R-Fiddle | Hacker News Reader

R-Fiddle

r-fiddle.org

112 pointskmfrk31 comments

Very cool! We've been working on a similar open-source project for collaboration via github gists of R scripts. Think iPython web notebook meets js-fiddle over a github backend: https://github.com/att/rcloud

One nice feature of RCloud is it knows how to make a Javascript-R bridge, so you can create d3 and webgl plots with R data and back.

The github page has more details: http://att.github.io/rcloud

Can't help but feel that you missed out on a much better name: fiddleR. (or perhaps fiddle-r.)

Awesome tool!

Fiddler is actually a really, really popular tool for web debugging and hacking (http://fiddler2.com/), and while I didn't make this, I like that it maintains the {{ lang }}(-)?fiddle nomenclature.

It gives people an idea of what it does just by seeing the URL. :)

That chat panel is highly distracting/annoying, it'd would be nice if there was a way to hide it.

Other than that, this is indeed cool.

The chat panel uses Disqus, so is blocked by browser plugins like Ghostery.

I have Ghostery installed at work, but for some reason it didn't block it. Maybe I had it disabled or something...

However, here at home (where I also have Ghostery), it did block Disqus -- however, the very large panel is still visible. So, the annoyance of having a large part of the view taken up by useless space is still there.

Agreed, we'll make the discuss part hidden by default. The idea was that discussion of the R code should be easy for educational purposes.

Custom user CSS via Stylebot or Stylish is invaluable for blocking / removing such distractions.

I can't say much about the actual environment, as I'm not an R programmer, but I can say that I really dig the tutorial you guys set up, with the flow of dialog boxes pointing out different features. Well done there.

Unlike JS-Fiddle, the computation is done server-side instead of client-side, which could cause problems given that R's a memory and resource hog.

Will be interesting to see how this scales with the server. I also have found R to have a LARGE foot print.

Memory is indeed the our biggest concern. We now allocate 50MB of memory to every R session..

Likely facing significant load, but I am not able to get the R console to do anything (print values, plot, etc).

Cool idea. RStudio is nice but it lacks the collaboration/pair coding opportunities that this could provide.

Sourcing data and package functionality are essential if anyone is going to use this.

The R console should be up again, we just added some servers to R-fiddle :-)! By the way, we came up with the idea for R-fiddle while working on www.datamind.org and www.Rdocumentation.org.

very cool. I can imagine this will become particularly useful for SO questions in R. In fact, I wonder whether there's some interesting SO/fiddle hybrid for programming questions- being able to actually run the original question, the answers, etc....

A related free service (but not a competitor) is http://rpubs.com/, to which you can publish R scripts, R-markdown files and more using a single mouse click from RStudio.

If you want a fiddle-like environment for d3.js, try tributary: http://tributary.io.

Very cool, though I can sense clearly the stress the server is experiencing.

I typically use RStudio[1] to setup a R server on EC2 when I am on Chromebook and mobile. However having a fiddle for R definitely helps a lot when I need to do something quick, saving tons of time signning into AWS and boot up my rstudio instance.

[1]http://www.rstudio.com/

Does anyone have extensive experience with Pandas vs R?

I'm a long term R user, though I'm not exactly a pro and I honestly don't really like some aspects of it. I've looked a little at Pandas and it seems to do some things a lot better than R. My only grip is that the Python environment is a lot more of a PITA than R's CRAN system.

Does anyone have some insight?

Yes. I switched to Python/Pandas this Spring and haven't looked back. In my opinion, between Pandas, iPython, and Scitkit-Learn, Data Science has taken a large step toward Python as the defacto language.

If you find the Python environment a PITA, try anaconda's free distro: https://store.continuum.io/cshop/anaconda/ It includes everything you'll need for 95% of data science projects in python, all with correct versions, installed correctly, etc.

Really? Everyone I know uses MATLAB. I work with engineers, so the stats I do is minimal.

Thank you for the feedback by the way!

Cool, definitely something that has needed to exist for a long time now!

What's the story with security? It looks like you are using Rserve, are you also using AppArmor or lxc or something to prevent users from doing bad things? I didn't do anything malicious but did play around with system() and was surprised how much I was allowed to do.

We do use Rserve and AppArmor. Security is something we're working on right now, but we're hiring ;-p

I suggest RAppArmor over the standard setup of AppArmor, and black-listing a few functions (which is a rather nasty way, but usually works) that should never be called in such shared environment. I did something with my sandboxR package, see some details at http://hackme.rapporter.net @jcheng: not just "system" but e.g. a simple "readLines" call can also return funny stuff :)

Thank you for the advice and the pointer!

On http://www.rdocumentation.org these r-fiddles are used to embed the code parts for all the R help files. That way you can test and run these code snippets within your browser as well interactively.

Works for me: http://www.r-fiddle.org/#/fiddle/ODzY3oqE/1

Wow, perfect timing. Just learned about R for the first time today and I will definitely be using this.

Bravo! Your tutorial was cool, this is another cool tool for people wanting to learn R.

Thanks

this is how I learned JavaScript

Finally! Thank you.

Reply on news.ycombinator.com