------+
                                                                                         
+---------------------------------------------------------------------------------------+
| Processes:                                                                            |
|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
|        ID   ID                                                             Usage      |
|=======================================================================================|
|    0   N/A  N/A      3291      G   /usr/lib/xorg/Xorg                            4MiB |
+---------------------------------------------------------------------------------------+

The issue I'm having is triggered by a code heavily inspired by the "volumeslices example" in the Makie documentation. The only addition I made (which seems to be relevant) is a Colorbar:

using GLMakie

x = let t=0:0.02:2π
    [sin(x)*cos(y)*cos(z) for x in t, y in t, z in t]
end

function volumeslices(x)
    (m, n, p) = size(x)
    fig = Figure()

    # Volume slices
    ax = LScene(fig[1, 1][1, 1], show_axis=false)
    plt = volumeslices!(ax, 1:m, 1:n, 1:p, x,
                        colormap=:bluesreds)
    Colorbar(fig[1, 1][1, 2], plt)  # This seems to be the problem

    # Sliders
    sgrid = SliderGrid(
        fig[2, 1],
        (label = "yz plane - x axis", range = 1:m),
        (label = "xz plane - y axis", range = 1:n),
        (label = "xy plane - z axis", range = 1:p),
    )
    sl_yz, sl_xz, sl_xy = sgrid.sliders
    on(sl_yz.value) do v; plt[:update_yz][](v) end
    on(sl_xz.value) do v; plt[:update_xz][](v) end
    on(sl_xy.value) do v; plt[:update_xy][](v) end

    set_close_to!(sl_yz, m÷2)
    set_close_to!(sl_xz, n÷2)
    set_close_to!(sl_xy, p÷2)

    fig
end

volumeslices(x)

The code above works fine with WGLMakie v0.8.13 (my originally intended backend) and GLMakie v0.8.9 (which I tested afterwards to check whether this was backend-dependent). But an error arises after upgrading to WGLMakie v0.8.14 or GLMakie v0.8.10:

ERROR: LoadError: Multiple colormaps found for plot Combined{Makie.volumeslices, Tuple{UnitRange{Int64}, UnitRange{Int64}, UnitRange{Int64}, Array{Float64, 3}}}, please specify which one to use manually. Please overload `Makie.extract_colormap(::Combined{Makie.volumeslices, Tuple{UnitRange{Int64}, UnitRange{Int64}, UnitRange{Int64}, Array{Float64, 3}}})` to allow for the automatical creation of a Colorbar.
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] extract_colormap_recursive(plot::Combined{Makie.volumeslices, Tuple{UnitRange{Int64}, UnitRange{Int64}, UnitRange{Int64}, Array{Float64, 3}}})
   @ Makie ~/.julia/packages/Makie/1hq9u/src/makielayout/blocks/colorbar.jl:79
 [3] Colorbar(fig_or_scene::GridSubposition, plot::Combined{Makie.volumeslices, Tuple{UnitRange{Int64}, UnitRange{Int64}, UnitRange{Int64}, Array{Float64, 3}}}; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})                                                                                                                                     
   @ Makie ~/.julia/packages/Makie/1hq9u/src/makielayout/blocks/colorbar.jl:86
 [4] Colorbar(fig_or_scene::GridSubposition, plot::Combined{Makie.volumeslices, Tuple{UnitRange{Int64}, UnitRange{Int64}, UnitRange{Int64}, Array{Float64, 3}}})
   @ Makie ~/.julia/packages/Makie/1hq9u/src/makielayout/blocks/colorbar.jl:84
 [5] volumeslices(x::Array{Float64, 3})
   @ Main ~/projets/DataViewer.jl/bug/bug.jl:15
 [6] top-level scope
   @ ~/projets/DataViewer.jl/bug/bug.jl:36
 [7] include(fname::String)
   @ Base.MainInclude ./client.jl:478
 [8] top-level scope
   @ REPL[3]:1
in expression starting at /home/francois/projets/DataViewer.jl/bug/bug.jl:36

All in all, I was surprised because I wasn't expecting anything to break for these releases. On the other hand, I'm very new to Makie, so it might very well be that I was using the Colorbar in an unintended way in the first place...

What do you think?

And many thanks for the great packages and the tremendous amount of work you put in them!

[ffevotte]

Looks like it's related to #3090

[jkrumbiegel]

All in all, I was surprised because I wasn't expecting anything to break for these releases.

This was an internal refactor, however Makie's API surface and combinatorial possibilities are very large. This case was not covered by tests otherwise it would've been fixed beforehand, it was not broken intentionally. The only thing that should have stopped working after the refactor were previously not well-defined cases that we classified as bugs basically.

[ffevotte]

This case was not covered by tests otherwise it would've been fixed beforehand, it was not broken intentionally.

Yes, of course. Sorry, I didn't mean to intend (at all!) that anything was broken intentionally. My comment was probably unclear, but what I meant to say was that I wasn't sure to which category this belonged:

1. an (unintended) breakage, and therefore a bug in Makie (which is very fine! Again, thanks for your hard work!)
2. an unintended use of Makie's API, i.e. a "not well-defined case" in my code (in which case I'd be more than happy to fix it and I'd be sorry to have bothered you...)

Is there anything I can do to fix this? After skimming through #3090, I have the (vague) impression that what would be needed is some sort of specialized Makie.extract_colormap method that works with volumeslices plots. Is that correct? Do you think it could be a "good first issue" for me to try and fix in a PR?

[SimonDanisch]

Yes that sounds correct and should be a good first PR :)

[ffevotte]

OK, I'll try and do this! Hopefully by the end of this week...