While mojo has support for Jupyter Notebooks and VSCode, I’m a bit more comfortable with a fully-featured IDE. Positron is one such IDE for data science (don’t know if it matters, but I used to work there…currently no affiliation).
Is anyone (besides me) interested in using an IDE like that? I’ve been messing around on my local computer and have a working version, but there are a handful of upstream things that need to be solved to make it a great experience. I opened an issue for one of them, but would love to have a conversation here if something like this is interesting to the community
I’m not really familiar with Positron, but I’m the developer of Mojo Pro (https://plugins.jetbrains.com/plugin/31943-mojo-pro), which brings support for Mojo 1.0/1.1 to JetBrains IDE’s. It supports things like syntax highlighting, code completion, refactoring, debugging, etc.
Would that suit your needs, or is there a special usecase you’re trying to address?
Yeah i would say Positron is mostly for statistics and data analytics and for languages like the R Programming Language. But we’ve had folks like @forfudan implement Mojo for Statisticians and Pythonistas before so he might have an edge on this-pedagogically speaking. And other folks like @kose-y implemented MojoR.
Positron already supports a VS Code Extension API meaning you can run VS Code extensions directly on it because it natively supports extensions from the Open VSX Registry and standard `.vsix`files designed for VS Code..
Sure…my question isn’t one of “Is it possible?”, but rather “Does anyone care?”
I have a working prototype on my computer, so if there was wild enthusiasm I’d clean it up and release it. But since there isn’t (yet?), it doesn’t seem like something to take on right now. Even if the cost of creating code is near zero these days, maintaining a community project isn’t…
Clarifying this… It’s note that we don’t have enthusiasts yet in our community yet it’s basically because not that no one is interested-I’m also quite interested in exploiting new stuff like this. I’ll try installing Positron in my system
I just installed Positron on my Linux Machine. The “does anyone care” question you asked you can now count me in. Do you have any Open Implementations for this on Data Analytics for Mojo ? On Positron?
No, I don’t have a publicly available implementation. My local one modifies the Mojo REPL to make the Positron variable explorer work, as well as some changes to the Mojo VSCode plugin. So it would require coordination between both projects.
Yes, Python is my main language for daily work (VS Code and Databricks). I don’t use R. Happy to try Positron at some point, but to me Mojo isn’t the best fit for REPL-style work. Python is better at that. I’d rather use Mojo for the backend.
If i tell you using Positron with Mojo is the most interesting thing I’ve done lately. The REPL-style sheeting gives me a much better Machine Learning workspace yet it works faster and way more efficient than running my ML models in some IDE like VS Code or Cursor. Would you stick to this? And i would say Mojo is pretty snappy in positron its much better when i do data analysis or basic big data analytics. So claiming Python is much better than Mojo in this isn’t what someone like me would say in a pedagogical analogy. You can explain further why you claim so?
While Modular deprecated standard REPL features in certain nightly toolchains to prioritize core GPU/compiler pipelines, the underlying execution architecture is designed specifically to support dynamic, line-by-line JIT execution.
I use positron semi-regularly. it is quite nice for visualization and mixed R and python work. I didn’t really have the thought to try mojo in it, but if you are willing to share the mojo extension, I can give it a try.
Glad it works well for you. I will also give it a try.
About my point. I value workflow and reliability more than raw speed. For interactive work, I prefer dynamic typing, no compile step, and a mature ecosystem, which is where Python shines. Python packages have been tested for years on edge cases, while Mojo is still young and evolving quickly. Building a library in Mojo and debugging all the edge cases is not something that can easily be done in a REPL, so I prefer VS Code for a better sense of control.
By the way, the IDE itself doesn’t change execution speed. I’m not a fan of Mojo’s REPL mode. What I value more in Mojo is its memory safety, its syntax similarity to Python, and its nice interoperability with Python.
I’m curious to see where Mojo + Positron goes, though.
Thanks everyone, looks like there is some interest, so I’ll pursue getting the changes into the Mojo REPL that will make Positron work better with Mojo
? On Positron?
its much faster and better for AI workloads than VS Code;
I have only used spyder, jupyter notebook/lab, and VS Code.