Publication Updates

November 3, 2026

Library review, reproducible labs, and source corrections

We re-read every guide against its sources and rewrote all twenty. Several problems described as fixed in the October 12 entry below were still present in the published text; this review corrected them and is recorded here so readers can see what changed.

• Added runnable labs to eighteen guides (seventeen in Go, and one on PostgreSQL 18.6 for the row-level security guide), with the exact program and the output recorded on November 3, 2026. The other two guides include code with passing unit tests. They are listed on the new Reproducible Labs page. Several labs found defects in code the guides previously published, including a job scheduler that never served late-arriving tenants, a deduplication order that lost usage events, and an audit-log verifier that skipped the first record.
• Replaced code that was unsafe or did not work: a tenant setting built with string formatting (an SQL injection pattern), an OpenTelemetry transport that did not compile, a Rego policy that did not parse, and a webhook ordering guard that Stripe's documentation advises against.
• Corrected source citations. Some previous entries carried titles that did not match the linked page, pointed to a home page instead of a document, or named the wrong organization or section. Every link was opened again and the label now matches the destination. The "Verified" label is now "Link checked", which is what the date records.
• Removed invented figures and examples, including a table of customer margins presented as real data, fixed latency and recovery numbers without a measurement, and a statistic about churn after data leaks.
• Removed TeX formulas that the site cannot render and that appeared as raw text in four guides.
• Corrected the text inside all twenty guide diagrams, which still showed claims removed from the articles, such as unmeasured latencies, invented recovery rates, "exactly-once" and "zero downtime" guarantees, and per-tenant point-in-time restore.
• Corrected technical errors, including the claims that separate PostgreSQL databases give per-tenant point-in-time recovery, that adding a column takes no locks, and that W3C baggage appears on spans automatically.
• Turned each topic page into a guide to its section, with an orientation, a suggested reading order, each guide's lab finding, and the relevant calculator.
• Added a revision history to every guide and build checks that reject raw TeX, duplicate or non-HTTPS sources, and broken contents links.

October 12, 2026

Planning tools and transparency review

• Added a multi-tenant capacity planner with visible traffic, transfer, and storage formulas.
• Added a tenant isolation decision planner that compares shared tables, schema-per-tenant, and database-per-tenant models using disclosed fit scores.
• Added a tenant unit economics calculator for equal-share and usage-weighted infrastructure allocation.
• Published the editorial standards used for sources, code examples, bylines, correction dates, and commercial separation.
• Rewrote the About and privacy descriptions to match the site’s actual search and tool behavior and removed unsupported claims of peer review, production testing, and anonymous professional credentials.
• Corrected the session-management guide’s refresh-token rotation example, updated it to current OAuth and NIST guidance, and documented concurrency, clock-skew, and revocation-store failure modes.
• Removed absolute or malformed claims from the migration, enterprise SSO, and tenant rate-limiting guides so the recommendations preserve workload and contract context.
• Corrected cloud-cost terminology and clarified that successful cross-tenant canary tests are evidence for review, not mathematical proof of compliance.
• Revised database-isolation, encryption-key, audit-log, queue-fairness, telemetry-sampling, webhook-idempotency, and migration guidance to replace universal performance or security promises with testable limits and failure conditions.
• Reworked the status-page and SLA guide around correct downtime math, representative probes, agreement-specific remedies, and reviewed credit proposals; removed a generic penalty schedule presented as a standard.
• Added contextual links between the tools and the implementation guides they support.

The new calculators perform their work in the visitor’s browser. Tool inputs are not posted to the publication, and each result includes its assumptions and exclusions.

September 2026

Initial guide library

The first publication set established four focused sections: multi-tenant architecture, billing and entitlements, enterprise security, and cloud operations. Each guide includes an on-page table of contents, key takeaways, implementation examples, related reading, and a list of references used during review.