docs: roadmap to 1.0.0 GA from a source review against the user-feedback corpus - #212
Merged
Conversation
…ack corpus Reviews XERJ's source against user-feedback/, the 130-file collection of real Elasticsearch pain points, and asks which of those reasons-to-leave XERJ actually removes today. Method: domain audits restricted to XERJ retrieval for code discovery (no grep, the repo's own selfaudit index), each adversarially verified by an independent pass that opened every cited file:line. Two verifiers returned not-sound and corrected real errors, including an audit that concluded pipeline aggregations were missing when aggs.rs implements them. That correction alone removed a roadmap item. Verdict: not GA today. Ten blocking items, none large. The pattern behind almost all of them is that the core answers are real and the edges lie: health endpoints that cannot report failure, scroll silently truncating exports at 10k, management state erased by a restart, sidecar corruption degrading silently, a Helm chart inverting the engine's own secure default, and published numbers that do not survive a count. The document states its own limits rather than hiding them. Ten is a floor, not a ceiling: it covers four audited domains, and security needs its own pass before the tag. GA here means single-node GA; the cluster data plane is deliberately unwired. Issues filed alongside this review: #195 #196 #198 #199 #200 #201 #202 #203 #204 #205 #206 #207 #208 #209 #210 #211.
xerj-org
force-pushed
the
docs/ga-roadmap
branch
from
August 8, 2026 05:20
1fb89a5 to
ac79186
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Reviews XERJ's source against
user-feedback/, the 130-file collection of real Elasticsearch pain points, and asks which of those reasons-to-leave XERJ actually removes today.Method. Domain audits restricted to XERJ retrieval for code discovery, using the repo's own source indexed into a selfaudit index. No grep. Each audit was then attacked by an independent verifier that opened every cited
file:line. Two verifiers came back not-sound and corrected real errors. One found that an audit had concluded pipeline aggregations were missing whenaggs.rsimplements them, because the audit trusted a stale doc instead of the code. That correction removed a roadmap item, which is the most useful thing a verifier can do.Verdict: not GA today. Ten blocking items, none large.
The pattern behind almost all of them is the same. The core answers to Elasticsearch's worst pains are real and verified: no shard decision to make, bounded unclean restart, memory budgets derived from the cgroup limit, auth on by default, nothing paywalled, no deep-pagination cliff, no refresh-interval consistency gap. The edges lie: health endpoints that cannot report failure, scroll silently truncating exports at 10,000 documents, management state erased by a plain restart, sidecar corruption degrading silently instead of loudly, a Helm chart inverting the engine's own secure default, and published numbers that do not survive a count.
The document states its own limits. Ten is a floor, not a ceiling: it covers four audited domains, and security needs its own pass before the tag. GA here means single-node GA, since the cluster data plane is deliberately unwired.
Issues filed alongside this review: #195 #196 #198 #199 #200 #201 #202 #203 #204 #205 #206 #207 #208 #209 #210 #211.
The biggest single finding is #204: eight independent cases of XERJ accepting an instruction, returning success, and doing nothing. That is a missing convention rather than eight oversights.