Somewhere in a backroom, there's a hamster named Julia. In a hamster ball.
And rolling around behind it is a smaller ball called M.
Don't forget the centipede crawling around in the sewer pipes named Fortran. We've all been trying to kill it for years and yet, somehow, it keeps going.
I honestly figured Fortran was still somewhere above M.
Jokes aside, I encounter Fortran in the applied physics community still fairly often. And have never encountered M in a professional context.
I loved Julia in my data science classes. Codes like python, runs like c. Can also use it with bash by piping values in
I love R, we are best friends. Life is wonderful when basically every function is vectorized by default.
The tidyverse is my favorite place
I used to love it, but all the non-standard evaluation started to give me a headache.
It’s easy enough to just not use it at all, except for ggplot which recently deprecated aes_() which fucking kills me; they really are dead-set on forcing tidy evaluation.
Probably barking up the wrong tree here, but boy do I hate R. The documentation is the worst, combined with the poor r studio experience. Vscode makes notebooks a bit better but lost a lot of functionality as far as I could tell.
Laughed so hard when I this course once they told us to do ML in R with Keras ... By calling the python API.
I've made such course and still feel sorry for the students. Was some legacy code just laying around.
Oh how I wish this was the data scientists I work with
This post was sponsored by the Matlab gang
Not me!
DON'T HURT THE SIMPLE CHILD
I like and despise R... WHY DO I HAVE TO COPY THE TEXT FROM MY CONSOLE INTO A SCRIPT TO ACTUALLY SAVE IT AS AN .R FILE????
Because it's a console, not an IDE. But I see your frustration and does seem ridiculous
Maan, I am way too much of a himbo chemical engineering student yo understand what IDE means, had to channel my inner parent and ask a software guy for help
I am currently arguing with par() and coming back here to shitpost when I get too angry.
EDIT: OMG https://stackoverflow.com/questions/56497422/using-jupyter-r-kernel-with-visual-studio-code finally an IDE with SENSE
You don't really have to. You could save the workspace along with the history of you commands to load it at a later time, and never have a script at all.
The reason nobody really does that (except maybe if they use R once in every decade) is that it's not really viable in the middle-term. That is because it doesn't distinguish between failed attempts and actual, final code and so quickly becomes a mess.
I feel like some people are using python just so they can say that they use python
It's more because when ds has to hand over their work to the eng team the eng team doesn't want to fuck with r
Trust me, the eng team doesn't want to fuck with Python either.
Meh lots of python stacks out there now, especially data pipes.
then typescript front ends.
Point is no one wants to support r in production
I considered looking at R once. As a data scientist and an experienced Python user, maybe I'll see something useful. Then I learned that R uses <- for variable assignment and = for equality evaluation, and I stopped learning because I would make that mistake if I learned to switch back and forth for the rest of my life.
You can also use = to assign variables in R it’s just not good practice
I remember trying to get shiny working for a statistician. Bad tooling. This was about 3 years ago.
I really wanted it to be severless but at the time it wasn't really possible.
I have since seen a cool web assembly method where it runs all the shiny stuff in the browser
Oooo do you have a link to that web assembly method?
A place for majestic STEMLORD peacocking, as well as memes about the realities of working in a lab.
Rules
This is a science community. We use the Dawkins definition of meme.
