Joins - Only plan upto 3 implemented in RxJava but Rx has 16

[sirinath]

Is it possible to have larger plans for joins.

(Also besace it is not is the core concerned about how optimised the implementation is.)

[akarnokd]

That 3 was tedious to implement and there was no one at the time who needed it or understood its use case.

[sirinath]

If you try to zip together about 16 streams in different combinations like DAG you will see how useful this is. Ideally there should a lot more than 16 as you will be dealing with a large number of streams to coordinate between in some cases.

[akarnokd]

One problem is that the operator requires a function callback with that arity, but we have only up to Func9. I didn't understand the operator enough so I could try with FuncN: there might be no good way to generalize to any arity with FuncN. Adding the remaining 4-9 arity version is mostly mechanical; one could always use Rx.NET as cheat-sheet, but writing tests is cumbersome.

[headinthebox]

I'b be really surprised any human can patterns over more than 3 streams by hand because of the combinatorial explosion. Are you generating the code?

[akarnokd]

I certainly didn't. I saw Rx.NET use code generators on this. I'm not sure if we have anything similar available with Gradle and/or annotation processing.

[sirinath]

Then stop at 9 and N. When you have more function arity then add accordingly.

[benjchristensen]

Up to 9 was added ... closing out.