oscerd opened a new pull request, #1711:
URL: https://github.com/apache/camel-website/pull/1711

   Adds a blog post about how the July security batch was handled, aimed at 
readers who saw "32 CVEs in one release" and want to know what that actually 
means.
   
   Rather than walking the list, the post is about the response process:
   
   - **Release cadence.** A table of every 2026 release that carried security 
fixes. The clustering is the point: five releases across three lines in nine 
days in April, three in four days in July.
   - **The LTS lines got more, not less.** 4.18.3 shipped 34 advisories and 
4.14.8 shipped 27, against 4.21.0's 32. The extras are findings whose main-line 
fix landed in an earlier minor.
   - **Incomplete fixes.** Both cases in this batch (CVE-2026-40048 → 
CVE-2026-46590, CVE-2026-40860 → CVE-2026-43866) were corrected in the next 
release on every line, each with its own CVE and advisory. Also covers 
CVE-2026-42527, the deserialization filter our own hardening series broke.
   - **Why 32 at once.** One report gets read as a pattern and swept across the 
component set, which is why 22 of the 32 are the same broken header boundary in 
22 components, and why 4.21.0 also renamed the header constants across 30+ 
components so the filter covers them by construction.
   - **Upgrade matrix**, including the one finding (CVE-2026-49042, 
`camel-langchain4j-tools`) that affects 4.14.x but was not backported because 
the fix changes behaviour, with the advisory workaround spelled out.
   - **Credits** for the six external researchers, and a reference table of all 
32.
   
   ### Data provenance
   
   Every number is derived from files in this repo and cross-checked before 
committing:
   
   | Claim | Source |
   |---|---|
   | 4.21.0 fixed 32 | 32 advisories in `content/security/` with `4.21.0` in 
`fixed:` |
   | 4.18.3 fixed 34, 4.14.8 fixed 27 | same query per version |
   | 26 of the 32 backported to 4.14.8 | intersection of the 4.21.0 set with 
`4.14.8` |
   | 5 HIGH / 26 MEDIUM / 1 LOW | `severity:` across the 32 |
   | Yu Bao credited on 24 | `credit:` across the 32 |
   | All release dates | `date:` in the `RELEASE-*` announcement posts |
   
   All 32 advisories in the batch appear in the reference table, and every CVE 
link resolves to an existing file under `content/security/`.
   
   ### Note on the Trust page
   
   While checking these numbers I found that `content/trust/_index.md` says the 
4.14.8, 4.18.3 and 4.21.0 releases "fixed and disclosed 32 vulnerabilities 
between them". Read strictly that is 34, since two findings shipped in 4.18.3 
and 4.14.8 with an earlier main-line fix. It under-counts rather than 
overstates, but it does not say what it computes, and it now reads 
inconsistently against this post. Fixed separately in #PLACEHOLDER.
   
   ### Not verified
   
   I could not render the page locally: no Hugo binary and no `node_modules` in 
the working copy. The post uses only Markdown already present in sibling posts, 
with no shortcodes or raw HTML.
   
   🤖 Generated with [Claude Code](https://claude.com/claude-code)


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to