AI search engine and MCP server for bots and agents. One Rust binary indexes code, docs, logs, JSONL/CSV and PDFs for BM25 full-text, vector and hybrid search, code search without grep, agent memory and RAG. Elasticsearch-compatible API; Corpus Hub serves hundreds of reference memory corpora.
Rust agent-memoryai-agentsbm25claude-codeclicode-searchcursordatabaseelasticsearchfull-text-searchhybrid-searchjsonllocal-firstmcpmcp-servermodel-context-protocolragrustsearch-enginevector-search
chore(release): a withdrawal that did not finish, and #405 wrong a second time
xerj-org3f830e4a2bca
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex fc044ad63..feae0fb80 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -415,7 +415,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0archives, cosign-compatible signatures and reproducible builds, was replaced bya `RELEASE INTEGRITY` section stating what actually ships: a `.sha256`, which isa checksum and not a signature or an attestation. The public-sector page lost- its `SBOM · SLSA` compliance row and its at-rest-encryption and BYOK claims.+ its `SBOM · SLSA` compliance row and one at-rest/BYOK line. Others on that page+ survive — two of them carrying no status marker at all — so the withdrawal is+ partial, not complete (#491).A verified air-gapped deployment recipe was added in their place (#430).diff --git a/ROADMAP.md b/ROADMAP.mdindex 129cd2d62..4609fc784 100644--- a/ROADMAP.md+++ b/ROADMAP.md@@ -108,13 +108,15 @@ measured repro:touch only the first member and report success([#450](https://github.com/xerj-org/xerj/issues/450)); scroll through such an alias reportsthe alias as `_index` ([#433](https://github.com/xerj-org/xerj/issues/433)); the scroll- snapshot cap is enforced on the summed total by `/{index}/_search?scroll=`, which fails- loudly, but `/{index}/_search_scroll` applies it per index and only raises a- `scroll_truncated` flag ([#405](https://github.com/xerj-org/xerj/issues/405)).+ snapshot cap is compared against the summed total by `/{index}/_search?scroll=` but applied+ per index by `/{index}/_search_scroll`, so a multi-index scroll that the first route refuses+ is accepted by the second. Both fail with the same 400 when they do refuse — the divergence+ is which total the ceiling is measured against, and it is permissive rather than lossy+ ([#405](https://github.com/xerj-org/xerj/issues/405)).- **Documents that do not come back the way they went in.** `POST /_bulk` drops explicit`null` fields from `_source`, and whether it drops them depends on request size- ([#415](https://github.com/xerj-org/xerj/issues/415)). The rc.18 notes cite all four of- these as open; they are listed here so this file and the CHANGELOG do not disagree.+ ([#415](https://github.com/xerj-org/xerj/issues/415)). #405 and #415 are cited as open by+ the rc.18 notes as well; #450 and #433 appear only here, which is the point of listing them.- **Autoindex.** Catalog IDs are global, so two corpora sharing one byte-identical filecollide ([#416](https://github.com/xerj-org/xerj/issues/416)); and widening an exclusionrule wedges the default graph path — the first rerun exits 1 and needs a documented