Proposal for ORAS Search Working Group

[sajayantony]

As a follow up to the ORAS calls I would like to propose creating the working group for search

Aligning with - #33

• The ORAS Search Working group is tasked with defining how OCI artifacts can be searched and will define the scenarios that the search would be applicable to.

• The following members would be a part of the initial working group and will present to the ORAS Maintainers the proposed next steps once the above scenarios have been defined.

  • @sabre1041
  • @SteveLasker
  • @TerryHowe
  • @toddysm

• The working group will be dissolved once this proposal has been made to the Maintainers and it is upto the maintainers to define the next step.

• The working group will validate the scenarios and proposals against prior art and other similar projects that align with similar scenarios.

• The maintainers will create a wg-search repository to facilitate issues and/or proposals for the scenarios. The working group may also use the #oras-wg-search to involve other members on cncf.slack.io.

[toddysm]

I believe the following folks also expressed interest in becoming part of the working group (i.e. maintainers for the WG repository): @afflom @jpower432 @sabre1041 @dmesser @rchincha @AaronFriel @scottrigby

[toddysm]

The above folks expressed interest in the following issue oras-project/artifacts-spec#72 Except @scottrigby who expressed interest privately on Slack

[SteveLasker]

Thanks @sajayantony. As discussed in the ORAS meeting last evening, we don't have a consensus for creating the definition or governance of a "working group" under ORAS. CNCF has several working group definitions, and I don't believe we should add yet another definition under ORAS.
Some CNCF Working Group Examples

• Cloud Native Network Function Working Group (CNF-WG)
• GitOps Working Group

ORAS Governance has a definition for sub-projects, that can have a scope defined. Just as we did for the ORAS Artifacts-spec.

Subproject owners are responsible for activities surrounding the development and release of content (eg. code, specifications, documentation)...

My suggestion would be to give the group that's been noodling around search a chance to assemble a scope and collection of "maintainers". The oras-org maintainers can then vote if they wish to accept the sub-project.
If we want to re-define what working groups are, under oras, that's a separate process that I personally don't think we should undertake.

As such, we should either retitle or close this issue, pending a proposal from the folks thinking about search.

[rchincha]

Happy to sign up as a maintainer, if no objections of course.

[sajayantony]

This issue is about building consensus on the WG creation. Time-boxing this is needed to make progress.

I am willing to take the effort to do the right thing here and define the process as needed.

If we call it a sub project, we repurpose governances for a much more fluid effort that should end and hopefully become a sub project.

Will put it on the call for next week if we can’t get maintainers to agree on GitHub on this issue.

Also would like comment from: @afflom @jpower432 @sabre1041 @dmesser @rchincha @AaronFriel @scottrigby

On being added to this effort.
I see @rchincha already responded.

[afflom]

Let's get this started and count me in!

[jpower432]

Please count me in as well!

[SteveLasker]

@sajayantony the only problem is trying to create a new definition of a "working group" under oras. This definition came out of left field, perhaps with some other discussions that aren't apparent to others.

I don't see a problem with saying the oras-search sub-project is timeboxed. It's the same result without creating a new category, so can you clarify why we need a new process, vs. defining the scope of the sub-project?

If you are insistent we use "working group" terminology, with new governance definition, I think we should defer to larger CNCF working group governance.

[AaronFriel]

I'm interested, I'm ambivalent on the question of sub-project or working group and will leave that to the current ORAS project leads to determine.

[sabre1041]

Count me in on this effort

[lachie83]

@sajayantony the only problem is trying to create a new definition of a "working group" under oras. This definition came out of left field, perhaps with some other discussions that aren't apparent to others.

I don't see a problem with saying the oras-search sub-project is timeboxed. It's the same result without creating a new category, so can you clarify why we need a new process, vs. defining the scope of the sub-project?

If you are insistent we use "working group" terminology, with new governance definition, I think we should defer to larger CNCF working group governance.

The concept of a working group has been modeled across a few different projects and the value I've seen is to create a space to come up with a scope and agree on a proposal from a focused group prior to writing a line of code or committing to a sub-project. I think that is a value process to adopt as the project can transparently iterate on the idea then come back with a formalize proposal. The following diagram is a good representation of working group to sub-project relationship from Kubernetes.

https://github.com/kubernetes/community/blob/master/kubernetes_governance_diagram.png

Specifically, it states the following regarding working groups:

• Organized to solve a specific problem, then dissolve
• Cross-SIG collaboration around a specific effort
• Does not own code
• Can spawn sub-projects in participating SIGs
• Define and address gaps

I don't think landing working groups in the governance docs for ORAS should be long process and believe the project stands to benefit form working groups in the long run. There is a already a proposal here - #33 (comment). I don't believe that the CNCF provides larger working group governance however I will research and let you know.

[FeynmanZhou]

Creating a WG seems not to conflict with creating a subproject. It appears to be more autonomous and open, with lower barriers to participation. It might be established temporarily to specifically discuss and solve a particular problem or requirement. Subproject might be the outcome of the WG.

The only challenging thing is the WG organizers need to run it proactively and independently, such as organizing dedicated meetings, Slack channel, etc.

I would suggest having a majority of consensus on the definition of WG from the org owners before creating the first WG in the ORAS and updating the governance doc. So @shizhMSFT @SteveLasker @sabre1041 @SteveLasker @TerryHowe would you be able to vote for the WG definition in @toddysm 's issue #33 and provide your justification?

[scottrigby]

The above folks expressed interest in the following issue oras-project/artifacts-spec#72 Except @scottrigby who expressed interest privately on Slack

Thanks @toddysm yes, publicly now, I am interested in and working on helping bootstrap the WG.

If you are insistent we use "working group" terminology, with new governance definition, I think we should defer to larger CNCF working group governance.

I don't believe that the CNCF provides larger working group governance however I will research and let you know.

@lachie83 Good point about Kubernetes Working Groups. And I would agree if there were no governance or structure for CNCF Working Groups then adding that here would make sense. But this does exist.

In CNCF yes there is a structure with governance in place. There are 2 kinds of CNCF WGs from a governance point of view:

1. those under a CNCF TAG, which follow that TAG's governance
2. those run by CNCF/LF staff, which follow CNCF TOC processes

This new registry search WG would be the first kind, under TAG App Delivery.

I'm interested, I'm ambivalent on the question of sub-project or working group and will leave that to the current ORAS project leads to determine.

@AaronFriel exactly

CNCF Working Groups are meant to be short-lived, with any lasting documents or code to live in projects or sub-projects. This is what we did with the GitOps WG and OpenGitOps project (currently sandbox). There is a relationship between the two.

We can do the same here:

1. A Registry Search Working Group (under TAG App Delivery)
2. A project to hold code and other implementation work in progress (a sub-project of ORAS)

[scottrigby]

Thomas Schuetz proposes that we follow the same approach as any other WG. To express our interested in establishing this, we need to prepare a charter, with the Scope, Goals, and interested people, and let's discuss this in a TAG meeting.

Let's do this in this Google doc in progress that folks above have put together: https://docs.google.com/document/d/1rcQROZP31q7BOjoZ977Ok7pt28z_UXfW0vAK3xC0wdI/edit#

I have to run right now, so can someone add the categories above to the doc? Other detail or topics could live in a linked, separate document if needed, but for now this all seems fine. This is the bootstrap phase.

[joshgav]

Hi folks - one of the capabilities we've posited for a cloud-native platform is artifact storage (last but not least in that list), from container images to software and system packages.

There are a lot of projects trying to create standard bundle formats within OCI too, like Acorn, Carvel kapp, rukpak, CNAB.

Perhaps a WG could take on the broader scope of reducing complexity for artifact storage generally. ArtifactHub would be a prime member of such a WG. We might get some ideas from PackagingCon too.

Wanted to put this on your radar, if you prefer something focused solely on ORAS that makes sense too!

[afflom]

Perhaps a WG could take on the broader scope of reducing complexity for artifact storage generally. ArtifactHub would be a prime member of such a WG. We might get some ideas from PackagingCon too.

@joshgav , I agree with this and this is something that we tried to address in the vision outlined within the gdoc that @scottrigby linked above. Would you broaden the scope from what is proposed in that draft?

[sabre1041]

I am encouraged by the trajectory of this thread as we are finding an amenable resolution to the registry search initiative.

@joshgav as we have seen, more and more content types is becoming applicable in the cloud native landscape, and this effort becomes even more important as a means to discover and search content. '
@scottrigby @lachie83 Thank you for your insight into some of the prior art and assets that can support this effort

I would also suggest the previously referenced document be used to collect everyones thoughts in the document and come to an agreement on the scope and goal for this effort.

In terms of where this effort should land, given that not only is there applicability within an existing CNCF TAG and there are participants from multiple communities, it could belong to the TAG App Delivery for which there is already an interest.

Finally, I will emphasize, that while the remaining logistical activities are completed, we provide a medium for those who want to collect their thoughts be able to so that innovation can begin to occur that can drive value for the entire community moving forward.

[toddysm]

Just pinging this thread to ask if there is any plan to continue with this effort in ORAS or somewhere else? The interest in this is growing from our side and I just wanted to hear back from the folks on this thread what are the next steps we should take.

[SteveLasker]

Hi @toddysm, @joshgav is setting up a working group under the CNCF TAG App Delivery to help with the breadth of scenarios across multiple projects.
While the group is spinning up, the details, specific to search, are being captured here

[scottrigby]

Hi @toddysm, @joshgav is setting up a working group under the CNCF TAG App Delivery to help with the breadth of scenarios across multiple projects. While the group is spinning up, the details, specific to search, are being captured here

Thanks, @SteveLasker. Plus – for those following along – the Artifacts WG bootstrap charter is now in a new doc, here, where you can add yourself to the interested parties list and collaborate async via comments. The initial artifact registry search doc Steve linked to above is one of several topic-based activities on the WG, and the doc is updated now to reflect that.