RC F'13, F2'17 Cryptogopher / Go cryptography maintainer Professional open source maintainer https://filippo.io / https://github.com/FiloSottile https://mkcert.dev / https://age-encryption.org https://sunlight.dev / https://filippo.io/newsletterhttps://bsky.app/profile/filippo.abyssdomain.experthttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mty4bpl74n2cOk, I think the ML-DSA performance side quest might be complete 🏎️ Very proud of how safe and clear the final incremental changes are, too.26 Aug 2026 10:19 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mty4bpl74n2chttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mtwlziccln2vI am so happy we can now divinate precise diagrams in minutes to help reason through complicated code. (This is the "butterfly" of the Number Theoretic Transform of ML-DSA. It's faster not to reduce non-overflowing intermediates, but I had to convince myself that they do not, in fact, overflow.)25 Aug 2026 19:56 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mtwlziccln2vhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mtm7l7id422dML-KEM and ML-DSA fill matrices/vectors with output from SHAKE, each element derived from slightly different inputs. This is perfect for SIMD instructions, so now we have func ReadMulti(s []*SHAKE, out [][]byte) backed by AVX2 on amd64, and by the existing two-lane asm on arm64. go.dev/cl/81872021 Aug 2026 16:46 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mtm7l7id422dhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mtho7sa54c2yWe got a lot of traces, thank you! The main category we are missing (unsurprisingly) is Windows. We got one (1) pure-Windows session 😅 Please... idk, ask a friend? help-test-crypto-passkey.exe.xyz Also, I worked around an interop issue in KeePassXC and ChiPass. If you use those please try again! [contains quote post or other embedded content]19 Aug 2026 21:25 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mtho7sa54c2yhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mteqjxnlml2vOh, I just realized quantum computers (and the massive size of ML-DSA signatures) will probably kill asymmetrycally-signed JWTs as bearer tokens. This is good. 90% of the time, asymmetric signatures are unnecessary complexity, and a session ID or HMAC would do just as well.18 Aug 2026 17:29 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mteqjxnlml2vhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mtcm4cd4ek2vI am mostly done implementing crypto/passkey, but now I need YOUR help collecting real-world traces for its test suite! Please go to help-test-crypto-passkey.exe.xyz and click the buttons. It should take 1–3 minutes. 𝘌𝘴𝘱𝘦𝘤𝘪𝘢𝘭𝘭𝘺 if you have some unusual Linux-on-the-desktop xkcd 1987 passkey setup.17 Aug 2026 21:04 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mtcm4cd4ek2vhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mt7c4it7tv2vFrom the "things that would have been too much effort before but take no time with LLMs and @exe.dev" category. Yay for personal software.16 Aug 2026 13:27 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mt7c4it7tv2vhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mt4ojj435c2gI think something broke in my relationship with the tech world. Not due to AI, but to threads like this. I thought my community cared about reality. I block ads with uBO Lite in Chrome. Anyone has access to verify this reality by experience. And yet pages and pages of comments on... not reality. https://news.ycombinator.com/item?id=4930320215 Aug 2026 12:31 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mt4ojj435c2ghttps://bsky.app/profile/filippo.abyssdomain.expert/post/3msxlkxd3l22rHuh, it’s not every day that an LLM suggests on the issue tracker a low-hanging fruit change that saves 30% on X25519 handshakes. We already had a much faster edwards25519 fixed-base scalar mult, which can be used for half of ECDH with a small mapping. (The other half of X25519 is “special.”) https://go-review.googlesource.com/c/go/+/81460113 Aug 2026 11:55 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3msxlkxd3l22rhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3msqid3r23q2bOn Saturday night, one of our CT logs rejected most submissions for 30 minutes. Here's the post-mortem. It was... a lot of fun? It involves Go mutex starvation, SQLite WAL behavior, and ZFS record sizes. I got to SIGKILL a VM 200 times, implement a turnstile (TIL!), and order a Nokia flip phone! https://groups.google.com/a/chromium.org/d/msgid/ct-policy/b1b5e8a2-b010-4e2a-b8c8-91c1d1b7ca91%40app.fastmail.com?utm_medium=email&utm_source=footer10 Aug 2026 16:08 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3msqid3r23q2bhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3msohm3xftk2wI'm watching this, and sure sure the agents coordinating is neat, but once again, WHY ARE WE CHILL WITH Artifactory HAVING SEVERAL RCEs, SSRFs, and unauthorized writes. Why are we chill with Hugging Face having RCEs. Why are we chill with GitHub having RCEs (unrelated, from April). [contains quote post or other embedded content]09 Aug 2026 20:50 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3msohm3xftk2whttps://bsky.app/profile/filippo.abyssdomain.expert/post/3msj4dfbihb2uIt’s very clear by now that if LLMs are not improving your software quality it’s either a revealed preference (yours or your org’s) for more volume vs more quality, or a skill issue. The level of testing and review they are enabling in the Go cryptography standard library is amazing.07 Aug 2026 17:45 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3msj4dfbihb2uhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3ms2dim2phs2xPaging @masnick.com to the Streisand effect courtesy phone. An “engine problem“ is a complete non-issue and it was corrected in 24h! But Go2Sky sued @avherald.com! Now I’m reading about their 2016 incident they wanted to hide, and I’d never, ever fly an airline with this approach to transparency. [contains quote post or other embedded content]01 Aug 2026 20:43 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3ms2dim2phs2xhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrxt2np2ws2lAfter a year and a quarter of operating the Tuscolo Certificate Transparency log (with a total of 8 minutes of planned downtime)… I'm happy to announce the second Geomys CT log: Trastevere! It's basically the same, except it's a Dell PowerEdge R6515 racked in Seeweb's Frosinone, Italy datacenter. https://groups.google.com/a/chromium.org/d/msgid/ct-policy/b5730b92-6a20-4067-8b61-49e2099e91eb%40app.fastmail.com?utm_medium=email&utm_source=footer31 Jul 2026 20:44 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrxt2np2ws2lhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrs25so75s2wI'm seeing folks draw the wrong conclusion (in good faith or not) from the HAWK attack. HAWK is a scheme that 1. cryptographers were suspicious of and 2. was still in the assessment process. A break is GOOD. It means the process is useful, and it INCREASES confidence in the selected algorithms.29 Jul 2026 13:35 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrs25so75s2whttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrrgu2p6rc2nI know it’s not most folks‘ primary concern, but LLMs or not, I’m unimpressed by how soft these infrastructure services are. What do you mean HF had a Jinja2 template injection. And I’m still not over GitHub’s unsandboxed RCE. Geomys might need to self-host code/CI to avoid a weak link.29 Jul 2026 07:50 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrrgu2p6rc2nhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrkytd7dwc2tI rigged up frood (my Alpine initramfs NAS running on an Ampere Altra, which continues to work great) to have a direct 2.5 Gbps Ethernet link to the secondary 5 Gbps fiber uplink router, so it can do downloads faster and without saturating the primary. I don't know why I didn't do it sooner! https://github.com/FiloSottile/mostly-harmless/commit/e5f289cae32cbebc219c62eb0a1c370c44f24b7026 Jul 2026 18:23 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrkytd7dwc2thttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrke6zps6c2fI keep finding new obvious LLM use cases. I have this old tool, md-ld, which collects orphan Markdown link references, so you can just type prose and then find links later, while preparing to publish. *Of course* Claude Code with just WebSearch and WebFetch can fill them out for me from context. https://github.com/FiloSottile/mostly-harmless/commit/9957940b67eced0097aca29e8ac8322aa15e38c626 Jul 2026 12:13 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrke6zps6c2fhttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrkcv4cyiu2tIt's not my usual beat, but I wrote a pure-Python ML-DSA verifier. pip install mldsa It's 350 lines, CC0/0BSD, single-file, no dependencies, and thoroughly tested. Signature verification handles no secrets, so it doesn't need to be constant-time. https://words.filippo.io/mldsa-py/26 Jul 2026 11:50 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrkcv4cyiu2thttps://bsky.app/profile/filippo.abyssdomain.expert/post/3mrhodqa53k25Get your own totally official Geomys FIPS 140-3 Entropy Source! 🎰 Only at @gophercon.com in Seattle, next month. Compatibility with any Go version is not guaranteed 👀25 Jul 2026 10:37 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrhodqa53k25https://bsky.app/profile/filippo.abyssdomain.expert/post/3mrg5dybmpk2hJFrog Xray is defective, flagging unaffected binaries as vulnerable, and the cost is borne by maintainers who have to field requests to work around it. This is a real open source sustainability issue. (The problem here is JFrog, not the person who politely opened the issue.) https://github.com/FiloSottile/age/issues/73024 Jul 2026 20:00 +0000at://did:plc:x2nsupeeo52oznrmplwapppl/app.bsky.feed.post/3mrg5dybmpk2h