drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jacques Nadeau <jacq...@apache.org>
Subject Re: RecordBatchLoader.load(...) SchemaChangeException
Date Mon, 04 May 2015 20:27:18 GMT
Daniel, your description is correct.  Let's remove if we're not using them
anymore.

On Mon, May 4, 2015 at 1:18 PM, Daniel Barclay <dbarclay@maprtech.com>
wrote:

> To clarify:
>
> load(...) returns a boolean value to indicate whether there was a schema
> change.
>
> Initially, I thought that throwing SchemaChangeException was an old way of
> indicating a schema change, and so thought that the throws declaration and
> various catches were obsolete.
>
> However, reportedly SchemaChangeException is or was for reporting
> /problems/ in making schema changes, not for reporting normal schema
> changes.
>
> Therefore, the code is not /necessarily /as obsolete as I initially
> thought.
>
> With that clarification, I re-submit the question:
>
>
> I wrote:
>
>> In RecordBatchLoader, the load(...) method is declared to throw
>> SchemaChangeException, but it never actually throws SchemaChangeException.
>>
>> It supposed to be declared to throw SchemaChangeException?  (E.g., are we
>> reserving the "right" for load(...) to throw that, and declaring "throws
>> SchemaChangeException" to help make sure callers already handle it in case
>> load(...) later changes to actually throw it sometimes?)
>>
>> Or is that "throws" a remnant that should be removed sometime?
>>
>>
>> Daniel
>>
>
>
> --
> Daniel Barclay
> MapR Technologies
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message