From 769544a1e83b6eb95eb84abe5c032444882dbcda Mon Sep 17 00:00:00 2001 From: bchanot Date: Sun, 18 Oct 2026 21:18:34 +0200 Subject: [PATCH] docs(docs): J1 device run recorded as reported (no failure; devices and measures pending confirmation) --- docs/development/PDF_IMPORT_HARDENING.md | 18 +++++++++---- docs/development/R4_J1_RESULTS.md | 32 ++++++++++++++---------- docs/development/RUST.md | 8 ++++++ 3 files changed, 40 insertions(+), 18 deletions(-) diff --git a/docs/development/PDF_IMPORT_HARDENING.md b/docs/development/PDF_IMPORT_HARDENING.md index 293236f9..71c858cc 100644 --- a/docs/development/PDF_IMPORT_HARDENING.md +++ b/docs/development/PDF_IMPORT_HARDENING.md @@ -88,11 +88,19 @@ hides its `/Keywords`, so a stripped attachment on such a file reads as - Option A merged and the CLI on the scanner; no `PDFDocument.load` on any untrusted input in the browser. -- A real-browser test of the inflate bound (Chromium, Firefox, Safari): the - vitest suite runs in Node and cannot see a browser inflater's read-ahead. - The audit of 60ee712 measured +2.2 GB RSS in Chromium on a 4 MB / 4 GiB - bomb before the inflater was fed in 16 KiB slices (+215 MB after); that - measurement is manual until a browser runner exists. +- Done at R4 (2026-10-18, `docs/development/RUST.md` §10.3 to §10.5): the + reader runs in Rust inside a disposable worker; the inflate bound and the + hostile classes were measured in Chromium 156, Firefox 157 and WebKit 27.2 + (headless): a 16 MiB + 1 bomb costs 63 MiB of linear memory and is + refused, 32 bombs 68, a 16 MiB token-dense document 16 MiB under the + 400,000-value cap (784 MiB with it off), the 1 GiB module ceiling traps; + the application build answers the same outcome and message as the + TypeScript reader on 666 of 678 (file, engine) pairs, the other 12 being + the four declared hostile files. Device run (J1, `R4_J1.md`): the owner + reports the functional checks executed on 2026-10-18 with no failure; + the device list, versions, scenario coverage, timings and memory + observations are recorded in `R4_J1_RESULTS.md` as they are confirmed. + The gate below stays open: imports are not offered to external users. - Done: the candidate CPU/byte budget on the scanner: one 16 MiB budget of retained and parsed content shared by every candidate (a cancelled inflation is not charged, so up to 32 oversized streams may each cost one diff --git a/docs/development/R4_J1_RESULTS.md b/docs/development/R4_J1_RESULTS.md index e584bcfb..f2a9c409 100644 --- a/docs/development/R4_J1_RESULTS.md +++ b/docs/development/R4_J1_RESULTS.md @@ -4,24 +4,30 @@ Procedure: `R4_J1.md`. One column per device actually run; `PASS`, `FAIL` or `NON TESTÉ` per cell, with a note under the table for every FAIL, incident, timing and memory event. A device or browser not run stays `NON TESTÉ`. +Status 2026-10-18: the owner reports the functional checks they executed as +passed, with no failure. Which devices, browsers and versions ran, which of +the twelve checks each one covered, the J1.8 / J1.12 timings and the J1.9 / +J1.10 memory observations have not been given yet: every cell below stays +`NON TESTÉ` until the owner confirms it, nothing is inferred. + | Device, browser, version | Surface (1 dev / 2 prod CSP) | Date | |---|---|---| -| | | | +| (to be confirmed by the owner) | | 2026-10-18 | | Check | Device A | Device B | Device C | |---|---|---|---| -| J1.1 reference PDFs open the editor | | | | -| J1.2 hostile folder: message per file | | | | -| J1.3 rescue copies | | | | -| J1.4 engine unavailable (`--csp without`) | | | | -| J1.5 engine recovered | | | | -| J1.6 interrupted import | | | | -| J1.7 offline import | | | | -| J1.8 L export (≈ 42 MB) opens, time to the editor | | | | -| J1.9 six L imports in a row, tab alive | | | | -| J1.10 twenty picks of `values-1m.pdf` | | | | -| J1.11 rescue copy under and over the cap | | | | -| J1.12 timings (demo, L) | | | | +| J1.1 reference PDFs open the editor | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.2 hostile folder: message per file | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.3 rescue copies | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.4 engine unavailable (`--csp without`) | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.5 engine recovered | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.6 interrupted import | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.7 offline import | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.8 L export (≈ 42 MB) opens, time to the editor | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.9 six L imports in a row, tab alive | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.10 twenty picks of `values-1m.pdf` | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.11 rescue copy under and over the cap | NON TESTÉ | NON TESTÉ | NON TESTÉ | +| J1.12 timings (demo, L) | NON TESTÉ | NON TESTÉ | NON TESTÉ | Notes (FAIL, incidents, timings, memory warnings, reloads): diff --git a/docs/development/RUST.md b/docs/development/RUST.md index 69d8dc7f..3d30e90f 100644 --- a/docs/development/RUST.md +++ b/docs/development/RUST.md @@ -1401,6 +1401,14 @@ off (−2.3 over its last five). Twenty consecutive dense imports in WebKit: all refused, no worker `error` event (the one seen in an R4.2e smoke run did not reproduce). +J1 device run (2026-10-18, `R4_J1.md`, owner decisions BDR-059: both +WebKit reserves accepted provisionally, PDF cap kept at 50 MiB): the owner +reports the functional checks they executed as passed, no failure. The +device list, versions, the checks covered per device, the timings (J1.8, +J1.12) and the memory observations (J1.9, J1.10) are recorded in +`R4_J1_RESULTS.md` as the owner confirms them; until then they are not +claimed here. Headless engines never stand for Safari or a phone. + Residual risks and open points (R4.3, for the J1 checkpoint): - WebKit, 50 MiB file: 145 MiB of content growth against 76 for the