mirror of
https://github.com/kiddin9/op-packages.git
synced 2026-11-02 15:03:56 +08:00
This commit is contained in:
github-actions[bot] committed
1 parent
c7bafc635d
commit
c71da84035
264 files changed
+8169
-3944
No files matched your search
@@ -3,5 +3,5 @@
|
||||
.DS_Store
|
||||
node_modules/
|
||||
|
||||
# Regents review verdicts. Written into the repo being reviewed; never committed.
|
||||
# Local review output; never committed.
|
||||
.regents/
|
||||
@@ -6,7 +6,7 @@ include $(TOPDIR)/rules.mk
|
||||
|
||||
PKG_NAME:=luci-app-appflow
|
||||
PKG_VERSION:=1.1.2
|
||||
PKG_RELEASE:=1
|
||||
PKG_RELEASE:=2
|
||||
PKG_LICENSE:=Apache-2.0
|
||||
PKG_LICENSE_FILES:=LICENSE
|
||||
PKG_MAINTAINER:=Bob <git16@bob7.com>
|
||||
|
||||
@@ -18,7 +18,7 @@ One file, every device. The package is `noarch` — only JavaScript and ucode
|
||||
inside — and the same `.apk` is verified installing on both aarch64 and
|
||||
arm_cortex-a15 hardware.
|
||||
|
||||
wget https://github.com/VolanticSystems/luci-app-appflow/releases/download/v1.1.2/luci-app-appflow-1.1.2-r1.apk
|
||||
wget -O luci-app-appflow-1.1.2-r1.apk https://github.com/VolanticSystems/luci-app-appflow/releases/download/v1.1.2/luci-app-appflow-1.1.2-r1.apk
|
||||
apk add --allow-untrusted ./luci-app-appflow-1.1.2-r1.apk
|
||||
/etc/init.d/appflowd enable && /etc/init.d/appflowd start
|
||||
|
||||
|
||||
@@ -135,11 +135,11 @@ participate and throttling is impossible by construction.
|
||||
accountable byte counts for sub-interval flows.
|
||||
|
||||
**CORRECTED 2026-08-25. The two paragraphs below understated this, and the
|
||||
review panel was right.** Once `purge_only_rescued` was instrumented and left
|
||||
reviewers were right.** Once `purge_only_rescued` was instrumented and left
|
||||
running against ordinary traffic, it reported **13 purge-only flows out of
|
||||
1,341 seen, 0.97%**. Flows purged without ever receiving a `flow_stats` DO
|
||||
occur on netifyd 4.4.7. Before the defensive path was added, every one of them
|
||||
lost its entire byte count silently, which is exactly what three carriers
|
||||
lost its entire byte count silently, which is exactly what three reviewers
|
||||
predicted and what this section briefly called refuted.
|
||||
|
||||
The refutation was under-powered, not wrong-headed: a 38-flow capture at a
|
||||
@@ -149,7 +149,7 @@ participate and throttling is impossible by construction.
|
||||
rate it claims to exclude. The event-level evidence below is still accurate
|
||||
about what it saw; it simply did not see enough.
|
||||
|
||||
**Re-tested at the EVENT level 2026-08-24, after an adversarial review panel
|
||||
**Re-tested at the EVENT level 2026-08-24, after an adversarial review
|
||||
independently reached the same prediction three times.** The 2026-08-22 result
|
||||
above is outcome-based: it shows the bytes landed, not why. Tapping netifyd's
|
||||
export socket directly for 170 s (`flow`/`flow_stats`/`flow_purge` per digest,
|
||||
@@ -355,7 +355,7 @@ purge:
|
||||
|
||||
> **The rest of this subsection is HISTORICAL and describes the state before
|
||||
> the fix above. It is kept because the reasoning is what led to the fix, but
|
||||
> nothing below is an open problem.** A review panel reading this document cold
|
||||
> nothing below is an open problem.** A reviewer reading this document cold
|
||||
> reported it as a live double-count, having entered mid-section, so if you are
|
||||
> skimming, stop here and go to §3.3.
|
||||
|
||||
@@ -582,7 +582,7 @@ Depends (from `Makefile` `LUCI_DEPENDS`): `netifyd`, `luci-base`,
|
||||
`conntrack.at`, i.e. on having read recently rather than on having read
|
||||
recently *and* found something.
|
||||
- **Byte conservation is now instrumented, not just asserted** (2026-08-24,
|
||||
suggested by a panel juror). `status.bytes` reports `reported` (what netifyd
|
||||
suggested by a reviewer). `status.bytes` reports `reported` (what netifyd
|
||||
said moved, before any clamp or shadow decision), `attributed` (what reached
|
||||
an aggregate), `shadowed` (what was deliberately dropped as a NAT twin) and
|
||||
`leaked` (the residual, which should be 0). Both silent byte-loss defects this
|
||||
@@ -602,7 +602,7 @@ Depends (from `Makefile` `LUCI_DEPENDS`): `netifyd`, `luci-base`,
|
||||
fixed without an established-flow dump upstream. Documented in the
|
||||
README rather than hidden.
|
||||
- ~~**reshadow() exempted the flows it existed to correct**~~, FOUND by an
|
||||
adversarial panel and FIXED 2026-08-24. `reshadow()` returned early on
|
||||
adversarial review and FIXED 2026-08-24. `reshadow()` returned early on
|
||||
`dev_key == "router"`, which is exactly how a WAN-side capture of a NAT-ed
|
||||
client flow is classified, because netifyd reports the post-translation
|
||||
identity (§3.2). The conntrack twin resolution added earlier that day went
|
||||
|
||||
@@ -232,7 +232,7 @@ var DEVICE_LABELS = {
|
||||
* normalised, and columnChart reads only p.dl/p.ul, so it drew an EMPTY framed
|
||||
* plot under a KPI row reporting megabytes. Worse, series.length was still 12,
|
||||
* so the honest "No time series available" branch never ran.
|
||||
* Found by a code-review panel, 2026-08-25. No daemon field is named dl or ul,
|
||||
* Found in code review, 2026-08-25. No daemon field is named dl or ul,
|
||||
* so the aliases are inert against real payloads. */
|
||||
var A_DL = [ 'download', 'bytes_down', 'rx_bytes', 'download_bytes', 'total_download', 'dl' ],
|
||||
A_UL = [ 'upload', 'bytes_up', 'tx_bytes', 'upload_bytes', 'total_upload', 'ul' ],
|
||||
@@ -485,7 +485,7 @@ return baseclass.extend({
|
||||
* bytes_down fell through to the byte counter and the Download KPI read
|
||||
* something like "12.4 GB/s" while nothing moved. Read with a -1 sentinel
|
||||
* and test presence instead, the way normTotals already does for its own
|
||||
* totals. Raised by a code-review panel, 2026-08-25.
|
||||
* totals. Raised in code review, 2026-08-25.
|
||||
*
|
||||
* Latent rather than live: the only caller is guarded by `summary.rates ?`,
|
||||
* and this daemon puts byte and rate counters together in `totals` and sends
|
||||
@@ -752,7 +752,7 @@ return baseclass.extend({
|
||||
* A device named `constructor` then rendered style="background:undefined",
|
||||
* which the browser drops, leaving white tile text on a transparent
|
||||
* tile; in the doughnut the slice silently did not draw while its legend
|
||||
* still claimed a percentage. Found by a code-review panel, 2026-08-25. */
|
||||
* still claimed a percentage. Found in code review, 2026-08-25. */
|
||||
if (Object.prototype.hasOwnProperty.call(CATEGORY_SLOT, s))
|
||||
return PALETTE[CATEGORY_SLOT[s]];
|
||||
|
||||
@@ -1257,7 +1257,7 @@ return baseclass.extend({
|
||||
* `.af-donut > svg` also carried transform:rotate(-90deg) until
|
||||
* 2026-08-25, the two stacked, and the first slice actually began at
|
||||
* 9 o'clock -- so this comment asserted something the render did not
|
||||
* do. Found by a code-review panel. If you re-add a CSS rotation,
|
||||
* do. Found in code review. If you re-add a CSS rotation,
|
||||
* take this offset out. */
|
||||
off = 25,
|
||||
ring = [ this.svg('circle', {
|
||||
@@ -1418,7 +1418,7 @@ return baseclass.extend({
|
||||
* It failed GREEN on the one indicator whose whole job is to tell the
|
||||
* operator the collector is down, which is also the most common reason
|
||||
* the page is empty. Absent evidence is not positive evidence; unknown
|
||||
* is its own state and is drawn neutral. Found by a code-review panel,
|
||||
* is its own state and is drawn neutral. Found in code review,
|
||||
* 2026-08-25. */
|
||||
known = (s.agent_connected != null) || (sock.connected != null),
|
||||
connected = known && (s.agent_connected !== false) && (sock.connected !== false),
|
||||
|
||||
@@ -300,7 +300,7 @@ return view.extend({
|
||||
* ordinary config change, so that was a hair trigger. paint() now drops
|
||||
* the buffer only when the gap exceeds the chart's own window, which is
|
||||
* the case where redrawing it really would mislead. Raised by a
|
||||
* code-review panel, 2026-08-25. */
|
||||
* code review, 2026-08-25. */
|
||||
dom.content(this.downNode, appflow.notRunning(appflow.classifyFail(err), err));
|
||||
this.liveNode.style.display = 'none';
|
||||
this.downNode.style.display = '';
|
||||
@@ -551,7 +551,7 @@ return view.extend({
|
||||
* caption reading "top, by cumulative bytes". The fallback
|
||||
* looked like belt-and-braces and was actually a silent unit
|
||||
* mix; a row with no bytes is filtered out below, which is
|
||||
* the honest outcome. Found by a code-review panel. */
|
||||
* the honest outcome. Found in code review. */
|
||||
value: t.total
|
||||
};
|
||||
})
|
||||
|
||||
@@ -333,7 +333,7 @@ return view.extend({
|
||||
* handler anywhere in this package. role=button plus tabindex puts
|
||||
* the row in the tab order and makes it announce as activatable;
|
||||
* Enter and Space are the two keys a button is required to answer.
|
||||
* Raised by a code-review panel. */
|
||||
* Raised in code review. */
|
||||
rows.push(E('tr', {
|
||||
'class': 'tr af-row',
|
||||
'title': _('Show per-device details'),
|
||||
@@ -530,7 +530,7 @@ return view.extend({
|
||||
* active catalog says. With the English catalog that is a constant in
|
||||
* this file; the moment a translation is accepted it is third-party
|
||||
* content, which is exactly what the pending zh_Hans PR proposes. Raised
|
||||
* by a review panel over that PR, which was careful to note the submitted
|
||||
* by a review of that PR, which was careful to note the submitted
|
||||
* Chinese contains no markup: this is closing the trust distinction, not
|
||||
* fixing an exploit. */
|
||||
ui.showModal([ _('Details') ], body, 'cbi-modal');
|
||||
|
||||
@@ -5,7 +5,7 @@
|
||||
// appflow browser-half suite.
|
||||
//
|
||||
// WHY THIS FILE EXISTS. Every defect this package's frontend has had was found
|
||||
// by a human or a review panel READING the code. Not one was found by a test,
|
||||
// by a reviewer READING the code. Not one was found by a test,
|
||||
// because until now no test could run any of it. The daemon has 54 checks
|
||||
// against it and the JavaScript had none, which is backwards: the daemon
|
||||
// mostly counts bytes, while the frontend is what decides whether an operator
|
||||
|
||||
@@ -155,7 +155,7 @@ test_conservation() {
|
||||
# A 38-flow event-socket capture had seen none, which at that rate has an
|
||||
# expected count of 0.4 -- so the sample was far too small to support the
|
||||
# "netifyd never does this" conclusion that was briefly drawn from it, and
|
||||
# the review panel that reported the byte loss was right.
|
||||
# the review that reported the byte loss was right.
|
||||
#
|
||||
# What is worth watching is the RATE. A sudden jump means netifyd's event
|
||||
# contract has shifted; zero across a busy run means this path stopped
|
||||
|
||||
@@ -5,7 +5,7 @@
|
||||
// functions can be tested without a browser.
|
||||
//
|
||||
// WHY THIS EXISTS. Every browser-side defect this package has had was found by
|
||||
// a human or a review panel reading the code. Not one was found by a test,
|
||||
// a reviewer reading the code. Not one was found by a test,
|
||||
// because no test could run any of it: LuCI view files are not CommonJS or ES
|
||||
// modules, they are evaluated by LuCI's own loader with the `'require x';`
|
||||
// directives intercepted. So the JavaScript half of a package whose entire
|
||||
@@ -73,7 +73,7 @@ function stripRequires(src) {
|
||||
// run of a suite that asserted exactly those outputs. Loading the module a
|
||||
// second time with a marking translator makes the two observationally
|
||||
// different, and equality assertions become sound again rather than
|
||||
// coincidentally true. Raised by a review panel, 2026-08-30.
|
||||
// coincidentally true. Raised in review, 2026-08-30.
|
||||
function load(file, stubs, opts) {
|
||||
const translate = (opts && opts.translate) || ((s) => s);
|
||||
const src = fs.readFileSync(file, 'utf8');
|
||||
|
||||
@@ -794,7 +794,7 @@ test_hostile() {
|
||||
feed_and_settle || return
|
||||
|
||||
# THIS RUNTIME CHECK BINDS NOTHING TODAY, AND THE STATIC ONE BELOW IS THE
|
||||
# REAL GUARD. A review panel converged on this and it was right.
|
||||
# REAL GUARD. Reviewers converged on this and they were right.
|
||||
#
|
||||
# The canary cannot be turned red by any edit that leaves appflowd doing
|
||||
# what it does, because appflowd never invokes a shell. The only edits that
|
||||
|
||||
Reference in new issue
Block a user