It can be a view, or a transformation. You might view data_a with schema_b. Or, you might take binary data, conforming to schema A and directly re-write it to binary data, conforming to schema B. Most Avro APIs don't yet handle workflows that are not 'read' and 'write' -- transformations to or from object representations to serialized forms.
The general case includes all transformation classes as well as views. On 6/8/13 10:16 PM, "Gregory (Grisha) Trubetskoy" <[email protected]> wrote: > > >On Sat, 8 Jun 2013, Scott Carey wrote: > >> In a more general sense it is simply "from" and "to" -- One might move >> from schema A to B without serialization at all, transforming a data >> structure, or simply want a view of data in the form of A as if it was >> in B. > >I'd like to zoom in on this specific point for a little, if I may. > >I think serialization is a red herring. It's always a transformation of >one data structure to another, because a claim could be made that one >cannot transform a serialized form without loading it into a data >structure first. > >In fact, I think it's always the latter case, a *view*, as you aptly >described it. Which makes it not so much a "from" and "to", but more a >"view A as B"? > >Something like: > >value_b = value_a.view_as(schema_b) > >Just my late-night $0.02. > >Grisha
