Makie Roadmap [WIP]

[SimonDanisch]

• colorsampler interface + alpha keyword
• resizing plots with multiple array attributes consistently (e.g. scatter(rand(5), color=rand(5)))
• data space + marker space
• text + scatter shader rewrite
• WidgetsBase.jl integration
• Improve how attributes work document & implement refactor goals JuliaPlots/AbstractPlotting.jl#360
• Merge MakieLayout
• Data transforms (log axis etc)
• Merge most code shared in WGLMakie & GLMakie
• clean up recipes

[asinghvi17]

I wonder if WidgetsBase could be integrated into MakieLayout now? It seems like nearly all of the widgets are, or will soon be supported there.

[SimonDanisch]

Yeah we should integrate them in the final version with the MakieLayout widgets!
But I may want to quickly do a first integration with Makie before that, to already figure out the problems with that ;) Makie widgets are much easier, so I assume it will be quite simple to this...

[asinghvi17]

Sorry for the noise, was trying to see if I could put this in multiple milestones at once, but it doesn't seem like it.

[KronosTheLate]

Shouldn't the "Log axis" box get a well-earned tickmark? ^_^

[Masacroso]

* [ ]  colorsampler interface + alpha keyword

* [ ]  resizing plots with multiple array attributes consistently (e.g. `scatter(rand(5), color=rand(5))`)

* [ ]  data space + marker space

* [ ]  text + scatter shader rewrite

* [ ]  WidgetsBase.jl integration

* [ ]  Improve how attributes work [document & implement refactor goals AbstractPlotting.jl#360](https://github.com/JuliaPlots/AbstractPlotting.jl/pull/360)

* [ ]  Merge MakieLayout

* [x]  Data transforms (log axis etc)

* [ ]  Merge most code shared in WGLMakie & GLMakie

* [ ]  clean up recipes

You forget here the more important point of all for any package: the documentation. As today the documentation is TERRIBLE, I wasted many hours trying to figure how to color a surface point to point... I just deduced a solution reading a post in the forum of Julia. After, I was trying to figure out what kind of types are valid for the function surface, I just find (by trial and error) just one valid type.

Seriously, its a pain to try to do something with this library due to the lack of documentation.

[kapple19]

Seriously, its a pain to try to do something with this library due to the lack of documentation.

Hi @Masacroso. I do understand your pain and frustration, I've had the same issue multiple times in different programming contexts.

Don't be afraid to search if someone's had the same struggle as you in the Julia Slack channel or on Julia's Discourse, otherwise you can ask your new question.

But @jkrumbiegel (edited wrong name, whoops) and the main developers of Makie already have a lot on their list of things to do, and they themselves have lamented on how behind the documentation is.

Seeing you're making this complaint, you must have found some value in Makie to use it. I like to tell people that instead of complaining that something they like is missing something, if it's open source then you can contribute to make it the product you're wanting it to be (up to some degree of democracy of course).

Your frustration is definitely more damaging than helpful in a context like this Makie Roadmap Discussion. It would help both you and everyone if you posted it in e.g. a gripe channel in Slack instead, designed as a lightning rod and frustration diffuser exactly for the frustration you're feeling. But as has been said, don't drop the negative energy in this inappropriate context.

I thought I might give this attention to your post here because we've all been frustrated before by what you're upset about, and maybe if we can convince you to join sides, both you and us will have gained a comrade 😁. None of us disagree with the point you're making and the frustration you're feeling that the Makie documentation is missing some things.

[jkrumbiegel]

Hey, leave such negativity somewhere else please. We don't need our hard volunteer work disparaged like that. If you find the documentation lacking, you can always help adding to it.

[KronosTheLate]

To add to @jkrumbiegel's response, I actually think it looks like documenting something as enormous as a plotting library is extremely difficult, and that the job done is really impressive. I generally find it sufficient to find out how to do everything I want to.

[asinghvi17]

AFAIK the third point (data + marker space) has been implemented in the :pixel, :clip, :data spaces as of 0.17?

[Moelf]

I'd like to help with cleaning up recipes and add stephist, any words on what's the guiding principle and desired outcome of clean up? maybe it takes a veteran to do it since it's quite nuance

[andreasvarga]

I would propose to add the following (if possible with high priority):

Allow to bring up a figure with CairoMakie from the REPL alone like Plots.jl can.

[SimonDanisch]

That should be working as of Makie@0.18

[SimonDanisch]

Ah, but you need to call Makie.inline!(false) though..

[andreasvarga]

Sorry, but I was not able to plot from REPL.

julia> using CairoMakie

julia> CairoMakie.scatter(rand(10))
FigureAxisPlot()

julia> display(CairoMakie.scatter(rand(10)))
FigureAxisPlot()

julia> CairoMakie.inline!(false)
ERROR: UndefVarError: inline! not defined
Stacktrace:
 [1] getproperty(x::Module, f::Symbol)
   @ Base .\Base.jl:31
 [2] top-level scope
   @ REPL[10]:1

julia> CairoMakie.inline(false)
ERROR: UndefVarError: inline not defined
Stacktrace:
 [1] getproperty(x::Module, f::Symbol)
   @ Base .\Base.jl:31
 [2] top-level scope
   @ REPL[11]:1

(@v1.8) pkg> status CairoMakie
Status `C:\Users\Andreas\.julia\environments\v1.8\Project.toml`
  [13f3f980] CairoMakie v0.9.3

[SimonDanisch]

Makie.inline!...somehow the re-export doesn't work, would be nice to fix 🤷‍♂️