search

IDEMSInternational / R-Instat

A statistics software package powered by R
http://r-instat.org/
GNU General Public License v3.0
38 stars 102 forks source link

Freedom from platform dependency and rdotnet! #3568

Open volloholic opened 7 years ago

volloholic commented 7 years ago

The current implementation is a .NET product and as such is tied into the associated platform dependencies. We are also subject to the limitations of the rdotnet package which is currently facilitating our link to R.

This is acceptable as the vast majority of our initial target audiences are Windows users and the initial development would not have been possible without these compromises under our chosen model.

However, if the ideas developed in R-Instat prove useful to a wider international audience than initially intended, then platform independence and improved stability of a connection to R will become higher priorities. Our initial thoughts in this direct would be to rebuild the front-end structures and dialogs from an RStudio base.

chrismclarke commented 5 years ago

If you're considering this then the main trade-off I see is how close to an excel-like grid system you require vs freedom to use popular tools/frameworks. If you want to get as close to excel as possible you'd probably want to be using c++, and could possibly start with a fork from the libreoffice calc tool (assuming open licence allows). If a reasonable grid will suffice then I would recommend building in html5 technologies with Electron as a backend (identical to how Jamovi made their app)... in fact that could also be a start for your link to R).

isedwards commented 5 years ago

Part of the motivation for building on RStudio was to engage with their community and to see how much support there may be for an "R-Instat-Studio" solution that is suitable for a wider user base than RStudio. The RStudio code looks clean an extensible and I can see how value could be added there.

LibreOffice calc is an interesting idea, especially if the goal is world domination (bringing stats to the masses). I've not looked at the code, but I suspect it would be a more difficult path. Although the grids in stats packages are spreadsheet like, their inflexibilities (e.g. not accepting different data types in a given column) are also their main strength. It may be difficult to make the R-Instat routines work as well with the adhoc sheets that most spreadsheet users create.

A Jamovi-style route would be the most exciting, but could R-Instat add much value duplicating what Jamovi are already doing? If this route was chosen, would it be better to cease R-Instat development and contribute to Jamovi instead?