Well - you like a challenge. :)

while i was writing up a response about 1684, i learned that somewhere
along the line that it was fixed! I don't see that behavior anymore on any
branch - here it is working on 3.2.10:

gremlin>
g.V(1).repeat(out()).emit().tree().by(union(__(),repeat(out()).emit().count()).fold())
==>[[v[1],5]:[[v[3],0]:[],[v[2],0]:[],[v[4],2]:[[v[5],0]:[],[v[3],0]:[]]]]

so - you fixed your first issue. i'm going to close that one.  You asked
for tips on how to proceed and I'd say that that's the first one....verify
that the bug is still a bug.

As for 1960 and 1643, they are going to be tricky I think. Luckily both
fail with TinkerGraphComputer so you can see the failures without spark in
the way. I'd start with 1643 as there is already a branch that has the
failing test in it. I'd say start by just being sure that you can get a
clean build on the tp33. then checkout the TINKERPOP-1643 branch in your
fork (i just rebased on tp33 and pushed). then, try to build and watch it
fail. From there, i'd say that since you are dealing with OLAP you need to
simplify as much as possible because you have parallelism going on all over
the place. One way to control that is to only use one worker in your test
runs which can be set here:

https://github.com/apache/tinkerpop/blob/d3b591c4db4e54111c296b4e3bc50d5247a77f53/tinkergraph-gremlin/src/test/java/org/apache/tinkerpop/gremlin/tinkergraph/process/TinkerGraphComputerProvider.java#L45

Then, in your debugger you run:

https://github.com/apache/tinkerpop/blob/d3b591c4db4e54111c296b4e3bc50d5247a77f53/tinkergraph-gremlin/src/test/java/org/apache/tinkerpop/gremlin/tinkergraph/process/TinkerGraphProcessComputerTest.java

you can limit to that one failing test by adding the `GREMLIN_TESTS`
environment variable in your IDE and setting it to the failing test case:

org.apache.tinkerpop.gremlin.process.traversal.step.branch.RepeatTest$Traversals

That should get you setup and going....after that it's just you and the
debugger.


On Wed, Nov 14, 2018 at 8:15 AM Atri Sharma <[email protected]> wrote:

> Thank you so much for the insightful pointers on how to start.
>
> I would love to look into the Neo4j upgrade process, but a personal
> objective is to improve Tinkerpop's OLAP capabilities. So I have very
> much gravitated towards the OLAP issues that you pointed to.
>
> To get warmed up a bit, I was thinking of looking at 1684 to get the
> grips of debugging in the core, and then look at 1960 and 1643. Does
> that sound like a reasonable strategy, or would it be too much to
> bite? Also, could you please help me with some pointers as to where I
> can start, which code module to look at etc? I am happy to do the
> specific conversation on the JIRA ticket as well if that makes
> documenting easier.
>
> Regards,
>
> Atri
> On Wed, Nov 14, 2018 at 6:13 PM Stephen Mallette <[email protected]>
> wrote:
> >
> > Thanks you for your interest. I would recommend that you start by having
> a
> > look at our dev docs on "Ways to Contribute"
> >
> >
> http://tinkerpop.apache.org/docs/current/dev/developer/#_ways_to_contribute
> >
> > and for information on how to setup your "Development Environment" for
> > TinkerPop.
> >
> >
> http://tinkerpop.apache.org/docs/current/dev/developer/#development-environment
> >
> > Not all of the latter will apply to you, but please have a look to get
> > familiar with things. We keep a small list of "starter" issues in JIRA
> that
> > you can look at:
> >
> >
> https://issues.apache.org/jira/issues/?jql=project%20%3D%2012316520%20AND%20resolution%20%3D%20Unresolved%20AND%20priority%20%3D%20Trivial%20ORDER%20BY%20priority%20DESC%2C%20key%20ASC
> >
> > in addition to those, there are a number of people who are interested in
> > seeing Neo4j upgraded to a more current version. You would probably be a
> > hero to a lot of folks if you decided to organize that effort. I'm not
> sure
> > that any of these item are in the area you said you were really
> interested
> > in (i.e. gremlin-core) but they should all be enough to get you familiar
> > with the general environment. If none of those items are interesting then
> > I'd say that you might consider rolling through the other open JIRAs that
> > aren't started and see if there's any you think you could tackle. Want a
> > real challenge - fix some OLAP bugs:
> >
> > https://issues.apache.org/jira/browse/TINKERPOP-1960
> > https://issues.apache.org/jira/browse/TINKERPOP-1643
> >
> > or get your debugger skills going on these problems in gremlin-core:
> >
> > https://issues.apache.org/jira/browse/TINKERPOP-1684
> > https://issues.apache.org/jira/browse/TINKERPOP-1992
> >
> > Whatever you choose to try, it would be helpful to know what you're
> making
> > an attempt at. Please let us know and we can try to be helpful along the
> > way. Thanks again for offering to get involved.
> >
> >
> > On Wed, Nov 14, 2018 at 4:09 AM Atri Sharma <[email protected]> wrote:
> >
> > > Hi All,
> > >
> > > I am looking to get my hands dirty in the ways of Tinkerpop and get
> > > insights into the internals and how it actually works.
> > >
> > > Could someone point me to some JIRAs that I can hack on? I am looking
> > > specifically at the innards of how a query is processed e2e, so please
> > > advice.
> > >
> > > Atri
> > >
>
> --
> Regards,
>
> Atri
> Apache Concerted
>

Reply via email to