flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Attila Bernáth <bernath.at...@gmail.com>
Subject Re: VertexUpdateFunction class has memory between the supersteps
Date Thu, 16 Oct 2014 11:15:05 GMT
So I thought I can use this fact to remember something about these nodes.


2014-10-16 11:25 GMT+02:00 Stephan Ewen <sewen@apache.org>:
> Yes, that is guaranteed.
>
> Am 16.10.2014 10:23 schrieb "Attila Bernáth" <bernath.athos@gmail.com>:
>
>> For example: an instance of the vertexupdatefunction processes certain
>> vertices in one superstep (in its updateVertex method). Can I assume
>> that it will process the same vertices in the next superstep, too?
>>
>> Attila
>>
>> 2014-10-16 1:34 GMT+02:00 Stephan Ewen <sewen@apache.org>:
>> > Hi!
>> >
>> > I am not sure that I understand the question. What memory between
>> > supersteps
>> > are you referring to?
>> >
>> > If you mean that the same instances of the functions are used across all
>> > supersteps and that their variables are kept: yes, you can rely on that.
>> >
>> > Stephan
>> >
>> > Am 15.10.2014 17:09 schrieb "Attila Bernáth" <bernath.athos@gmail.com>:
>> >
>> >> Dear All,
>> >>
>> >> I am using the spargel API. I found that the VertexUpdateFunction
>> >> class has memory between the supersteps. Is it safe to rely on this
>> >> memory?
>> >>
>> >> Attila

Mime
View raw message