Question 2:
to my knowledge (what I've found digging through the code). Each Operator
has a unique OperatorType that is used when writing the operator's stats in
the profile. So in this case, Each FormatPlugin implementation should have
a corresponding unique OperatorType added to CoreOperatorType.
I will get back to you after I get more information about the remaining
questions.
Thanks
On Thu, Sep 17, 2015 at 8:14 PM, Edmon Begoli <ebeg...@gmail.com> wrote:
> Hello all,
>
> Could some please with answering my questions 2-5 below?
>
> Thank you,
> Edmon
>
> On Wednesday, September 16, 2015, Edmon Begoli <ebeg...@gmail.com> wrote:
>
> > I am studying the JSON Storage plugin.
> >
> > Can someone please answer what is the meaning (or intent) of the
> following
> > methods:
> >
> > # Question 1:
> > # This is from inside the getter for RecordWriter. What is major/minor
> > fragment id?
> > String fragmentId = String.format("%d_%d", handle.getMajorFragmentId(),
> > handle.getMinorFragmentId());
> >
> >
> > # Question 2:
> > # Speficially, what is Operator Type, and what is JSON specific about it?
> > public int getReaderOperatorType() {
> > return CoreOperatorType.JSON_SUB_SCAN_VALUE;
> > }
> >
> > # Question 3:
> > # same, but for writer
> > @Override
> > public int getWriterOperatorType() {
> > throw new UnsupportedOperationException();
> > }
> >
> > # Question 4:
> > # I generally understand the concept of predicate pushdown, but is this
> > about predicate push down or something else?
> > # How does one implement a push down for storage format?
> > @Override
> > public boolean supportsPushDown() {
> >
> > # Question 5:
> > # Is this int value of 31 purely randomly selected, or is there an index
> > somewhere?
> > @Override
> > public int hashCode() {
> > return 31;
> > }
> >
> >
> >
> >
>
--
Abdelhakim Deneche
Software Engineer
<http://www.mapr.com/>
Now Available - Free Hadoop On-Demand Training
<http://www.mapr.com/training?utm_source=Email&utm_medium=Signature&utm_campaign=Free%20available>
