The latest release now also sorts workspace dependencies, as well as
different dependency sections. Keeping these things sorted reduces the
chances of merge conflicts when multiple PRs edit these files.
Tauri's `deb` and `rpm` bundler have support for configuring maintainer
scripts. We can therefore just use those instead of tearing apart the
`deb` file that it creates and rebuilding it ourselves.
Our `rpm` packaging is currently completely broken as well. I couldn't
get it to work on CentOS 9 at all due to missing dependencies, likely
introduced by our move to Tauri v2. It installs fine on CentOS 10
though, assuming that the user has the EPEL repository installed which
provides the WebView dependency. I extended the docs to reflect this.
Hence, with this PR, we drop support for CentOS 9 and now require CentOS
10. This allows us to remove a lot of cruft from our bundling process
and instead entirely rely on the Tauri provided bundler.
Lastly, for consistency with other platforms, the name of the
application in places like app drawers has been changed from "Firezone
Client" to just "Firezone".
---------
Signed-off-by: Thomas Eizinger <thomas@eizinger.io>
In #9072 this variable was removed in favor of populating it via an
Xcode build script. It appears however that the script does not take
effect properly when run from CLI and we need to populate this variable
again.
The name IPC service is not very descriptive. By nature of being
separate processes, we need to use IPC to communicate between them. The
important thing is that the service process has control over the tunnel.
Therefore, we rename everything to "Tunnel service".
The only part that is not changed are historic changelog entries.
Resolves: #9048
Our bump-versions script did not consider that we also have the version
stored in the `.spec` file for rpm builds and hence this did not get
bumped in a while (or ever?).
We already have a pretty powerful IPC framework in place to communicate
between the GUI and the service process. The deeplink implemenation uses
the same IPC mechanisms (UDS / pipes), yet it is effectively a
re-implementation of what we already have, just with less functionality.
In order to provide a more sophisticated handling of the case where
Firezone is launched again while it is already running, we refactor the
deeplink module to reuse the existing IPC framework. This makes it quite
easy to then reuse this in order to ping the already running Firezone
process that a new instance was launched.
For now, this doesn't do anything other than writing a log entry. This
however lays enough ground-work for us to then implement a more
sophisticated handling of that case in the future, e.g. open new windows
etc.
One caveat here is that we are now trying to connect to an existing IPC
socket on every startup, even the first one. Our IPC code has a retry
loop of 10 iterations to be more resilient on Windows when connecting to
pipes. Without any further changes, this would now delay the start of
Firezone always by 1s because we would try to connect to the socket 10x
before concluding that we are the first instance. To fix this, we make
the number of attempts configurable and set it to 1 when attempting to
the GUI IPC socket to avoid unnecessary delays in starting up the
Client.
Related: #5143.
When developing the macOS app, we always build the exact same version
and build code for each build. ~~This _may_ be one reason why we
constantly have to deactivate the extension before the new one will
launch.~~ Edit: Just tested, and I can verify that this does fix the
issue on dev builds, so no more having to uninstall the sysex between
builds.
Even if that's not the reason, this is a cleaner approach than building
it in our prod-only scripts.
---------
Signed-off-by: Jamil <jamilbk@users.noreply.github.com>
When updating the provisioning profiles (i.e. when changing anything the
Apple Developer Portal), we needed to manually update these build
scripts to point to the new UUIDs.
This can be made simpler to automatically pull it out of the profiles in
CI.
Somewhere between Xcode 16.0 and Xcode 16.3, the API for the libresolv
functions we call changed slightly, and we can now pass the return value
of `__res_9_state()` directly to the `res_9_ninit`, `res_9_ndestroy` and
`res_9_getservers` functions.
Microsoft Intune's DMG provisioner currently fails unexpectedly when
trying to provision our published DMG file with the error:
> The DMG file couldn't be mounted for installation. Check the DMG file
if the error persists. (0x87D30139)
I ran the following verification commands locally, which all passed:
```
hdiutil verify -verbose <dmg>
hdiutil imageinfo -verbose <dmg>
hdiutil hfsanalyze -verbose <dmg>
hdiutil checksum -type SHA256 -verbose <dmg>
hdiutil info -verbose
hdiutil pmap -verbose <dmg>
```
So the issue appears to be most likely that Intune doens't like the
`/Applications` shortcut in the DMG. This is a UX feature to make it
easy to drag the application the /Applications folder upon opening the
DMG.
So we're publishing an PKG in addition to the DMG, which should be a
more reliable artifact for MDMs to use.
---------
Signed-off-by: Jamil <jamilbk@users.noreply.github.com>
In #8480, we changed the location that `firezone-gateway` gets
downloaded to but forgot to update the knowledgebase with the new path.
---------
Signed-off-by: Thomas Eizinger <thomas@eizinger.io>
In order to get more recent tools like `cargo deny` that support Rust
2024, we need to bump the nixpkgs release we depend on to 24.11. As part
of doing so, we simplify the flake to not depend on `flake-utils` as we
only build for a single system anyway.
