I'll try my code on a more recent release to see if it works. Thanks for
pointing that fix out
On Fri, Feb 7, 2025 at 3:58 PM Joey Tran wrote:
> It is not different! I am using beam version 2.50 and it looks like the
> changes associated with that test were in 2.56. Were custom window types
> no
It is not different! I am using beam version 2.50 and it looks like the
changes associated with that test were in 2.56. Were custom window types
not supported before then?
On Fri, Feb 7, 2025 at 3:44 PM Robert Bradshaw wrote:
> On Thu, Feb 6, 2025 at 8:39 AM Joey Tran
> wrote:
>
>> Thanks all f
On Thu, Feb 6, 2025 at 8:39 AM Joey Tran wrote:
> Thanks all for the conceptual sanity check. I went ahead and tried
> implementing a data-based window type. I've gotten most of the way there
> but it was surprisingly difficult. Not really because of any complications
> due to the window not bein
It should help. Adding @Yi Hu @Steven van Rossum
@Sam Whittle who would be
able to give a more definitive answer.
On Fri, Feb 7, 2025 at 9:07 AM Utkarsh Parekh
wrote:
> Hi Team,
>
>
>
> I came across this PR and wanted to check if it addresses the issue of
> multiple kafka connections being c
Hi Team,
I came across this PR and wanted to check if it addresses the issue of multiple
kafka connections being created by Dataflow jobs.
https://github.com/apache/beam/pull/32928
Would appreciate any insights.
Thank you,
--
Utkarsh
Warning: All e-mail sent t
I see that there is an issue with caching of UnboundedReaders in the
UnboundedSourceAsSdfWrapper which is used by the v2 dataflow runner for
executing UnboundedSources (which JMS source is).
This is https://github.com/apache/beam/issues/32968.
This could result in more connections as each DoFn is
