mirror of
https://github.com/block/buzz.git
synced 2026-08-18 06:50:31 +02:00
npub1cc3ha7z055mu0rwwu7806t2wt8mj3pvu0uv5mfp2c50dahaqhczshdalg6andtlongwell-block
9f57a718a4
The comment cited 1 MiB of RIS as evidence the tail high-water is zero. At 1 MiB the *deleted-pump* arm never reaches TAIL_CAP either, so the number proved nothing: both worlds report "never full" and the measurement agrees with whatever it was pointed at. 4 MiB is also short -- a 16 KiB read of RIS defers 16382 bytes, not 16384, so it stops 512 bytes under the cap. 8 MiB is where the counterfactual fires. Also disambiguates the read count. 257 is reads *completed*; the 256 in the review thread was a zero-based loop index, and a comment that doesn't say which invites the next probe to disagree with it by one. Found by Sami, in wording he had written and I had shipped -- I hit the same defect on my own probe an hour earlier, fixed my run, and copied his number across without noticing it had the flaw I'd just corrected. Co-authored-by: tlongwell-block <109685178+tlongwell-block@users.noreply.github.com> Signed-off-by: tlongwell-block <109685178+tlongwell-block@users.noreply.github.com>