Yes. The writer+reader schema paired usage lets you do efficient partial deserialization. Even if the fields are non-consecutive, in which case it jumps over fields that it doesn't have to deserialize. In your case the deserialization would stop right after the header is deserialized.
On Sun, Nov 23, 2014 at 10:40 AM, Jun Rao <[email protected]> wrote: > Hi, > > Suppose that the writer schema has a header record, if I set the reader > schema to be just the header part in a GenericDatumReader, does it > construct the header record efficiently w/o deserializing the whole record > first? > > Thanks, > > Jun > -- Arvind Kalyan http://www.linkedin.com/in/base16 cell: (408) 761-2030
