Guide for SciRuby compatibility?

[Victor Shepelev]

Hi.

Is there any (maybe concise and incomplete?) guide "what should the gem can/implement for easy integration into SciRuby infrastructure?"

For example, I've made a (really simple and naive, yet somewhat useful) "map visualisation" gem: https://github.com/zverok/worldize

It seems to have some use for scientific context. What should it have to play well? Intuitively, there's at least:

* accept DaRu::DataFrame as data source (or even extend it? I've though about something like dataframe.worldize('Country codes', 'Population as of 2016') -- parameters are columns with countries and stats to plot)

* integrate into IRuby (seems that additional work is not necessary here: it produces Rmagick's Image, which IRuby fortunately able to show)

* sample notebooks with some useful examples.

What else?

And what about another kinds of libraries, like data-processing/data extraction ones?

V.

[Sameer Deshmukh]

One very important factor when trying to interface libraries is that they should be able to pass data to and from each other without much duplication or speed penalty. So I think if you make your library compatible with daru and nmatrix, it should suffice to say that any data you passed into or from your library will be usable by a variety of other gems like the statsample family (all of them use daru internally), nyaplot, gnuplotrb, iruby and mixed_models.

All these gems are compatible with daru, and if your library makes use of data storage and processing, I'd suggest you use daru/nmatrix as your data store depending on your use case.

Of course, this is just one consideration.

[Victor Shepelev]

Thanks!

--
You received this message because you are subscribed to the Google Groups "SciRuby Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sciruby-dev...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.