luci-app-ddns: replace dynamic_dns_lucihelper call on handleStopDDnsRule

[feckert]

Description

@champtar This should fix https://github.com/openwrt/packages/security/advisories/GHSA-hfpc-hw28-69pj

However, this change only works if the procd change is used to start the service.
This change is included in master and openwrt-25.12

Also, I'm no longer satisfied with the app. The app needs a little love. Please try it again with a different setup.

Checklist

• [ x ] This PR is not from my main or master branch 💩, but a separate branch. ✅
• [ x ] Each commit has a valid ✒️ Signed-off-by: <my@email.address> row (via git commit --signoff).
• [ x ] Each commit and PR title has a valid 📝 <package name>: title first line subject for packages.

[champtar]

Thanks @feckert, on my phone so hard to review your change, I've pinged the initial reporter for review.

Also, I'm no longer satisfied with the app. The app needs a little love. Please try it again with a different setup.

You mean you want to stop maintaining it ? Or you have improvement ideas but no time ?

[champtar]

CI is not happy with the empty commit body

[feckert]

CI is not happy with the empty commit body

That should work now.

You mean you want to stop maintaining it ? Or you have improvement ideas but no time ?

Well, I didn't create the app. I just took it over, because I'm also the maintainer of the ddns-script package (which I'm not happy about either), but no one else is doing it.
The problem is simply time

[openwrt-ai]

Reviewed 1 new commit. The core change (routing stop/reload through /etc/init.d/ddns per-section instead of dynamic_dns_lucihelper.sh, with the matching ACL update) is sound — the ddns init script uses USE_PROCD with per-section procd_open_instance "$cfg", so /etc/init.d/ddns stop <section> targets a single instance as intended. Two notes left inline on the reload.disabled change, which is functionally correct but diverges from the static render path and is outside the commit's stated scope.

---

Generated by Claude Code

[feckert]

@champtar I rebased the button handling again based on a note from the AI CI/CD on the overview page. It actually wasn't working right. The whole thing was a bit tricky. Like I said, the app needs some love. Let's see if I can find the time.

[openwrt-ai]

Reviewed 1 new commit (bf6c5d6 "fix start/stop button logic"; cfedd55 is content-identical to the previously-reviewed revision). One inline finding on the reworked reload-button gating, which throws a TypeError for enabled-but-not-running sections in both the poll and static render paths.

---

Generated by Claude Code

[openwrt-ai]

(!service[section_id] && !service[section_id].pid) is broken in both directions. When service[section_id] is undefined — which happens for an enabled section whose service isn't currently running (the surrounding code reads service[section_id]?.ip with optional chaining, so undefined is expected here) — !service[section_id] is true and the && then evaluates !service[section_id].pid, throwing TypeError: Cannot read properties of undefined (reading 'pid') and aborting the poll refresh. Because such a section has cfg_enabled != 0, the || doesn't short-circuit, so the crash is reachable. When service[section_id] is truthy the term short-circuits to false, so the pid check is dead and reload ends up gated only on cfg_enabled.

The stop.disabled line directly below`` already uses the correct guard — mirror it:

Suggested change

	reload.disabled = (cfg_enabled == 0) || (!service[section_id] && !service[section_id].pid);
	reload.disabled = (cfg_enabled == 0) || !(service[section_id] && service[section_id].pid);

The same bug is in the static render path at line 558:`` (!resolved[section_id] && !resolved[section_id].pid) should likewise become `!(resolved[section_id] && resolved[section_id].pid)`.

---

Generated by Claude Code