[Discussion] A new Bokeh binding?

Recently I’ve been looking for a interactive visualization solution (again), and realized that one of Bokeh’s core dev @Bryan is actually here on the discourse and at one point was looking for people to re-make a Bokeh binding(Invitation to VizCon 2 - 12-16 March 2020).

We have had a Bokeh.jl a long time ago, unfortunately it was discontinued.

Not sure how much interests people have in this, maybe most of us don’t need interactive visuals often and when do, Plotly is good enough for most cases. It may be find to get the basic going and worry about Plots.jl integration (as a backend) etc later. What do people think?

5 Likes

Not an expert on plotting libraries and their current capabilities, but a while ago a colleague of mine wrote a Bokeh application for viewing clustering results. It had the capability of generating matrix plots and one could select stuff on a plot and the other plots would be automatically highlighted. It also had some minimal data filtering options (don’t know if this is related to Bokeh). It was mindblowing and time saving so congratulations to the Bokeh devs. (it was written using Python)

I’d love to see that kind of capability in Julia as well. Not sure if other libraries cover such functionality, but diversity never hurts. I’d like to help, but I would need some guidance since I’m a total noob in this area.

2 Likes

You can find that level of interactivity with Stipple: https://github.com/GenieFramework/Stipple.jl
For simpler/quicker things, I love Pluto + Plots with PlotlyJS as a backend: https://github.com/fonsp/Pluto.jl

2 Likes

VegaLite.jl also has pretty extensive support for interactive figures, see here.

2 Likes

ah, right, I also enjoyed using VegaLite, I guess my personal weakness when it comes to using VegaLite is that I often need to ad-hoc some hack into the plot, for example, let’ say a title inside the plot using annotate!(), the way VegaLite works is that it’s very rigid, I have to prepare exactly the format and style of the plot etc. But I guess I can give it another try!

1 Like

Actually, I like bokeh quite a lot (in particular more than plotly) and I used it several times for visualization with python. It is an interesting tool, and even more in combination with holoviews, geoviews or similar frameworks. The server app are nice (you can check examples here)

I checked the bokeh implementation in julia sometime ago, but it had been programmed for Julia v0.6 (as far as I remember) so it has to be actualized to Julia v1 or later. I don’t think that this would be a huge amount of work but still some work is needed to make it functional again.

On my side, I would be a potential user of bokeh.jl for sure.