Using Ray components as library

Ray Core
1

How severe does this issue affect your experience of using Ray?

  • Medium: It contributes to significant difficulty to complete my task, but I can work around it.

Whether the different components of Ray (such as Ray core) can be used independently by another AI platform? If so, any guidance? How can I build each of those components from source and re-use them?

2

For building from source see [here](https://docs.ray.io/en/latest/ray-contribute/development.html) in Ray Docs.

All of Ray can be used on top of any AI platform, including Anyscale’s own [here](https://www.anyscale.com/). Which components (ie: libraries) you want to use is up to you.

However note that the stack is all built on top of Core (ie: Core < Data/Train/Serve). It doesn’t mean you have to actually interact with the Core libraries, if you use on of the higher stack libs it’ll basically interact with Core concepts for you.

That said, I do recommend you to learn Core or at least go through the documentation as the concepts in there are very important for you to efficiently troubleshoot/accomplish what you’re trying to do at a higher level.

3

Hi Sam,

Thanks for the pointers.

Do you want to provide with some specifics if I want to use Ray’s remote, task, actor & object artifacts in C++ with minimum wrappers?

Best regards,

Manoj

4

:thinking: it’s possible… but not something we recommend at the this point as the components are pretty interconnected so trying to build a tiny package of just cpp ray isn’t something we even do (and definitely don’t support).

is there a reason why you’re so interested in keeping it so lightweight?

5

Well, we want to integrate Ray core & cluster with our existing AI modules (which are in the line of Ray tune, data etc.) in such a way that entire software stack becomes manageable.

BTW, I have installed ray default and I was trying to build the following example in cpp:

But couldn’t do it so, are there any guidelines to build this?

Best regards,

Manoj