network applet: six correctness fixes, incl. multiple active WireGuard connections

[IvanTheGeek]

Six independent correctness fixes in network@cinnamon.org, one per commit so they can be taken or dropped individually. No new UI.

The bugs

1. updateAccessPoints() never loops — for (let i = 0; i < accessPoints; i++) compares the index to the array rather than its length, so the body never runs. _accessPoints is left empty and the notify::strength handlers are never reconnected, silently stopping signal-strength updates for that network.

2. splice(pos) truncates the connection list — Array.splice(pos) with no delete count removes everything from pos to the end, so removing one connection dropped every connection after it from _connections.

3. The VPN-over-wireless icon never resolves — set_applet_icon_symbolic_name() passes the name through unchanged and sets St.IconType.SYMBOLIC; St appends -symbolic itself at lookup time. The name built here already carried the suffix, so it asked for xsi-network-wireless-signal-<level>-secure-symbolic-symbolic, which does not exist. The panel icon renders as nothing while a VPN is active over wireless. Verified against the shipped icon set:

FOUND    xsi-network-wireless-signal-good-secure-symbolic
MISSING  xsi-network-wireless-signal-good-secure-symbolic-symbolic

The same commit guards active_access_point, which is null on a wireless device that is still activating.

4. _correctStateForTunnel() returns undefined — it falls through without a return when no tun/wireguard connection is active. Only reached from the currently-disabled connectivity check, but wrong either way.

5. An already-connected VPN draws no dot when its row is first built — _updateConnectionItemView() takes active as its third argument, but the first-build call omits it, so active is undefined. The row corrects itself on the next redraw, which is why it reads as intermittent. This is likely what #12550 is seeing.

6. Only one WireGuard connection can be active in the UI (#12178) — NMDeviceWIREGUARD kept a single _activeConnection slot, but every WireGuard connection is handed the same pseudo-device and _syncActiveConnections() calls setActiveConnection() once per connection, so each activation overwrites the previous one. With two tunnels up the menu shows one as connected and draws the other as if it were down; the section switch tears down only one and then flips straight back on.

This last one mirrors what #12930 did for NMDeviceVPN — keep a list, assign it once after the sync loop rather than per connection, and deactivate all of them. The VPN section got that treatment; the WireGuard section was left behind.

Rows become a switch each rather than a pick-one list, since WireGuard connections are independent of one another. That is the behaviour asked for in #12178 ("checkboxes instead of the radio items to toggle the Networks independently") and re-confirmed there in February.

Testing

Honest about what was and wasn't done:

• Built and node --checked against master.
• Smoke-tested by running this branch's applet.js on Cinnamon 6.6.9, which differs from master only by the four lines of network: Remove stale wireguard connections #13847. With two WireGuard tunnels up simultaneously the applet reported both, each with its own working switch:

  wg active list: [wg-devvps, wg-lab-inet]
  rows: wg-devvps=ON  wg-lab-inet=ON  wg-lab=off  wg-home=off
  

• Not run on master itself — I don't have a master build here.
• Bugs 1-5 are reasoned from the code and verified by inspection; 1, 2 and 4 are single-expression fixes.

Closes #12178

---

Disclosure: I worked on this with Claude (Anthropic's Claude Code). It did the bulk of the code reading, diagnosis and patch authoring, and drafted this write-up; I reviewed the changes and ran them on my own machine before filing. The commits carry a Co-Authored-By: Claude trailer to the same effect. Flagging it so you can weight the review accordingly.

[github-actions]

Best-practices scanner

This is a regex-based check for API usage that can pose security, performance or
maintainability issues, or that may already be provided by Cinnamon. Most findings
are advisory and do not automatically disqualify a pull request.

This check is not perfect and will not replace a normal review.

---

Found 2 potential issue(s):

⚠️ WARNING

⚠️ lang_bind

files/usr/share/cinnamon/applets/network@cinnamon.org/applet.js:1138

obj.item.connect('toggled', Lang.bind(this, function(item, state) {

Lang.bind() is deprecated. Use arrow functions (() => {}) or Function.prototype.bind() instead.

files/usr/share/cinnamon/applets/network@cinnamon.org/applet.js:1948

this._devices.wireguard.device.connect('active-connections-changed', Lang.bind(this, function() {

Lang.bind() is deprecated. Use arrow functions (() => {}) or Function.prototype.bind() instead.

---

Automated pattern check.

[IvanTheGeek]

The best-practices scanner findings above are addressed — all of the flagged Lang.bind() uses were in code this PR adds, and they are now arrow functions (which capture this the same way). No new Lang.bind() is introduced; the count in this file actually goes down versus master.

The scanner comment is stale rather than wrong: the workflow re-run is sitting in action_required, since runs from a first-time contributor need maintainer approval. It should clear once someone approves the run.