[DISCUSS] Algorithm DSL

[Stephen Mallette <sp...@gmail.com>]

Right now we have pageRank() and peerPressure() just sorta hanging out in
our core traversal API, but they really don't fit well there. Plus, as we
add more of those types of steps we risk polluting the core API further. I
think that these steps were originally placed there as a bit of an
experiment (and out of convenience) and the intention was to eventually
move them somewhere more permanent.

I think we should pull these methods and other ones that follow into a
special DSL designed to house these things. Not exactly sure what that will
look like at the moment, but I feel like we should build something more
concrete than what is produced by the more user oriented annotation-based
DSL system. I think we would look to try to deprecate pageRank() and
peerPressure() for 3.4.0 and remove them completely at some future release
where breaking changes were allowed.

Not sure what kind of hell awaits in the GLVs if we do this....probably
won't be pretty.

Any thoughts or concerns?

Re: [DISCUSS] Algorithm DSL

[Stephen Mallette <sp...@gmail.com>]

Create an issue in JIRA for this:

https://issues.apache.org/jira/browse/TINKERPOP-1991

On Fri, Jun 1, 2018 at 4:35 PM Stephen Mallette <sp...@gmail.com>
wrote:

> Right now we have pageRank() and peerPressure() just sorta hanging out in
> our core traversal API, but they really don't fit well there. Plus, as we
> add more of those types of steps we risk polluting the core API further. I
> think that these steps were originally placed there as a bit of an
> experiment (and out of convenience) and the intention was to eventually
> move them somewhere more permanent.
>
> I think we should pull these methods and other ones that follow into a
> special DSL designed to house these things. Not exactly sure what that will
> look like at the moment, but I feel like we should build something more
> concrete than what is produced by the more user oriented annotation-based
> DSL system. I think we would look to try to deprecate pageRank() and
> peerPressure() for 3.4.0 and remove them completely at some future release
> where breaking changes were allowed.
>
> Not sure what kind of hell awaits in the GLVs if we do this....probably
> won't be pretty.
>
> Any thoughts or concerns?
>