RStudio Is Becoming Posit | Hacker News Reader

RStudio Is Becoming Posit

rstudio.com

119 pointstosh35 comments

At first I thought they were renaming the IDE, which seemed like a clearly bad idea. Makes sense to rename the company I guess.

It tickles my brain every time I see it when I am reminded that the same guy who made ColdFusion and a diet app just decided to build a small company based upon a weird stats language 13 years ago. Good for him.

This. I thought they were renaming the software, which seemed like a really bad move.

Those all sound like good names. From the article:

> To avoid this problem and codify our mission into our company charter, we re-incorporated as a Public Benefit Corporation in 2019.

> Our charter defines our mission as the creation of free and open source software for data science, scientific research, and technical communication. This mission intentionally goes beyond “R for Data Science” — we hope to take the approach that’s succeeded with R and apply it more broadly. We want to build a company that is around in 100 years time that continues to have a positive impact on science and technical communication. We’ve only just started along this road: we’re experimenting with tools for Python and our new Quarto project aims to impact scientific communication far beyond data science.

> [...] What does the new name mean for our commercial software? In many ways, nothing: our commercial products have supported Python for over 2 years. But we will rename them to Posit Connect, Posit Workbench, and Posit Package Manager so it’s easier for folks to understand that we support more than just R. What about our open source software? Similarly, not much is changing: our open source software is and will continue to be predominantly for R. That said, over the past few years we’ve already been investing in other languages like reticulate (calling Python from R), Python features for the IDE, and support for Python and Julia within Quarto. You can expect to see more multilanguage experiments in the future.

Apache Arrow may be the best solution for data interchange in "polyglot notebooks" with multiple programming languages where SIMDJSON-LD isn't fast enough to share references to structs (with data type URIs and quantity and unit URIs) with IPC and ref counting. https://github.com/jupyterlab/jupyterlab/issues/2815

They _are_ renaming the (commercial) IDE while keeping the name of the (open source) IDE. Using two completely different brands doesn’t seem a good idea either.

> At first I thought they were renaming the IDE, which seemed like a clearly bad idea.

Which highlights exactly why it’s an amazing idea to rename the company.

ColdFusion was such a horrendous product and I was kind of put off of RStudio for a while when I found out Allaire started it.

ColdFusion was something that seemed cool and was kind of clever but it had so many runtime bugs and was way worse than PHP, ASP, and JSP.

I used it at a college internship because it was the only dynamic language we were allowed to use to access an old Oracle database. At the time (almost 20 years ago), it seemed kind of luxurious compared to PHP.

I just looked it up and realized it's still around and still expensive.

He is my college (Macalester) alum. His brother also built Circle (USDC).

Fun fact: both of them studied political science

It’s amazing how much value this company has built on top of R. In my opinion they’ve done more for R than pandas and numpy has for python.

You can't really compare rstudio to r and pandas+numpy to python.

Maybe jupyter, but even then those are completely different goals to different ecosystems.

When I say rstudio I guess I am implicitly saying also devtools, tidyverse, and ggplot2 because Hadley Wickham works at Rstudio.

I guess it is indeed true that Rstudio is much bigger than the IDE.

I think the most useful comparison is between languages have an org that makes a little money and uses it to hire top notch developers to work 100% of the time on OSS [1], VS languages that don't have that.

RStudio's contributions to the R ecosystem have been magical, most notably the RStudio IDE and the tidyverse set of libraries.

When I need to get anything done quickly that involves working with data, I reach for R immediately, long before even considering python, ruby, SQL or any of the more constrained tools like tableau or (uggh) SAS/SPSS/et al.

R is just so fast to develop in, and so reliable. I feel like I hit more walls in other languages ecosystems; not so much the actual syntaxes, but more oddities/bugs/incompatibilities in the way/place/environment in which I use the code.

[1] RStudio (the org) has 40 (!!) devs full time on OSS! (and many more who aren't full time)

This feels weird at first glance, but the IDE is staying as RStudio. I love the company and all they’ve done to make data analysis more accessible, so good on them and may they succeed even more in the future.

Still not sure what Quarto is above and beyond a normal RStudio notebook, though.

Quatro is a tool for converting notebooks into shareable formats (like PDF, HTML, Docx). Notebooks are very hard to share. You can't send it to your stakeholders, because Python code will scare them. You can use Quatro to convert notebooks into other formats and easily hide the code. It is like nbconvert on steroids.

I'm working on solution for notebooks sharing with non-technical users. Similar to Quatro it has YAML header, but it can also add widgets to the notebook and serve notebook as interactive web app. My framework is called https://github.com/mljar/mercury

I've been using a library to knit Rmarkdown files to OneNote API compatible HTML (https://github.com/CopenhagenCenterForGlycomics/knoter), which lets me output PDF figures (that I can download to go into Illustrator for the publication process), Excel files (which are great for validation and supplemental tables especially with wet lab scientists), and ad-hoc annotation on the document in one package (highlight text, scribble notes etc).

It would be awesome if there was a little bit more support for adding collapsible sections, and code blocks into OneNote pages so that I can look at everything in one spot.

You can use RISE (https://rise.readthedocs.io/en/stable/) for Jupyter notebooks to use it for presentations and also export to PDF.

FWIW repo2docker installs everything listed in /install.R or /.binder/install.R. I'll just cc this here because of the integration potential:

> repo2docker fetches a repository (from GitHub, GitLab, Zenodo, Figshare, Dataverse installations, a Git repository or a local directory) and builds a container image in which the code can be executed. The image build process is based on the [#REES] configuration files found in the repository.

Docs: https://repo2docker.readthedocs.io/en/latest/

  python3 -m pip freeze | tee requirements.txt

  conda env export -f environment.yml
  conda env export --from-history -f environment.yml

  python3 -m pip install jupyter-repo2docker

  repo2docker .

  # git branch -b mynewbranch; git commit; git push 

  repo2docker https://github.com/binder-examples/conda
  repo2docker https://github.com/binder-examples/requirements
  repo2docker https://github.com/binder-examples/voila #dashboards #ContainDS

jupyter-book/.binder/requirements.txt: https://github.com/executablebooks/jupyter-book/blob/master/...

  python -m webbrowser https://mybinder.org/

"#REES #ReproducibleExecutionEnvironmentSpecification" config files that repo2docker will build a container from at least one of: requirements.txt # Pip environment.yml # Conda (mamba) Pipfile install.R postBuild # run during build start # run in ENTRYPOINT runtime.txt Dockerfile

https://repo2docker.readthedocs.io/en/latest/config_files.ht...

repo2jupyterlite (WASM) sure would be useful for presentations, too.

Good to see the strategic decision to re-incorporate as a B-corp. Growth and return to investor is important, but shouldn't be the ultimate north star.

Do you know other B-corps in data science space?

Quansight is a B-corp on the Python side.

data.world is also a B Corp [0], although having worked for a different B Corp down the street from them, it doesn't mean much.

[0] - https://www.bcorporation.net/en-us/find-a-b-corp/company/dat...

It's more the PBC than the B corp status that make Posit's approach different.

From talking with people who have used R but not as their primary language, "R" and "Rstudio" are basically synonymous. I anticipate having conversations in the future about how Posit is to RStudio as Meta is to Facebook.

I think it will be more like Racket and PLT Scheme. It wasn't that long ago that the named changed but I suspect most people on HN wouldn't know what I was talking about if I mentioned PLT-Scheme and DrScheme.

The rebranding is similar as well, just as PLT-Scheme really was doing something different than the rest of the scheme out there (as far as keeping up with modern coding practices and expectations etc) RStudio and the tidyverse team have fundamentally changed what it means to write R.

I wouldn't be surprised if in 5 years people talk about "doing stats in Posit" as short hand for "modern R tooling with current (tidyverse related) best practices".

It really doesn't matter, people will adapt over time. Jupyter used to be iPython Web Notebooks. The project wasn't harmed by a rename and refocus.

I feel like they could have just come out with basically identical Pystudio and Jstudio products and kept the substantial value of the old brand name.

That said Quarto looks exciting and I'm looking forward to whatever they make next.

I’ve researched current platforms for technical/scientific blogging, and decided to go with Quarto. It’s powerful and quite pleasant to use to a person with nearly zero web dev experience.

It's really hard to explain to folks that "RStudio" makes software for languages besides R! The old brand will stick around for the RStudio IDE itself.

It’s been super easy for me actually. People are like “ok, cool.”

With VSCode they support lots of languages so it’s pretty easy to explain how RStudio runs Python now.

Hmm... I wonder if it will be possible some day to use posits in Posit:

- https://posithub.org/

- https://en.wikipedia.org/wiki/Unum_(number_format)#Posit

I love RStudio/Posit and really appreciate everything they’ve done for R, RStudio, tidyverse, etc but I’ve worried if they’ll stay in business because all the best stuff is free.

I pay for R Server Pro and it’s kind of nice for scientific computing. But I pay way less than for other “data science” server frameworks.

I hope this helps them stay solvent.

The beginning of the end for R

I guess this also rules out name confusion. For me R-Stuio always was a data rescue software. o_O

The company name is changing, but the open source IDE name isn't; it will still be called RStudio.

RStudio is my IDE of choice for Python. Works well with any conda env.

As a long time R user though, R is not synonymous with RStudio, as one previous HN commenter claims in this thread.

RStudio is a great env, but command line R or R straight out of emacs is eccellent too.

Reply on news.ycombinator.com