jupyter integration
lazarusA opened this issue · 3 comments
any plans for integration within jupyter notebooks? So far we can code there and run it, but always a new window is open with the plots.
Not in the near future, since I don't use Jupyter at all and I would not know where to start from:
Which gnuplot terminal would be suitable?
Will such terminal be available in all platforms?
How to send plots to Jupyter?
How to handle multiple plot windows?
Which additional dependencies would be needed by Gnuplot.jl?
Would such additional dependencies bloat user experience when Jupyter is not needed (or desired)?
Probably there's room for a separate package here...
Otherwise, If a simple and reliable solution can be implemented I have no particular concerns to include it in Gnuplot.jl
Oh I see... Gnuplot.jl in the present stage works flawlessly, I really like it as it is.
Also, I see that different people contributed to make this possible in Gaston (https://github.com/mbaz/Gaston.jl/blob/master/src/gaston_config.jl).
I guess Gaston.jl and Gnuplot.jl approaches are quite different: the former defines its own API, and hides the gnuplot process behind it. In principle it could use any other backend, not just gnuplot.
On the other hand Gnuplot.jl aims to be as transparent as possible, and you can't change gnuplot with something else...
Also, with Gnuplot.jl, you only need to learn two new macros: gp and gsp. No further names or syntax are required, just the original gnuplot one. If you know gnuplot, you can exploit Gnuplot.jl with minimal effort.
Finally, Gnuplot.jl aims to implement the most concise syntax ever. When I explore my data I want to focus on them, without struggling to remember the correct keyword or syntax to produce a simple plot.
Admittedly, Gnuplot.jl syntax may seems unintuitive at first. But only if you never used gnuplot. And if you're not used to gnuplot I see few compelling reasons to use Gaston.jl or Gnuplot.jl: there already are severl plotting packages widely used in the Julia community...