calcite-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Julian Hyde <jh...@apache.org>
Subject Re: [Avatica] Separating wire message from service implementation
Date Tue, 11 Aug 2015 04:04:52 GMT
I have logged https://issues.apache.org/jira/browse/CALCITE-839 <https://issues.apache.org/jira/browse/CALCITE-839>
for this task.

> On Aug 10, 2015, at 8:27 PM, Josh Elser <josh.elser@gmail.com> wrote:
> 
> Julian Hyde wrote:
>> If my understand is correct, this decision is an easy one to make, because whatever
we choose, we still comply with the specification.
>> 
>> Avatica specifies its API at two levels: (1) the Java API (Meta), (2) the JSON requests/responses
going over the wire.
>> 
>> I would accept any transport that complies with both of those levels. In principle
we could write a JSON-over-HTTP-marshalled-via-Jackson transport, specifying the mappings
elsewhere than in the Meta classes, and someone else could use another transport.
> 
> Assuming by "specifying the mappings elsewhere" you mean "by a means other than POJOs",
yes, I'd agree. I feel like the POJOs are bound to eventually cause troubles across versions
(as it's very easy to accidentally make some change that breaks something you forgot you did
before).
> 
> Any serialization library isn't foolproof obviously, but they are certainly much easier
to work around than doing it all by hand.
> 
>> Is it feasible to achieve this?
>> 
>> Julian
>> 


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