Just added another JPA extension to Cayenne Classic - a mechanism to
return scalars from SQLTemplate. Let me demonstrate how this works.
Take a normal query execution example:
List results = context.performQuery(query);
* In Cayenne Classic each element in the "results" list is either a
Persistent or a DataRow.
* In JPA each element is either a Persistent or a scalar (such as
java.lang.Integer), or an Object[] containing a mix of the previous
two types.
So the JPA difference is that (a) scalars are first-class citizens as
far as query result is concerned; (b) multiple scalar/objects per row
can be a part of the same result; [(c) no concept of DataRow, but
that's not relevant for this discussion].
So what I did now is adding support for scalars and arrays of scalars
to the "raw" queries. So now a count can be done as SQLTemplate in a
following manner:
// a bit involved query construction procedure
String sql = "SELECT count(1) AS C FROM ARTIST";
SQLTemplate query = new SQLTemplate(Artist.class, sql);
query.setColumnNamesCapitalization
(SQLTemplate.UPPERCASE_COLUMN_NAMES);
SQLResultSetMapping rsMap = new SQLResultSetMapping();
rsMap.addColumnResult("C");
query.setResultSetMapping(rsMap);
// very simple scalar result set retrieval
Number count = (Number) DataObjectUtils.objectForQuery(context,
query);
We can hide SQLTemplate creation internals inside a CountQuery, but a
user can easily take care of result processing without extra utility
API (see the last line in the code above - no need to unwrap a Map or
anything).
Regarding the earlier concern that DataObjectUtils is dealing with
non-DataObjects, looks like a conceptual shift of "what a query
result is" solves that as well. Now we consider any
"java.lang.Object" a valid result, only some objects are persistent
and have identity, and others - do not. Comments?
Andrus
On Jun 6, 2007, at 4:15 PM, Michael Gentry wrote:
I don't mind resurrecting QueryUtils, but is a count method the
only thing
that would go in it?
Thanks,
/dev/mrg
On 6/5/07, Andrus Adamchik <[EMAIL PROTECTED]> wrote:
[taking to dev]
There was an old QueryUtils class that was used for some internal
obscure queries. It's long gone (don't even remember which version
dumped it). So now we can reuse the name for the user-friendly
utility class (if we decide to go this way).
Andrus
On Jun 5, 2007, at 4:54 PM, Michael Gentry wrote:
> Hasn't QueryUtils gone poof in 3.0?
>
> /dev/mrg
>
>
> On 6/5/07, Andrus Adamchik <[EMAIL PROTECTED]> wrote:
>>
>> So where do we put it then? QueryUtils?
>>
>> Andrus
>>
>>
