Hi Ramesh, Apart from Till recommended, internal classes go with document around the source code define them, and for the "practices" part, maybe you can take a look at their tests.
But after all, these are all about Flink dev topics and so unstable to be used in production codes. Best, tison. Till Rohrmann <trohrm...@apache.org> 于2018年10月2日周二 下午9:51写道： > Hi Ramesh, > > the JobGraph, JobVertex, JobGraphGenerator, etc. are all internal classes > for which the community does not give any stability guarantees. Thus, they > can change from one version to another breaking all your programs. > Therefore, I would not recommend using these primitives if you are not > willing to resolve potential version upgrade problems. > > Apart from that, there is only little documentation for these components. > The best documentation is actually the code. > > Cheers, > Till > > On Tue, Oct 2, 2018 at 3:21 PM Ramesh Byndoor <in.rames...@gmail.com> > wrote: > > > Team, > > > > would like to create jobs from DAG's with details which is maintained > > externally(Filters, transform and Stream SQL). Idea is to create JobGraph > > and submit to Flink. I have started looking into JobGraph, Job, Vertex, > > JobGenerator from flink low level API's, Looks promising but is there any > > guidelines documented or any better practices for the same. Are these > used > > by any in the community.? > > >
