Re: python precommit error - google-auth depenedency?

BTW, the new pip resolver seems to do the right thing by installing rsa==4.0 instead of 4.2 in this case (upgrade pip to 20.1.1): pip --unstable-feature=resolver install apache-beam[gcp] https://discuss.python.org/t/an-update-on-pip-and-dependency-resolution/1898/4 On Thu, Jun 11, 2020 at 2:40

Re: python precommit error - google-auth depenedency?

> In python 2 oauth2client's rsa>3.14 requirement will resolve to latest python2 supporting version of rsa (4.0?) Unfortunately rsa 4.1 didn't set a python_requires stanza to prevent the breakage of Py2 users, opened: https://github.com/sybrenstuvel/python-rsa/issues/152. On Wed, Jun 10, 2020 at

Re: Remove EOL'd Runners

I have "archived" the `runner-gearpump` Jira component. This makes it so new issues cannot be associated. It can be unarchived any time. It does not close or hide associated issues (makes sense: in Jira's data model, issues are not "in" components, but associated with one or more) Kenn On Wed,

Re: Automation for Jira

Yes, my inbox is hit as well. I'm enjoying going through some old bugs actually. One takeaway is that we have a lot of early Jiras that are still relevant, and also that there are a lot of duplicates. I think some automation to help find duplicates might be helpful. Also, some accidental

Application For Google Summer Of Docs

Sir, First of all, I am Ayush Grover and I have been a technical content writer, for now, two years now and this year I was very excited to apply at Gsod and especially in Apache Beam If you be so kind, can you help me out just by briefing me about what all I can do to make the application look

Re: [External] Re: Ensuring messages are processed and emitted in-order

Hi, I'm afraid @RequiresTimeSortedInput currently does not fit this requirement, because it works on event _timestamps_ only. Assigning Kafka offsets as event timestamps is probably not a good idea. In the original proposal [1] there is mention that it would be good to implement sorting by

Re: [External] Re: Ensuring messages are processed and emitted in-order

I would not recommend using RequiresTimeSortedInput in this way. I also would not use ingestion time, as in a distributed environment, time skew between workers might mess up the order. I will ping the discussion on the sorted state API and add you. My hope is that this can be implemented

Re: DoFnSignature#isStateful deprecated

On 6/11/20 5:02 PM, Reuven Lax wrote: On Thu, Jun 11, 2020 at 1:26 AM Jan Lukavský > wrote: Hi, I'd propose the following:  - delete all DoFnSignatures.{usesState,usesTimers,...} helpers *except* for DoFnSignatures.isStateful Why? Actually seems

Re: DoFnSignature#isStateful deprecated

On Thu, Jun 11, 2020 at 1:26 AM Jan Lukavský wrote: > Hi, > > I'd propose the following: > > - delete all DoFnSignatures.{usesState,usesTimers,...} helpers *except* > for DoFnSignatures.isStateful > Why? Actually seems like maybe the opposite is better? (remove isStateful and keep the others).

Re: DoFnSignature#isStateful deprecated

Hi, I'd propose the following:  - delete all DoFnSignatures.{usesState,usesTimers,...} helpers *except* for DoFnSignatures.isStateful  - DoFnSignatures.isStateful would be equal to 'signature.usesState() || signature.usesTimers() || signature.processElement().requiresTimeSortedInput()'