Andre, Purbon

I'm certainly a fan of the idea and others seem to be as well as you
can see here [1].

This is certainly something which would be quite helpful as a way for
encouraging developers to really think about their implementations and
to help ensure they communicate the considerations along to a user
building a flow.  These can be a bit hard to judge and relative but a
great example are processors which read everything into memory.  Large
objects will have a large memory impact and users really do need to
know that so they dont go wild adding more and more threads, etc..

[1] https://issues.apache.org/jira/browse/NIFI-440

On Fri, Mar 24, 2017 at 4:01 AM, Pere Urbón Bayes <[email protected]> wrote:
> Hi,
>    i find this an interesting idea. I am new to nifi, so probably missing
> some history bits. however i can see situations where processors have to
> enhance data inside the data flow without having/willing to do the
> attr/data roundtrip.
>
> Another interesting annotation, or flag, would be for cpu intensive
> processors. What do you think?
>
> /purbon
>
> On Fri, 24 Mar 2017 08:42 Andre, <[email protected]> wrote:
>
>> dev,
>>
>> While I know we all would prefer if processors only used data streaming
>> within onTrigger calls, sometimes processors must load data into memory
>> prior to action.
>>
>> Should we have an annotation to flag those unfortunate processors that for
>> a particular reason must load flowfile contents into memory before doing
>> their magic?
>>
>> Cheers
>>

Reply via email to