Re: Why does `Combine.perKey(SerializableFunction)` require same input and output type

On Mon, Oct 31, 2016 at 8:39 PM, Kenneth Knowles wrote: > Manu, I think your critique about user interface clarity is valid. > CombineFn conflates a few operations and is not that clear about what it is > doing or why. You seem to be concerned about CombineFn versus >

Re: Why does `Combine.perKey(SerializableFunction)` require same input and output type

I'm a bit confused here because neither of them requires same type of input and output. Also, the Javadoc of Globally says "It is common for {@code *InputT == OutputT}, but not required" *If associative and commutative is expected, why don't they have restrictions like

Re: Why does `Combine.perKey(SerializableFunction)` require same input and output type

Thanks guys. My original confusion comes from that if the API allows me to have different types of input and output, why not make it easier. It's clear now. Do you think it's better to hide some interfaces we don't expect users to use ? The Combine API has lured me a lot to do more than it

Re: Why does `Combine.perKey(SerializableFunction)` require same input and output type

Manu, I think your critique about user interface clarity is valid. CombineFn conflates a few operations and is not that clear about what it is doing or why. You seem to be concerned about CombineFn versus SerializableFunction constructors for the Combine family of transforms. I thought I'd respond

Re: Why does `Combine.perKey(SerializableFunction)` require same input and output type

GlobalCombineFn and PerKeyCombineFn still expect an associative and commutative function when accumulating. GlobalCombineFn is shorthand for assigning everything to a single key, doing the combine, and then discarding the key and extracting the single output. PerKeyCombineFn is shorthand for doing

[ANNOUNCE] Beam 0.3.0-incubating Released

Hello Everyone, I'm very pleased to announce that we just release a new version. Ideally, you should just have to update your Maven POMs, update your code and be good to go. This is still very much a learning release where we are not releasing a specific new feature set. Nevertheless, these are

Re: [ANNOUNCE] Beam 0.3.0-incubating Released

Excellent! On Mon, Oct 31, 2016 at 9:49 AM Aljoscha Krettek wrote: > Hello Everyone, > I'm very pleased to announce that we just release a new version. Ideally, > you should just have to update your Maven POMs, update your code and be > good to go. > > This is still very

Beam App: MEM<-->Disk Space Issue

Hi Colleagues,Perhaps some of these questions should be asked in Flink forum. Pls let me know if thats the case so I can post it there  as well. I am facing something new when running my Beam app in a 4 nodes Flink Cluster.I list the behavior items:1- Dashboard shows all nodes actively running2-