Fork me on GitHub
#onyx
<
2016-04-24
>
lucasbradstreet14:04:30

@lvh @michaeldrogalis you can always use a lifecycle to parameterise the function by pulling a larger dataset from somewhere else on a peer’s task start

lvh15:04:11

right; I was just thinking something like storing the actual data somewhere and just using a URL or whatever

lucasbradstreet15:04:42

Yeah, that would work. You would then pull it down and inject it into the params via the lifecycle instead of parameterizing directly in the task-map entry

michaeldrogalis16:04:12

Yep, good point simple_smile

Drew Verlee19:04:09

does anyone have a short introductory opinion on why flink has two apis rather then one like onyx or dataflow?

michaeldrogalis19:04:20

@drewverlee: I believe they're working towards having 1 now

gardnervickers19:04:45

Historically they were batch-orientated I believe.

Drew Verlee19:04:14

ok cool, ill dig around a bit. Thanks for a starting point.