Hacker News Reader

Comment by jarpineh | Hacker News Reader

Thank you for comprehensive comment. Hopefully I can understand enough about the video to get the gist. Idea of Scheme style functional features with APL's power tools. If you get around writing more about these things I'll definitely be reading.

Since most of Python and JavaScript I do at work is mostly searching and using examples and libraries from where ever having full complement already with the system feels impossible. On the other hand, APL implementations have had some time to make things right. I'll take a look at that big APL book and see what I can gather (my favorite way to learn).

BTW: I got my first taste of APL from one of the FinnAPL fellows. I saw some incredible things by him I've since tried to replicate after a fashion.

View on HN

The Mastering Dyalog APL book is available as a free PDF, too. Though the paper version is more enjoyable, and serves as an excellent monitor stand in need, as well.

As a Scheme programmer, one of the first things I did when learning APL was to create a new REPL/environment in Chez Scheme called Sapling that replicated APL in Scheme using Scheme implementations of the primitives.

I quickly discovered that at least for my meager mind, there was no way to combine the two as one. I could use one within the other without trouble, but I couldn't mix the two easily.

If you're interested, the Co-dfns compiler is designed to integrate with existing languages. If your language has a C FFI, then you'll be able to integrate with Co-dfns compiled code. The workflow is basically for you to produce a namespace of the functions you want to use, and then you can compile them and link against that in your Python or C or C++ code. You can then just call those functions with the appropriate arrays and things are a go. The main then you need to do is write a function to create a Co-dfns array to and from the data types you are using in your program.

This got me a bit confused. Is the purpose of Co-dnfs to compile apps that run on a GPU? Or does it only itself compile on the GPU for apps than can run on any target architecture?

Other than that the idea of getting using my domain specific Python stuff (and few general libraries) with data processing tools is intriguing.

These two goals are not mutually exclusive. The Co-dfns compiler is a compiler that is designed to self-host on the GPU, but it is a compiler for the Co-dfns language, which is a lexically scoped syntax in Dyalog APL. The compiler compiles dfns programs and supports Mac, Windows, and Linux, targeting CUDA, OpenCL, and Intel CPUs. The compiler itself is written as a dfns program, hence the idea of self-hosting on the GPU.

Target applications for the compiler include document processors, cryptographic libraries, neural networks programming, data analytics, high-performance web applications, vector databases, financial algorithms, bioinformatics, and HPC/Scientific computing tasks.

So, in short, yes, it compiles apps to run on the GPU, or the CPU, whichever you prefer. And yes, it is meant to compile itself on the GPU (so, you compile a GPU program on the GPU).

Reply on news.ycombinator.com