Tag a new release

[saschagrunert]

Hi there,

Do you think it would be possible to tag a new release according to semver? Some go module based dependencies have to rewrite their go.mod to get the latest features in. 😇 Thanks in advance! :)

[cyphar]

The main thing I'd like to see be ready before a new runtime-spec release is the resolution of #1008 -- which will allow runc to have a 1.0 release. (Tagging releases, especially spec releases, in OCI is a bit of a hassle.)

[h-vetinari]

It seems the resolution of #1008 is held up mainly by waiting for someone to review the PR. ;-)

[onlyjob]

Formal release is long overdue... Because of that some projects (namely runtime-tools) already deviated from good practice of vendoring by tag...

[vbatts]

to echo @h-vetinari, we are waiting on #1008 for a next release of the spec, and of runc-1.0.0 😬

[h-vetinari]

#1008 has been merged a week ago. Is all that's necessary a PR with a tag? I'm guessing this is better left to the maintainers, but according to several comments all over the place, runtime-spec 1.0.2 should be ready?

[vbatts]

vote for release is posted

[h-vetinari]

#1026 failed due to not meeting the 2/3 quorum of maintainers; since then, two inactive maintainers have stepped down, meaning the quorum could now be reached with the LGTMs from the last vote alone. Since a nit** was brought up by @kad on the vote thread anyway, I thought it'd be good to double-check the state of open PRs - not to further delay the release, but to pick up already finished changes, since runtime-spec releases are few and far between. In descending order of approval:

• config-linux: add SHOULD to linux.namespaces.type #1025; 2 LGTMs (done)
• implementations: add railcar #1003; 2 LGTMs (obsolete)
• config-windows: Added support for network namespace #978; 1 LGTM (obsolete)
• config-linux: describe more about rootfs mount propagation #975; 1 LGTM (done in config-linux: describe more about rootfs mount propagation (cont'd) #1035)
• remove unneeded indent #937; 1 LGTM (done in Rebase #937 #1031)
• Fix typo in RELEASES.md #1033: 1 LGTM (done)
• Clarify case with pre-configured Intel RDT closID #1034: 1 LGTM (Fix for @kad's nit**) (done)

For completeness, the remaining open PRs are probably not ready / necessary for a last-minute merge:

• PRs introducing additional (conditional) MUSTs: config: Do not allow runtimes to ignore properties defined by the spec #680, add RFC requirement for mountLabel #934, runtime.md: only required properties are MUST applied #935, add definite limit for username mappings #939, runtime: Document state annotations as a copy of config annotations #946, maskedPaths and readonlyPaths: skip non-existent paths #982
• TPM support seems reasonably complete but needs at least a rebase: Add vTPM specification #920
• Repo hygiene: RELEASE: document how to do the release #1027, PullApprove: No need for 3 DCO checks #1029

---

**

@kad: There is one questionable statement introduced by PRs #988 and #932 about Intel RDT closID state, which in theory should require small clarification:
One case that is not covered in the description is

"If closID is set and neither l3CacheSchema nor memBwSchema is set".
In this case, I think, runtime should check if closID exists in the resctrl and assign container to it.
If this closID does not exist in resctl - error.

[vbatts]

Good list. Would want to try your hand at carrying over #937 ?

[h-vetinari]

Updated the list.

[h-vetinari]

All the open PRs that could still be considered for the next release are exceedingly trivial and could be finished if two devs took half an hour each.

In any case, now that two months have passed since the first quorum, it's really time to get this release done, IMHO.

[h-vetinari]

Only two PRs left that could realistically still squeeze into 1.0.2. If anyone except @tianon (who already LGTM'd) could look at #1034 #1035, there'd be nothing left but to actually release it. :)
@crosbymichael @mrunalp @vbatts @dqminh @hqhq

[h-vetinari]

All done; thanks @vbatts! 🎉
Any ETA on sending out the vote?