Thank you all for your feedbacks.
@Kenn, @Robert I got what you mean now. I think it's good to make
window/timestamp/paneinfo values configurable while also avoid redundant
encoding and decoding.
Besides, I think the idea from @Luke is good, that we can replace
WindowedValue with T in some cases
On Fri, Nov 8, 2019 at 9:23 AM Luke Cwik wrote:
>
>
> On Thu, Nov 7, 2019 at 7:36 PM Kenneth Knowles wrote:
>
>>
>>
>> On Thu, Nov 7, 2019 at 9:19 AM Luke Cwik wrote:
>>
>>> I did suggest one other alternative on Jincheng's PR[1] which was to
>>> allow windowless values to be sent across the
On Thu, Nov 7, 2019 at 7:36 PM Kenneth Knowles wrote:
>
>
> On Thu, Nov 7, 2019 at 9:19 AM Luke Cwik wrote:
>
>> I did suggest one other alternative on Jincheng's PR[1] which was to
>> allow windowless values to be sent across the gRPC port. The SDK would then
>> be responsible for ensuring
I did suggest one other alternative on Jincheng's PR[1] which was to allow
windowless values to be sent across the gRPC port. The SDK would then be
responsible for ensuring that the execution didn't access any properties
that required knowledge of the timestamp, pane or window. This is different
I think there is some misunderstanding about what is meant by option
2. What Kenn (I think) and I are proposing is not a WindowedValueCoder
whose window/timestamp/paneinfo coders are parameterized to be
constant coders, but a WindowedValueCoder whose
window/timestamp/paneinfo values are specified
Thanks for your feedback and the valuable comments, Kenn & Robert!
I think your comments are more comprehensive and enlighten me a lot. The
two proposals which I mentioned above are to reuse the existing coder
(FullWindowedValueCoder and ValueOnlyWindowedValueCoder). Now, with your
comments, I
Yes, the portability framework is designed to support this, and
possibly even more efficient transfers of data than element-by-element
as per the wire coder specified in the IO port operators. I left some
comments on the doc as well, and would also prefer approach 2.
On Wed, Nov 6, 2019 at 11:03
I think the portability framework is designed for this. The runner controls
the coder on the grpc ports and the runner controls the process bundle
descriptor.
I commented on the doc. I think what is missing is analysis of scope of SDK
harness changes and risk to model consistency
Approach 2:
Hi all,
I am trying to make some improvements of portability framework to make it
usable in other projects. However, we find that the coder between runner
and harness can only be FullWindowedValueCoder. This means each time when
sending a WindowedValue, we have to encode/decode timestamp, windows