Hi Ashish, 1) I think it would be great with better function and expression support. Right now we support just a few (aggregate) functions and I also especially would like to see some scalar functions such as type-casting/conversion functions, coalesce functions or maybe even pluggable functions that the user can implement himself. Better formula/expression support is a part of that as well IMO. I wouldn't say that there are "plans" as much as there are certainly wishes :-)
2) I agree to the problem, but could also imagine a more generic solution that would not just be about JOINs nor just about CompositeDataContexts. The reality is that we have a memory space problem in many cases when the data gets big. If we for instance have a GROUP BY query on a unsorted datasource then we currently also use an in-memory structure to solve it. That obviously doesn't scale too well... So a generally pluggable storage mechanism for the query engine would be great. Kasper 2015-04-02 7:14 GMT+02:00 Ashish Mukherjee <[email protected]>: > Hi, > > Would like to bring up couple of features for brainstorming - > > 1. As of now, formula or computations are unsupported in the MetaModel SQL > xyntax. eg. SELECT col1, col2, col1/col2 FROM foo; > > This works only on JdbcDataContext but not with others. > > Is there any plan to allow this kind of SQL in future? > > 2. Currently, MetaModel performs joins etc. on composite data context in > memory within the application space. If my data set is large and I want to > pass this off to an external SQL database, then I am considering the > following approach - > > In CompositeDataContext class - > > Introduce a new public method executeQueryExternally(Query) > > If this method is called, then the results returned by delegate executor > contexts are passed off to the external DB for joining. This method would > be similar to executeQuery() in the same class, which seems to determine > the delegate data contexts and executes parts of the query on them. > > Please validate my thought process. > > Regards, > Ashish >
