JB, Ayush, thanks for sharing the resource. Looks good to me!

Robert, thanks for the explanation. To restate my concern from another
thread, VEX lets Polaris explain whether a known dependency vulnerability
actually affects a particular Polaris release. However, a dependency
missing from the inventory could still escape detection, so missing
information should not be interpreted as evidence of safety.

I think it’s fine to proceed as long as we clearly document these
limitations and set the right expectations for users about what our VEX
statements cover.

Yufei


On Sat, Nov 14, 2026 at 10:37 PM Robert Stupp <[email protected]> wrote:

> JB,
>
> Thanks for clarifying.
> The ATR presentations from the ASF Infrastructure track [2] from Community
> Over Code Glasgow are provides useful background.
>
> Yufei,
>
> Which specific SBOM question do you consider unresolved?
> I've answered JB's questions in the SBOM thread.
>
> Your coverage concern points to the limitation already documented in the
> README [2]: dependencies embedded inside Iceberg's bundled JAR are not
> fully inventoried.
> The generated SBOMs explicitly mark completeness as unknown.
> Are you asking for additional disclosure, or for that coverage to be
> expanded before publishing the initial SBOMs?
>
> Suitable upstream SBOMs from Iceberg and other dependencies could help
> expand coverage, provided they can be matched to the bundled artifacts.
> Pointers to those, or help making them available upstream, would be
> welcome.
> Reconstructing those inventories within Polaris would be separate work
> requiring its own validation; it is outside the proposed initial scope.
>
> Let's continue that discussion in the SBOM thread so the requested change
> and its scope are clear.
>
> Robert
>
> [1] https://www.youtube.com/watch?v=Vdw-0mryT2M&list=PLdDygsL9TxLk
> [2]
>
> https://github.com/snazy/polaris/blob/a00eaa5e465acba9e35956a72491f840ffea3229/gradle/sbom/README.md?plain=1#L35-L36
>
>
> On Sun, Nov 15, 2026 at 7:20 AM Jean-Baptiste Onofré <[email protected]>
> wrote:
>
> > Hi Yufei,
> >
> > ATR does not replace our current release process, but it will help
> automate
> > several steps. As a result, I don't expect a major change in how we work.
> >
> > I am currently conducting a complete review of ATR's usage and will share
> > more details soon.
> >
> > Regards,
> > JB
> >
> >
> > On Sat, Nov 14, 2026 at 8:30 PM Yufei Gu <[email protected]> wrote:
> >
> > > Thanks, JB, for raising this and volunteering to do this. The timeline
> > > looks good to me. +1 on trying ATR. Does ATR replace the current
> release
> > > process? If not, any change is needed? I'd also document the new
> process
> > so
> > > that people can refer to it next time.
> > >
> > > Robret, it'd perferrable to resolve open questions before moving on
> > > regarding SBOM.
> > >
> > > Yufei
> > >
> > >
> > > On Sat, Nov 14, 2026 at 1:21 AM Robert Stupp <[email protected]> wrote:
> > >
> > > > Hi JB,
> > > >
> > > > Thanks for volunteering!
> > > >
> > > > I agree that we should move towards ATR. It's been on our radar for a
> > > long
> > > > time.
> > > >
> > > > We're pretty close with the SBOM work for the Maven artifacts for
> 1.10.
> > > > However, we'd also need reproducible SBOMs for the Python parts.
> > > >
> > > > Targeting mid-December sounds reasonable to me.
> > > >
> > > > polaris-tools releases are a separate effort and should not block ATR
> > for
> > > > Polaris "main" releases.
> > > >
> > > > Robert
> > > >
> > > > On Sat, Nov 14, 2026 at 6:35 AM Jean-Baptiste Onofré <
> [email protected]>
> > > > wrote:
> > > >
> > > > > Hi folks,
> > > > >
> > > > > Now that Polaris 1.9.0 is almost out, I would like to propose
> Polaris
> > > > > 1.10.0 by mid December (just before Christmas).
> > > > > I'm volunteering to be the release manager on this release.
> > > > >
> > > > > I would also like to propose to evaluate the use ATR for this
> > release (
> > > > > https://releases.apache.org/).
> > > > >
> > > > > ATR is the Apache Trusted Releases platform. It's a centralized
> > > > > release-management and distribution platform built by The ASF
> Tooling
> > > > team
> > > > > to modernize how Apache projects build, verify, vote on, and
> publish
> > > > > releases.
> > > > > ATR provides a standardized pipeline that ensures every artifact
> > > > downloaded
> > > > > by end users is authentic, unmodified, and compliant with ASF
> policy:
> > > > > For that, the capabilities are:
> > > > > - Automated Compliance & Security Checks: During the initial
> Compose
> > > > phase,
> > > > > ATR automatically validates release candidates for cryptographic
> > > > > signatures, checksums, license headers, and Software Bills of
> > Materials
> > > > > (SBOMs), giving Release Managers immediate feedback before a vote
> is
> > > even
> > > > > called.Streamlined
> > > > > - Voting & Publishing: ATR manages the entire release lifecycle
> > across
> > > > > three stages (Compose, Vote, and Finish) automating vote
> tabulation,
> > > > > stripping release-candidate (RC) tags upon approval, and tracking
> > > > > distribution to downstream channels like Maven Central and PyPI.
> > > > > - Trusted Publishing via CI/CD: Through dedicated GitHub Actions
> and
> > > > OpenID
> > > > > Connect (OIDC) authentication, projects with reproducible builds
> can
> > > > > trigger and manage releases directly from their CI/CD pipelines
> > without
> > > > > relying on long-lived credentials.
> > > > > - Release Catalog & Attestations: ATR generates release
> attestations
> > > and
> > > > > maintains a public catalog of current and archived ASF artifacts,
> > > helping
> > > > > downstream users verify provenance and meet regulatory standards
> such
> > > as
> > > > > the EU Cyber Resilience Act.
> > > > >
> > > > > You can find more details https://releases.apache.org/docs/
> > > > >
> > > > > There are few pre-requisites to use ATR (reproducible build, etc),
> so
> > > I'm
> > > > > doing a pass.
> > > > >
> > > > > I will keep do an analyze soon and I will keep you posted.
> > > > >
> > > > > Regards
> > > > > JB
> > > > >
> > > >
> > >
> >
>

Reply via email to