[CI] Integration failure: staging instance

[github-actions]

Staging instance failure

A scheduled test against Sigstore's staging instance has failed.

This suggests one of three conditions:

• A backwards-incompatible change in a Sigstore component;
• A regression in sigstore-python;
• A transient error.

The full CI failure can be found here:

https://github.com/sigstore/sigstore-python/actions/runs/7772634242

[woodruffw]

Context: this failed because we failed to parse the checkpoint record in Rekor's response.

Triaging: looks like Rekor 1.3.5 removed the timestamp field from the checkpoint: sigstore/rekor#1888

This should be trivial for us to fix, per convo in the PR above.

[woodruffw]

Opened #891 for this.

[jku]

Yes, even the "spec" on transparency-dev says:

The first 3 lines of the body MUST be present in all Checkpoints.

So we're not compliant when we require 4 lines.

[woodruffw]

Yep, opened #891 with the fix.

[haydentherapper]

Whoops, sorry that broke tests! Yea, the timestamp is part of extra data (which should be parsed, just not required), and it served no purpose in the ecosystem.

[woodruffw]

No problem at all! This was definitely us not reading the spec closely enough 🙂

[jku]

We should do a release in near future so we're ahead of this changing in production rekor.

[woodruffw]

We should do a release in near future so we're ahead of this changing in production rekor.

Yeah, agreed. We're mid-burn for 3.0, so we may have to cut another 2.x release and backport this.

[woodruffw]

I've created https://github.com/sigstore/sigstore-python/tree/series/2.1.x to track a point release to 2.1.x.

[woodruffw]

I've cut 2.1.2 as a backport release. We can continue to use series/2.1.x for further releases on that line 🙂