Hi all,
Would a committer be willing to cut a pulsar-client-go release from the
current master? The last tag, v0.21.0, went out on Jul 31; master is seven
commits ahead and has been quiet since Aug 26.
What is on master and untagged:
b3d8930e fix: avoid infinite retry in blue-green migration test (#1502)
8213b20c feat: add RemoveBacklogQuotaByType to namespace (#1525)
b47b6901 [fix][consumer] Initialize batch ack tracker from the broker
ack set on
redelivery (#1521)
bf4809d8 [fix][consumer] Use resolved partition topic for
single-partition zero queue
consumer (#1531)
fd604fa7 feat(pulsaradmin): add PIP-401 batchingConfig to ProducerConfig
(#1530)
61d7a95e fix(pulsaradmin): preserve explicit zero receiver queue size
(#1529)
(plus the 0.21.0 changelog commit.) The two consumer fixes, #1521 and
#1531, look like the ones users would most want in hand.
My concrete interest, for transparency: terraform-provider-pulsar currently
pins the pseudo-version v0.21.1-0.20260826023458-61d7a95e66cd, because it
needs #1530 to expose producer batching configuration. That commit,
61d7a95e66cd, is master's current tip — so a tag cut today would contain
exactly the tree that provider already builds and tests against, making the
switch a one-line go.mod edit with no behavioural change to validate.
Downstreams pinning pseudo-versions to reach a merged feature doesn't seem
like a good steady state, and this is the cheapest possible moment to end
it.
On the version number: I had assumed v0.21.1, but master carries two feat
commits (#1525 and #1530), so v0.22.0 is arguably the more honest tag. I
have no stake in which — what I need is simply a release containing
fd604fa7. I do see the 0.22.0 milestone still has 24 open issues against
it, so if a full 0.22.0 is some way out, either a patch cut now or a
re-scoped milestone would work equally well from my side. That tradeoff is
yours to make, not mine.
RobertIndie has been RM for the last three releases — no expectation that
it falls to you again, just flagging in case you already have a cadence in
mind here.
Happy to take on whatever would make this cheaper, including drafting the
changelog for the seven commits.
Thanks,
David
