samiandClaude Sonnet 5
31eadffce5
DAEMON landed packaging/nativehost/, packaging/systemd/{velox.service,velox.socket}
(real sd_listen_fds() socket activation) and cli/man/velox.1 on main while this was
in flight, duplicating paths this branch had drafted independently. Per CLAUDE.md
(packaging/nativehost/ is DAEMON's; the rest of packaging/ is PKG/QA's), DAEMON's
copies are kept as the single source and this branch's packaging/native-messaging/,
packaging/man/velox.1 and stale packaging/systemd/velox.service are dropped rather
than maintained twice. The root CMakeLists.txt's install() rules now point at
DAEMON's paths directly, install both systemd units (not just velox.service), and
docs/07-packaging.md / packaging/README.md no longer claim there's no socket
activation — there is, and this package now ships it.
Adds the local-test-only .deb itself: debian/{control,rules,postinst,postrm,
changelog,copyright}, a placeholder icon set, the desktop entry and PKG/QA's own
man pages (veloxd.8, velox-gui.1 — velox.1 stays DAEMON's). libveloxcore stays
static (no consumer needs a .so yet); pairing has no real approval UI in this
build (D1 unbuilt), so postinst and packaging/README.md both say VELOX_PAIR_AUTO=1
is required and print it prominently. No PPA, no GPG — local dpkg -i only.
Verified live: dpkg-buildpackage -us -uc -b, dpkg -i, socket activation (systemctl
--user enable --now velox.socket), `velox ls` against the running daemon, lintian
clean, dpkg -r removes the manifest/units and leaves $XDG_DATA_HOME/velox alone.
.gitignore now excludes dpkg-buildpackage's debhelper build tree and obj-*/ (the
debuild output dir) — debian/control et al. stay tracked, everything debhelper
regenerates does not.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_0176u3fTrxegrGNm2yC7r69W
102 lines
6.0 KiB
Markdown
102 lines
6.0 KiB
Markdown
Owner: lane PKG/QA. See ../docs/07-packaging.md.
|
|
|
|
## `.deb` — local test build only
|
|
|
|
There is no PPA, no GPG signing, no Launchpad wiring here. This produces a `.deb` you
|
|
install with `dpkg -i` on the box you built it on (or an identical one), for testing —
|
|
not something to hand to a user or publish anywhere. See `docs/07-packaging.md`'s
|
|
package matrix for the eventual PPA plan.
|
|
|
|
```sh
|
|
sudo ./tools/bootstrap.sh --packaging # debhelper, dpkg-dev, lintian, devscripts, fakeroot
|
|
dpkg-buildpackage -us -uc -b # from the repo root; -b = binary only, no signing
|
|
sudo dpkg -i ../velox_0.1.0-1_amd64.deb
|
|
lintian ../velox_0.1.0-1_amd64.deb
|
|
```
|
|
|
|
Uninstall: `sudo dpkg -r velox` (or `sudo dpkg -P velox` to purge — see below for what
|
|
purge does and does not do here).
|
|
|
|
### Known limitations of this build — read before relying on any of these
|
|
|
|
- **`libveloxcore` is static, not `.so.1`.** The install layout in `docs/07-packaging.md`
|
|
originally called for a shared, versioned `libveloxcore.so.1`; the user decided against
|
|
building one for now (no consumer needs it as a shared object yet, and it would need a
|
|
`SOVERSION` added to CORE's own `core/CMakeLists.txt`). The doc was corrected in the
|
|
same commit as this packaging work — it no longer describes something unshipped.
|
|
|
|
- **Placeholder app icon.** `gui/resources/icons/` (GUI's lane) is empty — no real icon
|
|
art exists anywhere in the repo yet. `packaging/icons/hicolor/*/apps/velox.png` is a
|
|
plain generated placeholder (flat color, a download-arrow glyph), good enough to
|
|
install correctly and pass `desktop-file-validate`/lintian's icon checks, and nothing
|
|
more. **Replace it before any real release** — and do the icon-set licence audit
|
|
`docs/07-packaging.md`'s release checklist asks for once real art exists (this
|
|
placeholder was drawn from scratch for this package, so it has no licence question of
|
|
its own, but it is not real Velox branding either).
|
|
|
|
- **No license has been chosen for Velox's own source yet.** There is no top-level
|
|
`LICENSE` file in the repo. `debian/copyright` says so plainly (`License:
|
|
Unlicensed-TODO`) rather than inventing one — that is a real project decision, above
|
|
any one lane. Expect `lintian` to flag this package over it; that is lintian doing its
|
|
job, not a bug in this packaging, and it stays a known exception to "lintian clean"
|
|
until the project adopts a license. The release checklist's "Licence audit" item (Qt,
|
|
libcurl, SQLite, ffmpeg, icon set) is separate and still open regardless.
|
|
|
|
- **Pairing has no real approval UI in this build.** `libdbus-1-dev` is not in this
|
|
build's `Build-Depends` (D1's GUI-dialog/desktop-notification approver needs it and
|
|
isn't wired up yet — `daemon/src/main.cpp` still uses `EnvAutoApprover`
|
|
unconditionally). That approver auto-**rejects** every pairing request unless
|
|
`VELOX_PAIR_AUTO=1` is set in `veloxd`'s environment — there is no prompt, no
|
|
accept/decline, it is all-or-nothing. `postinst` prints this prominently after
|
|
install; it is repeated here so it isn't missed by anyone reading only one of the two.
|
|
**Do not set `VELOX_PAIR_AUTO=1` on a machine or account where auto-accepting every
|
|
pairing request is unacceptable.**
|
|
|
|
- **Socket activation is real.** `velox.service` and `velox.socket` both ship, verbatim
|
|
from DAEMON's `packaging/systemd/README.md` — `veloxd` receives the pre-bound fd via
|
|
`sd_listen_fds()` (`daemon/src/rpc/systemd_activation.cpp`) rather than binding its own
|
|
socket. `debian/rules` does **not** disable `dh_installsystemduser` (only the
|
|
system-level `dh_installsystemd`, since these are `--user` units); debhelper finds both
|
|
units under `lib/systemd/user/` in the install tree and enables the pair together
|
|
(`Also=velox.socket` in `velox.service`'s `[Install]`).
|
|
|
|
- **The systemd `--user` service starts enabled for your *next* login, not this one.**
|
|
debhelper's `dh_installsystemduser` (verified in the built package's `postinst`) marks
|
|
`velox.service` enabled automatically — that takes effect the next time each user's
|
|
`systemd --user` instance starts. It cannot reach an *already-running* session from a
|
|
root maintainer script, so for the session you're in right now, `postinst` prints:
|
|
```sh
|
|
systemctl --user daemon-reload
|
|
systemctl --user start velox.service
|
|
```
|
|
|
|
- **`dpkg -P velox` (purge) does not delete user data.** The uninstall test this package
|
|
is built to pass is "manifests, units and sockets removed; user data untouched *unless
|
|
purged*" — this build is deliberately conservative and leaves `$XDG_DATA_HOME/velox`
|
|
(normally `~/.local/share/velox`, the task database) alone on **both** `remove` and
|
|
`purge`, for every account. A root maintainer script deleting per-user data across
|
|
every account on the system is exactly the kind of destructive, hard-to-test operation
|
|
that deserves its own design pass, not a first cut bolted on here. `postrm` only ever
|
|
undoes what `postinst` did: the per-user native-messaging manifest it copied in.
|
|
|
|
### The uninstall test, concretely
|
|
|
|
What `docs/07-packaging.md`'s release checklist asks for, and what this build actually
|
|
does:
|
|
|
|
| Removed on `dpkg -r`/`-P`? | |
|
|
|---|---|
|
|
| `/usr/bin/{veloxd,velox,velox-gui}`, `/usr/libexec/velox/velox-nmhost` | yes — dpkg itself, ordinary package files |
|
|
| `/usr/lib/systemd/user/{velox.service,velox.socket}` | yes — same |
|
|
| `/usr/lib/mozilla/native-messaging-hosts/com.velox.host.json` | yes — same |
|
|
| Per-user `~/.mozilla/native-messaging-hosts/com.velox.host.json` | yes — `postrm`, both `remove` and `purge` |
|
|
| `$XDG_DATA_HOME/velox` (task DB, settings) | **no, on either `remove` or `purge`** — see above |
|
|
|
|
### Native messaging
|
|
|
|
`velox-nmhost` (DAEMON's `nmhost/`) and its manifest (DAEMON's
|
|
`packaging/nativehost/com.velox.host.json`) now ship in this package — the top-level
|
|
`CMakeLists.txt`'s install rules are guarded on `if(TARGET velox-nmhost)`, the same way its
|
|
`add_subdirectory()` calls are, so this lit up automatically the day that lane merged, no
|
|
packaging change needed on this end.
|