Comment by uptownfunk | Hacker News Reader

Comment by uptownfunk | Hacker News Reader

R was that language I learnt so many years ago. I love it actually, have so many fond memories. It’s hard to find something better especially for prototyping. I will admit productionalizing has almost drag and drop simplicity in python compared to R.

I love love love what Hadley has done with Dplyr for the most part, at least in spirit, though I think the implementation could have been done better so as not to be so clunky, esp wrt NSE. But I think he is just trying to work within the current R ecosystem.

Which makes me ask.. is it then time for R2? (Like a Python 3). Before you shoot me.. Do we need to save the good things we have innovated from within the R ecosystem over the years and consider doing things from scratch?

Is this what Julia tried to do? I haven’t gotten around to trying it yet.

That said, I think R is always going to be there and have it’s place.

Frankly if they could just make an IDE like Rstudio that ran python, I’d probably be happy enough with that. I heard with reticulate you can run both, curious to hear of others experience with this..

`reticulate` is a brilliant package for running Python within R. I guess it's good as long as it's local machine. I've not tried productionizing a code that has both R+Py. But it can work within Shiny too, helping bring Python's Datascience stack (esp. scikit-learn) within R.

You should try Jupyter lab. It has notebooks, text editors and more.

The whole problem with notebooks is that it's not a simple text file. On the other hand has anyone tried Rodeo for Python?

I've used it. It does have the feel of RStudio IDE if that's what you're getting at.

I wanted to like it but it kept crashing when I tried it a year or two back, I may give it another shot. I do recall it was closest to what i wanted..

You can save notebooks to simple text files. You can write simple textfiles in jupyter lab and have a split view where you can execute it. You can have a notebook, console and graph view with the same kernel on the same page as split windows.

You can make jupyter lab behave mostly like rstudio, but it can do a lot more, especially in terms of visualization and rich display of data objects.

jupytext --sync

If you need a multi-language notebook that plays nicely with version control then RMarkdown in RStudio is worth checking out

I like Jupyterlab, but there's a huge gap between it and something like RStudio.

Reply on news.ycombinator.com