openthread-br: drop luci-app-openthread, proposed for luci repo

[LorbusChris]

📦 Package Details

Maintainer: @stintel

Should land after #30091

Description

Removes the luci-app-openthread subpackage from net/openthread-br. The LuCI
web interface is proposed for the openwrt/luci repository as
luci-app-openthread (openwrt/luci#8871), following the standard convention for
LuCI apps, as requested in #29791 and openthread/ot-br-posix#3431. Dropping it
here means the UI is no longer shipped from two places.

Split out of #30091 at @stintel's request so the removal can be reviewed on its
own. The branch sits on top of #30091: the first three commits here are #30091's
(the 2026.08.0 update, the REST listen-address option, the internal mDNS
switch), and only the last commit is what this PR proposes. It is meant to land
with or after #30091 — merging this merges that too.

PKG_RELEASE is bumped to 2 on top of #30091's version reset, since the
package set changes without a version change.

Note the app being removed is the old Lua/luci-lua-runtime implementation; the
openwrt/luci proposal is a rewrite using the modern client-side JS + rpcd ucode
conventions, so this is not a straight move of the same files.

openwrt/luci#8871 is still open, so this should ideally land together with (or
after) it, otherwise there is a window with no OpenThread web UI available at
all. The in-tree copy in ot-br-posix is proposed for removal separately in
openthread/ot-br-posix#3326 (draft).

🧪 Run Testing Details

The removal commit itself changes no code in the openthread-br package — the
remaining Package/openthread-br recipe is untouched. The src/openwrt/ tree
in ot-br-posix stays in use: src/openwrt/ubus/ builds the otbr ubus object
that the netifd proto handler waits on (ubus -t30 wait_for otbr, then
ubus call otbr threadstart), so -DOTBR_OPENWRT=ON is untouched here and must
stay.

The luci-app-openthread arms are dropped from test.sh and test-version.sh
(both added in #30091 underneath): those scripts refuse to pass a package they
don't know, so a case for one that is no longer built would defeat that.

✅ Formalities

• I have reviewed the CONTRIBUTING.md file for detailed contributing guidelines.

[openwrt-ai]

Reviewed 2 new commits; both commit messages match their diffs. Three inline notes, none blocking.

---

Generated by Claude Code

[openwrt-ai]

Reviewed 2 new commits; no new issues found.

The only new content since dad9226 is the README.md "### LuCI" hunk, which resolves the earlier nit — test.sh and the Makefile are byte-identical to what was already reviewed. Both commit messages still match their diffs, and the three FormalityCheck runs on 300b53e are green.

---

Generated by Claude Code

[openwrt-ai]

Reviewed 1 new commit (300b53e → fabe4a6, the test.sh commit amended to also add test-version.sh). The amended commit message matches its diff.

The only new content since the last review is net/openthread-br/test-version.sh; test.sh, the Makefile and README.md are unchanged from what was already reviewed. One inline note on the new file. All three FormalityCheck runs on fabe4a6 are green.

---

Generated by Claude Code

[openwrt-ai]

Reviewed 1 new commit (fabe4a6 → 03936f7, the test commit amended again); no new issues found.

The only change since the last review is net/openthread-br/test-version.sh switching case "$1" / '$1' to $PKG_NAME, which resolves the earlier note and now matches the documented contract in CONTRIBUTING.md:264-266 ("The script receives PKG_NAME and PKG_VERSION as environment variables") and the worked example below it. test.sh correctly keeps "$1", since that script is documented as receiving positional arguments. test.sh, the Makefile and README.md are byte-identical to what was already reviewed, and the amended commit message still matches its diff.

All three FormalityCheck runs on 03936f7 are green.

---

Generated by Claude Code

[Copilot]

Pull request overview

Removes the luci-app-openthread subpackage from net/openthread-br so the LuCI UI is no longer shipped from this feed and can instead live exclusively in the openwrt/luci repository, avoiding duplication and aligning with LuCI app conventions.

Changes:

• Drop the luci-app-openthread package definition and install recipe from net/openthread-br/Makefile and bump PKG_RELEASE.
• Add test.sh and test-version.sh to replace the generic version probe with functional smoke tests suitable for this package.
• Update README.md to reflect that the LuCI UI is no longer shipped from this package.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
net/openthread-br/Makefile	Removes the LuCI subpackage definition and bumps PKG_RELEASE due to content change.
net/openthread-br/README.md	Updates documentation to point users to the LuCI-provided UI package.
net/openthread-br/test.sh	Adds a functional smoke test exercising otbr-agent and ot-ctl.
net/openthread-br/test-version.sh	Overrides generic version detection which cannot work for this package.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[openwrt-ai]

Reviewed the force-push since the last review (03936f7 → b7e4ecc). The branch was rewritten rather than extended: the two previously reviewed commits are gone, and the head now carries #30091's three commits plus a README-only commit.

Commit checks

• b7e4ecc "openthread-br: drop luci-app-openthread, proposed for luci repo" — the message says "Remove the luci-app-openthread subpackage (and its install recipe)", but the commit only touches README.md. The Makefile hunk that the earlier version of this commit (d9419b7, 27 lines removed) carried is no longer present, so the PR does not currently do what its title, body and commit message say. Details inline.

The other three commits (55c50a7, 39d8999, c4254e9) are the ones from #30091 and their messages do match their diffs. c4254e9 is #30091's current head, so the two PRs now hold the same three commits — worth deciding whether #30096 should still be the independent LuCI-only split described in its body, or whether the two have merged back together.

Non-blocking notes: v2026.07.0 exists upstream (ec16e39), the PKG_VERSION / PKG_RELEASE:=1 pairing is correct for a version bump, and all three FormalityCheck runs on b7e4ecc are green.

---

Generated by Claude Code

[openwrt-ai]

Reviewed the force-push since the last review (b7e4ecc → 268dca4). The previous blocker is resolved: the head commit now carries the Makefile hunks again — define Package/luci-app-openthread, its install recipe and the BuildPackage call are all gone — and the now-dead luci-app-openthread arms are dropped from test.sh and test-version.sh, so the README and the build agree again. Nothing else in the tree references the subpackage. The other three commits (55c50a7, 39d8999, c4254e9) are unchanged from what was already reviewed.

Commit checks

• 268dca4 "openthread-br: drop luci-app-openthread, proposed for luci repo" — the message says "Split out of #30091 so it can be reviewed and land independently of the 2026.07.0 update", but the commit's parent is c4254e9, which is #30091's current head. The branch still carries all three of #30091's commits below this one, so as it stands this cannot land independently of the 2026.07.0 update — merging #30096 merges #30091 with it. Same mismatch in the PR body ("#30091 now contains the version update … and will be rebased on this"), which describes the opposite stacking from the one that exists. This was raised on the previous force-push and the Makefile half of it was fixed; the topology half is unchanged, so either the branch wants rebasing onto master or the body and commit message want updating to say #30096 sits on top of #30091.

  Secondary, non-blocking: the message describes only the Makefile removal, while the diff also rewrites README's ### LuCI section and edits both test scripts. Both are in scope for the change, just undocumented.

Non-blocking note inline on PKG_RELEASE. All three FormalityCheck runs on 268dca4 are green. For the record, luci-app-openthread is not yet present in openwrt/luci master (8b5f528), which is what the README's "a LuCI feed recent enough to carry it" caveat already covers.

---

Generated by Claude Code

[openwrt]

Formality Check: Suggestions Available

We checked this pull request against the contribution guidelines. Here is what needs your attention:

⚠️ STYLISTIC WARNINGS & SUGGESTIONS

Package Release Audit:

• ⚠️ Package net/openthread-br version updated from '2f3c799c7463c8f674754e65c53f78bc0bbcbd58' to '2026.08.0', but PKG_RELEASE was not reset to 1 (currently: '2')

---

_{Something broken? Consider reporting an issue.
Running version 0e15d1c deployed on 2026-08-12 22:36:32 CEST}

[openwrt-ai]

Reviewed the force-push since the last review (268dca4 → 27ab7f9). The branch was rewritten rather than extended: #30091's three commits were replaced by their 2026.08.0 versions (df88201, 6875a57, 6e05e4c), and the drop commit was reapplied as 27ab7f9. Those three are byte-identical to #30091's current head, so the PR body's description of the stacking now matches the branch — the earlier topology mismatch is resolved.

All four commit messages match their diffs, so no ## Commit checks section this time.

Grounding for the new content, checked against v2026.08.0 (3377117) and the openthread submodule it pins (c7a3a19):

• The tag exists and is the newest on openthread/ot-br-posix (git ls-remote --tags).
• -DOTBR_VERSION=$(PKG_VERSION) works as the commit message describes: CMakeLists.txt:72-95 only synthesises a version when OTBR_VERSION is unset, and it feeds OTBR_PACKAGE_VERSION, which is what realmain.cpp:192-194 prints for --version (before the vendor/product check, so it exits 0).
• Dropping test-version.sh is safe. The generic check only hard-fails when every executable misses the version — test_entrypoint.sh:245-256 fails on exec_checked -eq version_missing and only warns otherwise. otbr-agent now reports 2026.08.0, so ot-ctl not reporting it downgrades to a warning. The case "$PKG_NAME" in test.sh is fine too: the harness exports PKG_NAME/PKG_VERSION and passes them positionally (L327-L360).
• ot-ctl -h is still the right probe: client.cpp:211-215 has no --version entry, and the terminating {nullptr, 0, nullptr, 0} is present, so the old segfault is indeed gone.
• The CMAKE_OPTIONS rework checks out: options.cmake:104-121 is the Only one Advertising Proxy can be enabled. FATAL_ERROR, OTBR_DUA_ROUTING no longer exists, and OTBR_MDNS=openthread is a valid value that forces OT_MDNS/OT_SRP_ADV_PROXY on. OTBR_BORDER_AGENT_MESHCOP_SERVICE defaults OFF in that mode, so the OTBR DNSSD support must not be enabled with OT mDNS. guard is not tripped, and the MeshCoP instance name survives: OTBR hands its OTBR_VENDOR_NAME " " OTBR_PRODUCT_NAME base name to OT core via application.cpp:311, rather than falling back to OT's "OpenThread BR (unspecified vendor)" default.
• Both dropped patches are genuinely upstream — third_party/CMakeLists.txt now carries the pkg_check_modules(CJSON libcjson) logic verbatim — and the remaining 100-fix-gcc14-build.patch still matches its context in the mbedtls revision this release pins.
• Nothing in the tree references luci-app-openthread any more except the README paragraph that points at openwrt/luci.

Two inline notes, neither blocking. The first retracts an earlier openwrt-ai comment of mine on PKG_RELEASE that no longer holds now that the branch stayed stacked on #30091.

---

Generated by Claude Code

[openwrt-ai]

Reviewed the force-push since the last review (27ab7f9 → 31e07e1); no new issues found.

The branch was rewritten but the only new content is in net/openthread-br/README.md (+17/-4). Per-commit, the patches are unchanged apart from these two hunks and rebase line offsets:

• 07cd2dc "update to 2026.08.0" — gains the new ### Version string section.
• 2586fcc "make the REST API listen address configurable" — rewrites the required/optional paragraph under "Create network".
• 2c803bc "use OpenThread's internal mDNS" and 31e07e1 "drop luci-app-openthread" — byte-identical to 6e05e4c / 27ab7f9.

All four commit messages still match their diffs, so no ## Commit checks section. 2c803bc is also #30091's current head, so the two PRs still hold the same three commits below this one, matching the stacking the body describes.

Grounding for the two new paragraphs:

• "Only backbone_network, device and radio_url are required" is accurate and fixes the previous wording: the proto handler hard-fails on exactly those three plus the resolved backbone ifname (openthread-proto.sh:55-60), and dataset is applied only when set (L81-L83). "or if backbone_network names an interface that has no device" is the MISSING_BACKBONE_IFNAME path. This resolves my earlier nit on that line.
• "the repacked source tree has no git directory for git describe to read" holds for this package: PKG_SOURCE_URL is a github.com URL, so dl_method_git selects github_archive, and both that path and its rawgit fallback end with rm -rf .git .gitmodules before packing (download.mk:232-234).
• The [REST Server](#rest-server) link still resolves to the ### REST Server heading, and nothing outside the README references luci-app-openthread any more.

All three FormalityCheck runs on 31e07e1 are green.

---

Generated by Claude Code

[openwrt-ai]

Correction to the download.mk citation in my review just above: the commit SHA in that permalink is wrong. The correct link is include/download.mk:233-234 (openwrt/openwrt at 83e93e0) — the rm -rf .git .gitmodules in DownloadMethod/rawgit. The point stands unchanged; only the link was broken.

---

Generated by Claude Code

[openwrt-ai]

Reviewed the force-push since the last review (31e07e1 → f8a2e50); no new issues found.

The branch was rewritten but the tree delta is four lines in net/openthread-br/README.md and nothing else (git diff 31e07e1..f8a2e50 --stat). Per-commit:

• 07cd2dc "update to 2026.08.0" — unchanged, same SHA as last review.
• caacf17 (was 2586fcc) "make the REST API listen address configurable" — carries the whole delta: the two -- in the required/optional paragraph become em dashes and the "check ifstatus for the OpenThread interface" line is rewrapped. This matches the em dashes the README already uses at README.md:90 and :178, so it makes the file internally consistent. No wording changed, so the accuracy of that paragraph (verified against the proto handler last time) is unaffected.
• 233cfaa (was 2c803bc) "use OpenThread's internal mDNS" and f8a2e50 (was 31e07e1) "drop luci-app-openthread" — content-identical, only rebase blob hashes differ.

All four commit messages are byte-identical to the ones already reviewed and still match their diffs, so no ## Commit checks section.

233cfaa is also #30091's current head, so the stacking still matches what this PR's body describes: the three commits below the drop commit are #30091's, and merging this merges that. All three FormalityCheck runs on f8a2e50 are green.

---

Generated by Claude Code