e3e3a68cb3
T-505 item 3: the owner's defaults by performance — the parts by part on, the express second frame off; item 4's start
main
NotAlexNoyleandClaude Fable 5.12026-11-03 13:00:06 -05:00
d331e4ed29
T-505 item 3: the parts' pictures by part (engine 0073), the express second frame back; item 4 next
NotAlexNoyleandClaude Fable 5.12026-11-03 12:47:59 -05:00
4e1e9a9ae4
PICKUP: the engine on RB_20261027 across the three variants; the push script for both engine branches
NotAlexNoyleandClaude Fable 5.12026-11-03 10:27:17 -05:00
62af6e82c7
The engine on UXP RB_20261027 (T-551..T-554): the queue rebased, 0060 dropped, every pin, the device checked
NotAlexNoyleandClaude Fable 5.12026-11-03 10:25:46 -05:00
6c0a12ca30
Desktop tests: the harness caught up with the engine host; the ARM build takes a jobs count
NotAlexNoyleandClaude Fable 5.12026-11-03 09:53:27 -05:00
1d16d55855
PICKUP: T-505's gate and verify frame closed; RB_20261027 found, the owner: the engine update next
NotAlexNoyleandClaude Fable 5.12026-11-03 08:53:06 -05:00
c29f396c21
T-505: the long page's wrong verify frame found and fixed (engine 0073); the second review's fixes
NotAlexNoyleandClaude Fable 5.12026-11-02 20:30:35 -05:00
1a0026e923
PICKUP: paused for a fresh agent; T-505 state, the open items, the device on this head
NotAlexNoyleandClaude Opus 5.52026-10-26 14:54:54 -04:00
81f56bef6d
T-505: the parts' picture in the frame's own job (E2, engine 0072); the review's fixes
NotAlexNoyleandClaude Opus 5.52026-10-26 14:43:50 -04:00
2bd3e6f1ae
T-505: the sticky parts on the CPU (behind a switch); own-slot frames diffed without the scroll (engine 0070, 0071)
NotAlexNoyleandClaude Opus 5.52026-10-26 13:44:33 -04:00
4586bf6731
Kits: low-power engine tuning, the engine's next release, the webOS SDK, the Atlas references
NotAlexNoyleandClaude Opus 5.52026-10-26 12:26:53 -04:00
554ed514dc
T-505: the cause proven, the card measured; the card's frame wrapped without alpha
NotAlexNoyleandClaude Opus 5.52026-10-26 11:52:33 -04:00
a507ee867c
T-504: a double tap out after a pan while zoomed in drawn at once from a half-resolution overview (engine 0069)
NotAlexNoyleandClaude Opus 5.52026-10-26 09:56:42 -04:00
d9816d2fe2
PICKUP: the scripted pan moves the card, R3 re-opened, T-544 done; the device on the packaged head (engine 078a0942)
NotAlexNoyleandClaude Opus 5.52026-10-26 07:20:29 -04:00
f50e49c128
T-544: the card holds sticky sidebars and fixed bars still between frames of a pan
NotAlexNoyleandClaude Opus 5.52026-10-26 07:02:14 -04:00
fe3fa819ea
The scripted pan moves the card, as a finger does; with it, desktop Wikipedia under a finger is measured at ~7 frames a second
NotAlexNoyleandClaude Opus 5.52026-10-26 06:09:09 -04:00
000e1c580e
Build site: T-516 DONE (both levers, the owner's choice); T-504 and T-505 in progress with the owner's decisions (a bounded overview; the card shifting its frame, behind a switch)
NotAlexNoyleandClaude Opus 5.52026-10-26 05:30:40 -04:00
292d857b9d
T-516: a gesture's first frame on two cores and its view read alone — double taps 20/20 within 250 ms; desktop Wikipedia's pan 24.0 fps with the sensors running
NotAlexNoyleandClaude Opus 5.52026-10-26 05:30:11 -04:00
ec46afd103
T-516: both levers measured — the first zoom frame's view read alone (40/40 within 250 ms, slowest 227) and the second core online for the gesture (40/40, slowest 161)
NotAlexNoyleandClaude Opus 5.52026-10-26 05:16:29 -04:00
a0c9061c78
T-516 measured read-only: no GPU clock ramp; the slow double tap's readback runs on one contended core with the second unplugged after a pause
NotAlexNoyleandClaude Opus 5.52026-10-26 04:48:14 -04:00
5abea8bcb3
T-414: the live add-on site checked — smoke 43/43 from the PC, a colour Persona (Dracula) worn from it on the TouchPad by finger and taken off; left: the same in Pale Moon (the owner's)
NotAlexNoyleandClaude Opus 5.52026-10-26 04:40:33 -04:00
48a4e21e46
PICKUP and tracking: T-527 done; the three decisions for the owner; the device on the packaged head
NotAlexNoyleandClaude Opus 5.52026-10-26 03:24:12 -04:00
f19af6acaf
Build site: T-527 DONE, T-515 DONE, T-517 DONE; T-504, T-505, T-516 blocked on the owner's calls (a page overview for a zoom out after a pan; the card composing over its frame; a GPU clock floor)
NotAlexNoyleandClaude Opus 5.52026-10-26 03:23:44 -04:00
a1213c7418
Kits: three criteria from T-527's second review, met (a rootless page's zoom painted at its resolution; the zoom-back frame from before the zoom in; the search results' length follows the filter)
NotAlexNoyleandClaude Opus 5.52026-10-26 03:01:43 -04:00
31fce6143d
T-527: the second review's fixes — the zoom-back frame kept by the lane that sees the zoom first, its publish clear of reads in flight; about:addons' search filter re-measured
NotAlexNoyleandClaude Opus 5.52026-10-26 03:01:26 -04:00
dc7bae02be
Engine 0066 (T-527 review): the paint skip repaints after a resolution change and asks the child widget; a failed band surface retried after 30 s
NotAlexNoyleandClaude Opus 5.52026-10-26 03:01:03 -04:00
a10a9a7937
T-527 records: double taps in runs of ten by finger — 20 of 20 zoomed, out 0-3 ms every time, in 52-251 ms
NotAlexNoyleandClaude Opus 5.52026-10-26 02:21:56 -04:00
45a63a9209
about:addons: the search results' rows say Uninstall too; their list pans by finger (chrome-look R2 ticked)
NotAlexNoyleandClaude Opus 5.52026-10-26 02:10:22 -04:00
384fd3c6ec
PICKUP: about:settings by finger done in all three; the device's pages are head's
NotAlexNoyleandClaude Opus 5.52026-10-26 01:59:55 -04:00
4b05514102
about:settings by finger, the rest: the Fonts box, Forget, Update, the go buttons in Enyo; every tab and each kind of control in Mochi and Mojo (preferences-ui R6 criterion 1 ticked)
NotAlexNoyleandClaude Opus 5.52026-10-26 01:58:43 -04:00
3aaa6148d9
T-527 (R3): the scripted pan at 24 fps or better on the owner's bench — desktop Wikipedia 24.6 / 24.2, the long page 34.2 (sensor daemons stopped); the gaps classified
NotAlexNoyleandClaude Opus 5.52026-10-26 01:08:29 -04:00
15f6612ef9
Impl tracking: iteration 20 (T-527's review fixes, engine 0064/0065, the zoom-back frame; the finger items)
NotAlexNoyleandClaude Opus 5.52026-10-26 00:59:31 -04:00
062f3d8c3a
PICKUP: the 2026-10-26 delta — T-527's review fixes, engine 0064/0065, the double tap out's frame, about:settings and about:addons by finger, the device state
NotAlexNoyleandClaude Opus 5.52026-10-26 00:59:12 -04:00
08993767c9
about:addons' long lists pan by finger in Mochi and Mojo too (the packages); the test pads removed from all profiles
NotAlexNoyleandClaude Opus 5.52026-10-26 00:58:33 -04:00
5e1d6f405f
Build site: T-527 — the review's fixes, engine 0064/0065, the double tap out's frame at once, two levers measured and not kept
NotAlexNoyleandClaude Opus 5.52026-10-26 00:33:49 -04:00
1d614f46d6
T-527 records: two compositor-job levers measured and not kept (a region-based read of the change; the own slot with fixed layers)
NotAlexNoyleandClaude Opus 5.52026-10-26 00:33:25 -04:00
286a926569
T-527 (R6): a double tap out shows a frame at once — the frame kept at the zoom in is published again on the way back
NotAlexNoyleandClaude Opus 5.52026-10-26 00:16:45 -04:00
623dc1acf7
T-527: the review's fixes — XUL layout height per document, pan frames under a zoom of 0.5, the rootless offset per document, the fling lead bounded, the adoption's bookkeeping
NotAlexNoyleandClaude Opus 5.52026-10-26 00:03:03 -04:00
eb9987b742
Engine 0064, 0065 (T-527): a layer paint with nothing pending skipped, a moved displayport base repainted; a failed band surface not retried every frame
NotAlexNoyleandClaude Opus 5.52026-10-26 00:02:53 -04:00
f43a016509
about:settings by finger (preferences-ui R6): every table row changed and put back by the emulated finger in Enyo, the Browser pane's controls and a pane reset
NotAlexNoyleandClaude Opus 5.52026-10-25 23:34:27 -04:00
fc2f9de3fc
Build site: T-504, T-505, T-515, T-516, T-517 blocked by T-527 (their open criteria are T-527's work since the owner re-opened them under it)
NotAlexNoyleandClaude Opus 5.52026-10-25 22:50:29 -04:00
a597e4a697
about:addons' long lists pan again (chrome-look R2): a XUL page's whole length reported and painted, and moved by the compositor's world offset on the GPU path
NotAlexNoyleandClaude Opus 5.52026-10-25 22:41:12 -04:00
0481ca9410
T-527: a flick no longer shows grey while the card coasts — the express lane composes each frame where the card will be when it is shown (its speed times 80 ms, JIHAD_GPU_EXPRESS_PREDICT_MS)
NotAlexNoyleandClaude Opus 5.52026-10-25 22:18:08 -04:00
867687f19e
T-527: records — the double tap's sharp frame broken down, the texture pool at 96 measured (kept at 48), desktop Wikipedia's pan mean with the sensors (23.8/24.4 fps) and its CPU split; pan-run's PAN_TOP reads the Enyo daemon's threads
NotAlexNoyleandClaude Opus 5.52026-10-25 22:12:11 -04:00
d5f50632e3
T-527 (R8) / T-511 DONE: the WebGL cube at 25-28 fps on the card — the card redraws only what changed, a tall band surface (engine 0063); the PDK adapter builds again
NotAlexNoyleandClaude Opus 5.52026-10-25 22:04:28 -04:00
911034810b
T-527 (R6): the zoom's sharp frame 3-4x sooner — the critical displayport, the express lane publishing the main lane's frame, engine patch 0062
NotAlexNoyleandClaude Opus 5.52026-10-25 21:14:05 -04:00
ef9691eca8
Cavekit 3.1 alignment: tools/host/cavekit-sync.py regenerates .cavekit/tasks.json and reconciles task-status.json from every tier table of the build site (the loop saw 171 of 327 tasks; now all, 304 complete / 16 blocked / 7 in progress); cavekit-overview.md's Status column counted from each kit's criteria with the Finger Emulation row added; plan-overview.md and impl-overview.md point at the current sources; the plugin registration refreshed (the 3.1 marketplace's ck, which reports 3.0.0 from its own manifest); the rotate item done by the card's own orientation request (the Enyo test hook's rotate scenario); PICKUP's device state after the fling-lead push
NotAlexNoyleandClaude Fable 5.12026-10-25 19:48:38 -04:00
c0aee8d02b
T-541 DONE: a fast flick on an image-heavy page no longer lands on grey — the GPU displayport at rest is a height each way (JIHAD_GPU_DP_Y 0.25 → 0.75; the A/B on desktop Wikipedia cost nothing: 80 vs 74 express frames over the same five flicks, the main lane 5-8 fps either way) and a fling adds a lead of up to a height in its direction from the card's scroll speed (BrowserPageGoanna's velocity → SetGpuLead → asymmetric margins); about:personas' flicks by finger: the grey lower third and the white upper third are gone, a sliver at the very bottom on the first flick after load
NotAlexNoyleandClaude Fable 5.12026-10-25 19:30:59 -04:00
5f042ccfe0
T-542: no defect — the file-installed Persona's fixture has no author and a flat grey picture; the card shows exactly that
NotAlexNoyleandClaude Fable 5.12026-10-25 19:16:18 -04:00
e2ff0d79bc
FINGER-TESTS: about:personas by finger — no card flashes at rest or on a slow scroll; a fast flick far down an image-heavy page leaves the uncovered rows grey for up to a second (T-541 opened); Remove right after installing a Persona keeps its question until answered; a Persona installed from a .xpi lacks its author and picture (T-542 opened)
NotAlexNoyleandClaude Fable 5.12026-10-25 19:14:23 -04:00
2abac1beb7
PICKUP: the device state after the proxy round's push (packages 16305f22 / 0db117c5 / 40f7526b, daemon 45fae433)
NotAlexNoyleandClaude Fable 5.12026-10-25 19:09:20 -04:00
88a3c2bc58
Proxies by finger in all three shells: My own proxy typed on the keyboard and used, a login-asking proxy (tools/host/auth-proxy.py) answered through each shell's dialog, the section's redraw, the Relay when chosen (https pass-through, the site's own issuer in the padlock) — and the owner's rule recorded: the Riverstone Relay is never a default, a fallback for webOS older than 3.0 only (the shipped defaults audited: direct connection, no automatic selection, the Relay's CA trust off); the settings page says so; a login dialog's reply deadline raised to 180 s
NotAlexNoyleandClaude Fable 5.12026-10-25 19:01:24 -04:00
0865747a4a
PICKUP: the device state at the end of the 2026-10-25 session (packages 075a9626 / 61d4b5f4 / dfb05e57, daemon 2de5bdc8), the next steps
NotAlexNoyleandClaude Fable 5.12026-10-25 15:14:13 -04:00
1d85d84d47
T-539/T-540 DONE: the slow-script dialog by finger — the card's confirm names the engine's buttons in its text (OK = Continue, Cancel = Stop script; Cancel stops the script and the card stays usable); a dialog whose card closes meanwhile no longer holds the daemon's main thread for the 60 s reply wait — the wait watches the card's socket (a hang-up ends it, measured 6 s after the close) and a failed send to the card (YapProxy remembers EPIPE; the sink says clientAlive/clientFd)
NotAlexNoyleandClaude Fable 5.12026-10-25 15:05:32 -04:00
1107473ffe
FINGER-TESTS: everyday browsing by finger in all three shells — DuckDuckGo by keyboard and Enter, back and forward, a result page in Enyo; the JS results page's render time and the bfcache-less back noted
NotAlexNoyleandClaude Fable 5.12026-10-25 14:46:34 -04:00
9bc3bd5d2d
PICKUP: the device state after the T-537/T-538 push (packages fb8bc730 / 9a3f8159 / dcda6c9b, daemon f7b5ffdb)
NotAlexNoyleandClaude Fable 5.12026-10-25 14:42:47 -04:00
697226111b
T-537/T-538 DONE: Enter in a textarea reaches the page first (a real keydown + keypress; duckduckgo.com's box now searches, a plain textarea still gets its line break); a field inside a frame by finger — the edit target follows a tapped frame into its document and a tap on the frame of the current target re-raises the keyboard; the followed layout width (T-499) survives the keyboard's height-only resize, so a fit-zoomed page neither re-flows nor re-zooms around the keyboard (it went 0.78 → 1.0 and moved the field 88 px); the everyday-browsing round in Enyo (DuckDuckGo by keyboard, a result, back, forward) recorded; jihad-frame-field.html added to the harness
NotAlexNoyleandClaude Fable 5.12026-10-25 14:34:13 -04:00
1a18b7492a
T-536 DONE: about:addons by finger — a closed popup's pixels stayed on the card (the GPU frame rebuilt on the slot that still carried the composited menu): the popup-close transition now voids every slot, the software path's and the GPU path's (gpuSlotsReset), verified gone after Check for Updates and after a dismiss; engine patch 0062 (a content invalidation on popup close) proved unnecessary and is dropped, the engine back at ec8f0ed7 (patches through 0061); the row after Undo relabelled Uninstall by a mutation observer; the tools menu styled readable; the three packages rebuilt and pushed — the device is the repository head
NotAlexNoyleandClaude Fable 5.12026-10-25 14:07:42 -04:00
16afea29fe
about:addons by finger (part): Uninstall → Undo and the update check act on one tap; three findings fixed in source, untested on the device at the cut-off — engine patch 0062 repaints the content when a composited popup closes (the tools menu's pixels stayed on the card after Check for Updates), the row after Undo relabelled Uninstall by a mutation observer, the tools menu styled readable (#utils-menu); PICKUP says exactly what is left
NotAlexNoyleandClaude Fable 5.12026-10-25 11:14:59 -04:00
1401096d35
Personas by finger after T-535: Wear, Take off, Undo, Remove (two taps), Install from file, a wear from the add-on site and the addons.mozilla.org question all act on one tap (personas-ui R6's finger criterion, FINGER-TESTS); the 2026-10-24 oddities were the double click and the worn Persona moving to the top of the list
NotAlexNoyleandClaude Fable 5.12026-10-25 11:05:12 -04:00
7825c9ba5d
T-535 DONE: one tap, one click — the adapter sent a clickAt from the gesture single-tap on top of the pen path's hit-test click (added 2026-07-07 when the daemon answered no hit test), so every tap was two clicks: a checkbox toggled twice, about:addons' Disable disabled and re-enabled, Preferences opened one run in five; the gesture's clickAt removed, the three adapters rebuilt and verified by finger in each shell (cavekit-input-bridging R1, FINGER-TESTS about:addons)
NotAlexNoyleandClaude Fable 5.12026-10-25 10:45:14 -04:00
e3ca9ab237
T-534 DONE: a certificate error is not TLS intolerance — engine patch 0061 (PSM's retryDueToTLSIntolerance remembered a certificate failure as version intolerance, retried at TLS 1.2 and tripped the TLS 1.3 server's downgrade sentinel, so a revoked certificate read 'malformed Server Hello'); the three variants pushed as packages built at this commit (libxul 161d7a88, daemon a26899d7, roots f8e9ebea) — the device is the repository head; verified on the packaged build: ssl.com's and DigiCert's revoked hosts say SEC_ERROR_REVOKED_CERTIFICATE; the overview row, the handoff's device state
NotAlexNoyleandClaude Fable 5.12026-10-25 10:23:06 -04:00
241988a8f0
T-529/T-530/T-531 DONE, T-532/T-533 (new) DONE: the certificate-trust round on the device — the ten new TLS roots' test sites load with the platform import off, the import now trusts 0 webOS roots (the five it added are built in), badssl's cases recorded, OCSP revocation enforced (CRL-only revocation invisible: no CRL fetch or CRLite in UXP, recorded); the TLS floor raised to 1.2 (the engine's default accepted TLS 1.0/1.1; the Settings row can lower it); the Trust dialog only for the engine's overridable errors (a revoked certificate got Trust Always and accepting reloaded into the same failure; now a failed load with the engine's alert); the daemon's start line counts the built-in roots on the removed report (0); update-engine.sh --pins checks UXP release's certdata; T-534 opened for NSS's TLS 1.3 label of a revoked refusal
NotAlexNoyleandClaude Fable 5.12026-10-25 10:04:52 -04:00
124bca3801
T-528/T-529 (part)/T-531 (part): the certificate-trust kit and Tier 90; the root audit (tools/host/audit-roots.py: the engine's roots vs UXP release and the CCADB removed report by SHA-256) found 24 removed roots shipped and 19 missing; the roots rolled to UXP release 037cc5f6 as engine patch 0060 with the pins and a check; the platform import's removed list rebuilt from today's report (106); docs/CERTIFICATE-TRUST.md
NotAlexNoyleandClaude Fable 5.12026-10-25 08:59:02 -04:00
30a94b6111
PICKUP: the commits to push named (Jihad-Browser de4fb43c..HEAD, the engine fork 7a872e8f..c79d16f8), the push script to use
NotAlexNoyleandClaude Fable 5.12026-10-24 23:52:44 -04:00
64413a7347
T-527 (part): the GPU floor bench without the sensor daemons — desktop Wikipedia 21-22 fps with sensord/atlas-sensord, 25-27.5 without (sensord is ~30% of a core at idle); the long page 30-35 / 33-37; the sensors restored (cavekit-gpu-rendering R3)
NotAlexNoyleandClaude Fable 5.12026-10-24 23:52:06 -04:00
17f05b7c50
T-526 DONE: Mochi's Paste from the field's hold sheet lands — the sheet had taken the plugin's focus and the engine's field blurred with it, so the insert had no target; pagePaste clicks the held point first (the hit test's document point kept from the page-hold) and sends the text after; the round's part 3 recorded (cavekit-mochi-ui R2, cavekit-finger-emulation R3)
NotAlexNoyleandClaude Fable 5.12026-10-24 23:45:12 -04:00
82e086d186
T-525: Mojo's long-press on a link or an image gets its own commands — the shell claims those holds through the widget assistant's eventFired (the framework widget had none, so the adapter's hit test never reached it) and offers Open in New Card / Share Link / Copy URL (cavekit-mojo-ui R6); Mochi logs its clipboard copy and paste sizes (the paste reads 5 chars and still inserts nothing — under investigation)
NotAlexNoyleandClaude Fable 5.12026-10-24 23:35:17 -04:00
414d284291
T-526 (part): the caret and the Backspace repeat, Copy from the address and Paste into a page field, a download of two types (the CE bench opens neither: no handler app); a generated one-line PDF as the download fixture
NotAlexNoyleandClaude Fable 5.12026-10-24 23:26:55 -04:00
24cf4210a5
T-525 DONE, T-526 (part): the round's parts 2 and 3 by the emulated finger — the fonts picked and restored, the add-on site's nav, search and Wear dialog on the TouchPad, the records in the kits, the site and the handoff; open: Mojo's link hold, the extension row's buttons, Mochi's Paste
NotAlexNoyleandClaude Fable 5.12026-10-24 23:19:34 -04:00
51d1e288cb
T-526 (part): part 3 by the emulated finger — fields take the keyboard on every page tried, keys reach about:addons' search and about:preferences, the hide key and the app menu over a focused address behave, a Persona worn and taken off, Palefill disabled and re-enabled (the row's buttons do not take a tap, the detail view's do), YouTube's mobile page plays on its play button, a tap pauses a video; Mochi's Paste left the field empty (open)
NotAlexNoyleandClaude Fable 5.12026-10-24 23:12:26 -04:00
bf59709845
T-525 (part): part 2 by the emulated finger — Mojo's back and forward act on their boxes (the earlier miss was a tap on the disabled forward box and a stale JS cache), Mojo's address selects on one tap and a typed search runs, its picker and text settings stick, the long-press menus of all three (Mojo's link hold offers the text commands only), the select popup, the certificate dialog, the preferences tabs, a Persona worn and taken off, the add-ons panes and plugin menu (the extension row's buttons open)
NotAlexNoyleandClaude Fable 5.12026-10-24 22:55:29 -04:00
02b9edfe7a
T-525 (part): the chrome buttons of the three shells with the emulated finger — all act but Mojo's back and forward, which emit no command (open); the daemon logs pageBackward/pageForward; Enyo's long-press on text and a field
NotAlexNoyleandClaude Fable 5.12026-10-24 19:37:38 -04:00
937419d1b8
T-523/T-524/T-514: the finger round's part 1 with the emulated finger — the owner's grey found and fixed: the smart zoom's block and the viewport's maximum, the root's displayport base and clip at a resolution (engine 0059), the GPU-path hit test, the resolution before the displayport; the results in the kits (cavekit-finger-emulation R2-R4, cavekit-gpu-rendering R6, R8, R12)
NotAlexNoyleandClaude Fable 5.12026-10-24 18:50:58 -04:00
ffaa022ba4
T-523 (part): an emulated finger on the real input path — the panel's datagrams to LunaSysMgr's socket; tap, drag, flick, pinch, long press, double tap and keyboard taps proven; the finger-emulation kit and Tier 89 (cavekit-finger-emulation R1, R2, R4)
NotAlexNoyleandClaude Fable 5.12026-10-24 17:13:25 -04:00
fb3ca2f6d5
Records: the 2026-10-24 push made (engine fork 706d4782..7a872e8f, main 62df604c..de4fb43c), the handoff says so
NotAlexNoyleandClaude Fable 5.12026-10-24 13:59:23 -04:00
de4fb43cd3
T-521: the commits made, unsigned by the owner's rule (the Gitea key is Gitea's only); the records say so and name the push (cavekit-gpu-rendering R11)
NotAlexNoyleandClaude Fable 5.12026-10-24 13:50:05 -04:00
318869f2a6
T-518/T-519/T-520/T-522: the kits' open criteria closed on the current build — the records, the GPU verify sweep and per-page table, the blocker's device criteria (cavekit-gpu-rendering R2-R4, R11; cavekit-content-blocking R3, R4, R6, R7; cavekit-modern-web R1, R6); the generic hiding sheet's 4 s cost found and fixed
NotAlexNoyleandClaude Fable 5.12026-10-24 13:49:05 -04:00
3f4e1a27e8
T-513/T-499: the review of Tier 87 fixed and verified, Codex's nine on top; the layout viewport per document (cavekit-gpu-rendering R2, R10; cavekit-modern-web R6)
NotAlexNoyleandClaude Opus 5.52026-10-24 13:49:04 -04:00
506c41144d
T-487: the blocklist updater tested on the device; EasyPrivacy's truncation guard and the skipped early look fixed (cavekit-content-blocking R5)
NotAlexNoyleandClaude Opus 5.52026-10-24 13:49:04 -04:00
ac4ea1c9a7
T-490: the blocklist count's independent tokenizer (inspector F-21), the review of Tier 86 done (cavekit-content-blocking R2, R7)
NotAlexNoyleandClaude Opus 5.52026-10-19 14:31:04 -04:00
61d25726f7
T-512: fit to keep, done — R11's diagnostics, patch queue and docs criteria met (cavekit-gpu-rendering R11)
NotAlexNoyleandClaude Opus 5.52026-10-19 14:24:12 -04:00
1c633a64e0
T-505/T-517/T-511 records: the evening's cuts measured, the band frame GPU-bound, the circuit breaker taken (cavekit-gpu-rendering R3, R8)
NotAlexNoyleandClaude Opus 5.52026-10-19 14:23:31 -04:00
e8b0e4250d
T-512 (part): the GPU path's diagnostics off by default behind JIHAD_GPU_LOG (cavekit-gpu-rendering R11)
NotAlexNoyleandClaude Opus 5.52026-10-19 14:23:23 -04:00
7a4873febb
T-517 (part): a band frame's regions in one composite, tiles outside the clip skipped early (cavekit-gpu-rendering R3)
NotAlexNoyleandClaude Opus 5.52026-10-19 12:46:20 -04:00
8c1544f855
README: "Where it stands", the browser's progress to release, counted from the kits
NotAlexNoyleandClaude Opus 5.52026-10-19 12:15:27 -04:00
33814005c5
T-504 records: double taps after the review fixes, the GPU-swizzle tries, the task's budget spent (cavekit-gpu-rendering R6)
NotAlexNoyleandClaude Opus 5.52026-10-19 11:20:52 -04:00
902187aebe
T-517/T-511 review fixes: the express lane's own-slot base guarded, slots invalid while written, WebGL off after a give-up (cavekit-gpu-rendering R3, R8, R11)
NotAlexNoyleandClaude Opus 5.52026-10-19 11:13:45 -04:00
20ccf6b553
T-512 (part): the README and the rendering options say what the GPU path does and measured; every JIHAD_GPU_* switch listed (cavekit-gpu-rendering R11)
NotAlexNoyleandClaude Opus 5.52026-10-19 10:20:46 -04:00
897e0c55ac
T-517 (round 333): a still-column rule for see-through sticky panels tried and reverted
NotAlexNoyleandClaude Opus 5.52026-10-19 10:16:46 -04:00
4b8a595b5a
T-517 (part): frame windows, a scrolled frame moves within its segment instead of its rows (cavekit-gpu-rendering R3)
NotAlexNoyleandClaude Opus 5.52026-10-19 09:57:37 -04:00
e0416cbc96
T-517 (part): no cache clean of the frame rows when the segments are mapped write-through (cavekit-gpu-rendering R3)
NotAlexNoyleandClaude Opus 5.52026-10-19 08:48:37 -04:00
NotAlexNoyleandClaude Fable 5.1
