Yes that's an important point, and one that may be Plots most important use. For me, I want to write complex plotting/visualizations for OnlineStats and OnlineAI, and its too restrictive to assume that a user will have Gadfly, or Qwt, or Winston installed. (Right now, Plots is only dependent on Colors and Reexport... Everything else is imported dynamically as needed... Even DataFrames) So I either have to force huge dependecies on them or accept that some people can't ever use my visualizations. Now with Plots they only need any backend that supports the plot type, and the user can choose what backend they like, so package authors may be more inclined to expose plotting recipes for their problem domain.
On Thursday, September 24, 2015, Kristoffer Carlsson <kcarlsso...@gmail.com> wrote: > It's a really nice package when you write your own module that does some > plotting but you don't want to enforce a certain backend on the user. > > > On Thursday, September 24, 2015 at 4:04:50 AM UTC+2, Tom Breloff wrote: >> >> I'm happy to announce a new tagged release of >> https://github.com/tbreloff/Plots.jl, a plotting package that is simple >> and easy to use, and which I hope will one day offer the union (not an >> intersection) of functionality between underlying backends. This is still >> a work in progress, but it's possible to create many varied plots already. >> Check out the recently expanded readme and linked examples to see more. I >> currently support 6 different plotting backends to varying degrees, all of >> which use exactly the same plotting commands: >> >> >> - Gadfly >> - Immerse >> - PyPlot >> - Winston >> - UnicodePlots >> - Qwt >> >> Immerse is the new package from Tim Holy which allows for viewing Gadfly >> plots in a Gtk gui window, with some cool functionality to zoom/pan and >> select points for numerical analysis... it is my recommendation for the >> best backend, and it is (along with Gadfly) the best supported. I also >> appreciate UnicodePlots (REPL only) for the minimal dependencies... for >> quick plots it's pretty cool. >> >> Some of the big improvements coming down the pipeline are support for >> partitioning incoming data, various datapoint-specific settings (such as >> varying size/color of individual points), and rounding out the standard >> library of graph manipulation (specialized axes, scales, etc). If Plots >> covers most of your needs right now except for a few tweaks, you can always >> access the underlying plot objects for the backend of your choice and >> update directly. (however please ask questions and request features so >> that I know what to prioritize, and of course report any bugs that you >> find). >> >> In the longer term I plan to include some recipes and example IJulia >> notebooks for creating more complicated plots, as well as other types of >> visualizations (graphs, 3D) and interactivity. >> >> Feedback and wishlists are appreciated! Happy plotting. >> >> Tom >> >
