Hey Kyle,

It probably belongs here -
https://github.com/apache/incubator-metron/blob/master/metron-platform/metron-common/src/main/java/org/apache/metron/common/dsl/functions/StringFunctions.java
There is an existing JOIN function for strings - might this suit your
needs? I didn't see a unit test for it, so it would probably be good for us
to backfill with a test here as well. I can submit a PR for it, or if
you're already in that code, you're welcome to also -
https://github.com/apache/incubator-metron/blob/master/metron-platform/metron-common/src/test/java/org/apache/metron/common/dsl/functions/StringFunctionsTest.java

e.g.
Object joined = run("JOIN(['A','B','C','D'], ':')",new HashedMap());
System.out.println(joined);
Object joined2 = run("JOIN(['A','B','C','D'], '')",new HashedMap());
System.out.println(joined2);

Output I get is:
A:B:C:D
ABCD

AFA where to put these functions, I believe we have a number of options now
that we have the ability to add to the storm topology classpath and
sideload jars.
- https://github.com/apache/incubator-metron/pull/204
- https://github.com/apache/incubator-metron/pull/468

I think that if the functions are unique to a customer, they should
probably be built as a stand-alone Maven project. I believe Otto is working
on this ATM if I'm not mistaken. If there is universal (across all
functions in the system, whether parsing, analytics or otherwise) then they
should probably go in with the dsl package in metron-common. At some point
we might want to make Stellar its own module, but there is some work there.

Best,
Mike


On Sun, Apr 9, 2017 at 2:26 PM, Kyle Richardson <kylerichards...@gmail.com>
wrote:

> I have the need for a new Stellar function to perform string concatenation.
> I have it implemented but am curious about where new functions should live
> given the new capabilities around 3rd party Stellar function libraries.
>
> So, I guess my question is, should this function live in:
> 1) metron-common with the other string functions
> 2) another metron project
> 3) as a standalone project and not part of the metron source tree
>
> While I'm specifically asking about this case, I think it's also worthwhile
> that we think about where other new functions should live in the long term.
>
> Thanks!
>
> -Kyle
>

Reply via email to