flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Fabian Hueske <fhue...@gmail.com>
Subject Re: sorting groups
Date Thu, 18 Jun 2015 10:08:29 GMT
The reason for this restriction is that KeySelector keys (i.e., keys that
are extracted using a function) require special case handling at runtime.
If we allow combinations of KeySelector keys with other keys for grouping
and groupSorting, we have four different cases to cover compared to two. So
this is not a technical limitation but mainly due to code complexity and
development time.

I agree, that restriction should be better documented.

2015-06-17 15:34 GMT+02:00 Michele Bertoni <michele1.bertoni@mail.polimi.it>
:

>  Got it,
> I solved the problem changing the grouping function:
> instead of group by r._1 (int the topk case)
> now I group using a function that return r._1
>
>  thus both, group and sort, have a function as parameter
> is there a reason why it works in this way? probably you should make it
> very clear in the documentation
>
>
>  michele
>
>
>  Il giorno 17/giu/2015, alle ore 08:35, Michele Bertoni <
> michele1.bertoni@mail.polimi.it> ha scritto:
>
>  Hi Fabian,
> My dataset is of this type
> RegionType (Long, String, Long, Long, Char, Array[GValue])
> Where GValue is a case class implemented by
> GString(v:String)
> GDouble(v:Double)
>
> I have two case of sorting:
> In the first (topk) i have to group by the first field of the regions and
> sort by a set of fields of the GValue array
>
> In the second (topg) i have to sort by the first field of the regions and
> by a set of fields of the array, then sort by one field of the array
>
> For grouping i am using the groupby function with a function as parameter
> that creates the hash of the desired fields, something like
> ds.groupby((r:RegionType) =>
>   s = new stringBuilder
>   s.append(r._1)
>   grouping.init.foreach((index:int) =>
>     s.append("#")
>     s.append(r._6(index))
>   )
>   Md5.hash(s.toString)
> )
>
> Then i sort it using (in the topg case, the second)
> .sortGroup(((r:RegionType)=>
>   r._6(grouping.last ) /*here i am doing some cast, i am writing from my
> smartphone i don't remember all the details sorry*/ ),Order.ASCENDING)
>
> in the first case instead i group only by r._1 and i have a recursive
> function that appends sortgroup operator to the grouoed dataset
>
> Is there a way to solve this?
>
> I think i don't understand what a keySelector is
>
>
> Thanks!
> Michele
> ------------------------------
> *Da:* Fabian Hueske <fhueske@gmail.com>
> *Inviato:* martedì 16 giugno 2015 23.43.03
> *A:* user@flink.apache.org
> *Oggetto:* Re: sorting groups
>
>    Hi,
>
>  the error is related to the way you specify the grouping and the sorting
> key.
> The API is currently restricted in the way, that you can only use a key
> selector function for the sorting key if you also used a selector function
> for the grouping key.
>
>  In Scala the use of key selector functions is often not very obvious.
>
>  If you post the groupBy().sortGroup() command and the input type, I can
> help you getting it right.
>
>  Cheers, Fabian
>
> 2015-06-16 23:37 GMT+02:00 Michele Bertoni <
> michele1.bertoni@mail.polimi.it>:
>
>> Hi everybody,
>> I am trying to sorting a grouped dataset, but i am getting this error:
>>
>> Exception in thread "main"
>> org.apache.flink.api.common.InvalidProgramException: Sorting on KeySelector
>> keys only works with KeySelector grouping.
>>         at
>> org.apache.flink.api.scala.GroupedDataSet.sortGroup(GroupedDataSet.scala:113)
>>         at
>> it.polimi.genomics.flink.FlinkImplementation.regionOperation.OrderRD$.sort(OrderRD.scala:82)
>>         …
>>
>> can anybody help me understanding the error?
>> i have no idea what it means and google is not helpful in this case
>>
>>
>> thanks!
>> cheers
>> Michele
>
>
>
>

Mime
View raw message