frostvex https://frostvex.icu/ Releases, internals, and benchmarks from the frostvex project. en-us Tue, 14 Apr 2026 12:00:00 +0000 https://frostvex.icu/assets/og-preview.png frostvex https://frostvex.icu/ Frostvex 0.3 — sub-millisecond diffs and a saner manifest https://frostvex.icu/blog/v0-3-released.html https://frostvex.icu/blog/v0-3-released.html Tue, 14 Apr 2026 12:00:00 +0000 Manifest engine rewrite, 60% faster steady-state diffs, breaking changes, migration notes. Benchmarks 2026 — frostvex on a real home network https://frostvex.icu/blog/benchmarks-2026.html https://frostvex.icu/blog/benchmarks-2026.html Sat, 04 Apr 2026 09:00:00 +0000 Four nodes, three different ISPs, one of them on flaky LTE backup. Throughput, retransmits, manifest churn, and where we still lose to plain rsync. Why I rewrote the merge engine in Rust https://frostvex.icu/blog/lockfree-merging.html https://frostvex.icu/blog/lockfree-merging.html Sun, 29 Mar 2026 10:00:00 +0000 The old merge engine used a coarse mutex. Three readers and two writers were enough to bottleneck it. Here's what I replaced it with. Quarantine, not deletion: how 0.2.x learned to forgive flaky disks https://frostvex.icu/blog/quarantine-not-deletion.html https://frostvex.icu/blog/quarantine-not-deletion.html Thu, 12 Feb 2026 14:00:00 +0000 Why frostvex 0.2.x stopped deleting files even when asked. The case for soft removal in sync tools. Picking BLAKE3 over xxh3, and why it took me a week to commit https://frostvex.icu/blog/blake3-over-xxh3.html https://frostvex.icu/blog/blake3-over-xxh3.html Thu, 08 Jan 2026 11:00:00 +0000 Why I switched chunk hashing to BLAKE3 in 0.2.5, and why it took me a week to commit. 2025 year-end retrospective: 11 releases, 3 bugs that ate sleep https://frostvex.icu/blog/year-end-2025.html https://frostvex.icu/blog/year-end-2025.html Sun, 21 Dec 2025 16:00:00 +0000 11 releases, 3 bugs that ate sleep, and what I learned shipping a small CLI in 2025. QUIC for sync: choosing quinn, then patching it https://frostvex.icu/blog/quic-quinn-patches.html https://frostvex.icu/blog/quic-quinn-patches.html Tue, 04 Nov 2025 09:00:00 +0000 Why frostvex transports over QUIC, why I picked the quinn library, and the three small patches that made it work. Reed-Solomon parity for backup pools — a small benchmark https://frostvex.icu/blog/reed-solomon-parity.html https://frostvex.icu/blog/reed-solomon-parity.html Tue, 30 Sep 2025 12:00:00 +0000 A small benchmark of Reed-Solomon parity overhead across 4+2, 6+3, and 8+4 schemes — and why we ended up at 8+4 for the default. Why Rust, after years of Go https://frostvex.icu/blog/why-rust.html https://frostvex.icu/blog/why-rust.html Tue, 19 Nov 2024 14:00:00 +0000 Picking Rust for frostvex was not obvious. Here's the case I made to myself.