Re: Discussion about a Flink DataSource repository

2016-05-06 Thread Fabian Hueske
Yes, you can transform the broadcast set when it is accessed with RuntimeContext.getBroadcastVariableWithInitializer() and a BroadcastVariableInitializer. 2016-05-06 14:07 GMT+02:00 Flavio Pompermaier : > That was more or less what I was thinking. The only thing I'm not sure is > the usage of the

Re: Discussion about a Flink DataSource repository

2016-05-06 Thread Flavio Pompermaier
That was more or less what I was thinking. The only thing I'm not sure is the usage of the broadcasted dataset, since I'd need to access tot the MetaData dataset by sourceId (so I'd need an Map. Probably I'd do: Map meta = ...;//preparing metadata lookUp table ... ds.map(MetaMapFunctionWrapper(new

Re: Discussion about a Flink DataSource repository

2016-05-06 Thread Fabian Hueske
Hi Flavio, I'll open a JIRA for de/serializing TableSource to textual JSON. Would something like this work for you? main() { ExecutionEnvironment env = ... TableEnvironment tEnv = ... // accessing an external catalog YourTableSource ts = Catalog.getTableSource("someIdentifier"); tEnv.

Re: Discussion about a Flink DataSource repository

2016-05-05 Thread Flavio Pompermaier
HI Fabian, thanks for your detailed answer, as usual ;) I think that an external service it's ok,actually I wasn't aware of the TableSource interface. As you said, an utility to serialize and deserialize them would be very helpful and will ease this thing. However, registering metadata for a table

Re: Discussion about a Flink DataSource repository

2016-05-04 Thread Fabian Hueske
Hi Flavio, I thought a bit about your proposal. I am not sure if it is actually necessary to integrate a central source repository into Flink. It should be possible to offer this as an external service which is based on the recently added TableSource interface. TableSources could be extended to be