Alexey,

Any reason why you propose this only for raw data? I think this should be
an option for binary marshaller (probably per type). If we also add special
methods to Binarylizable, then they should be applied to both raw and
non-raw data.

BTW, there are couple other tickets regarding this, and looks like there is
already some progress:
- https://issues.apache.org/jira/browse/IGNITE-5097
- https://issues.apache.org/jira/browse/IGNITE-6418

-Val

On Fri, Sep 29, 2017 at 5:37 AM, Alexei Scherbakov <
alexey.scherbak...@gmail.com> wrote:

> Dmitriy,
>
> Not arrays, just primitives.
>
> Using special binary representation, ints and longs can be represented by
> 1-9 bytes, depending on their cardinality.
>
> I want to add such implementation in raw binary API to help implementing
> custom serializers.
>
>
>
> 2017-09-29 15:12 GMT+03:00 Dmitriy Setrakyan <dsetrak...@apache.org>:
>
> > Alexey, are you talking about arrays of ints and longs?
> >
> > On Fri, Sep 29, 2017 at 3:29 AM, Alexei Scherbakov <
> > alexey.scherbak...@gmail.com> wrote:
> >
> > > Guys,
> > >
> > > I notices we do not have support for packed ints and longs in raw
> binary
> > > API [1] [2]
> > >
> > > Such methods are essential for implementing efficient custom
> compression
> > > schemes.
> > >
> > > Their addition can simplify implementing custom serializers for the
> cases
> > > then default binary marshaller is not enough, without additional
> library
> > > dependencies.
> > >
> > > Proposed API extension for rawReader/rawWriter:
> > >
> > > org.apache.ignite.binary.BinaryRawWriter#writePackedInt
> > >
> > > org.apache.ignite.binary.BinaryRawWriter#writePackedLong
> > >
> > > org.apache.ignite.binary.BinaryRawReader#readPackedInt
> > >
> > > org.apache.ignite.binary.BinaryRawReader#readPackedLong
> > >
> > > JIRA ticket: [3]
> > >
> > > Thoughs ?
> > >
> > > [1] org.apache.ignite.binary.BinaryRawReader
> > >
> > > [2] org.apache.ignite.binary.BinaryRawWriter
> > >
> > > [3] https://issues.apache.org/jira/browse/IGNITE-6426
> > >
> > > --
> > >
> > > Best regards,
> > > Alexei Scherbakov
> > >
> >
>
>
>
> --
>
> Best regards,
> Alexei Scherbakov
>

Reply via email to