[Buildroot] [PATCH v6 3/6] package/polkit: add systemd service file.
aduskett at gmail.com
Sun Dec 8 21:32:38 UTC 2019
On Sun, Dec 8, 2019 at 1:31 PM Adam Duskett <aduskett at gmail.com> wrote:
> On Sun, Dec 8, 2019 at 11:57 AM Thomas Petazzoni
> <thomas.petazzoni at bootlin.com> wrote:
> > Hello,
> > On Sun, 8 Dec 2019 10:46:58 -0800
> > Adam Duskett <aduskett at gmail.com> wrote:
> > > > Why aren't you using the one from polkit itself? There is one in
> > > > data/polkit.service.in, which gets compiled to polkit.service, and
> > > > apparently installed if HAVE_SYSTEMD is enabled. There also a D-Bus
> > > > service file at data/org.freedesktop.PolicyKit1.service.in.
> > > I guess I could.
> > >
> > > >
> > > > > - Add a polkitd user upon which /usr/lib/polkit-1/polkitd relies.
> > > > > - Set appropriate directory permissions to allow the polkitd user to
> > > > > access rules and actions.
> > > > > - Set appropriate default attributes for the pkexec binary.
> > > >
> > > > These could be done separately from the systemd service file addition.
> > > >
> > > No, they can't. The user is required to access those particular
> > > directories during
> > > the service startup.
> > They can be added before, as a separate patch. Note that if I hadn't
> > come up with the systemd unit question, I would have applied as-is.
> > > I don't have time to make a new patch set, so I am going to just mark the entire
> > > patch series as rejected and move on.
> > >
> > > If anybody else wants to resurrect this patch series they may do so.
> > That's a bit said. What you've done is very good, it includes
> > additional test cases.
> > I'm surprised you're giving up at v6. polkit was a complicated topic,
> > other folks tried in the past, and gave up way before they reached the
> > point where you are now, which is essentially very close to what can be
> > merged.
> > Best regards,
> > Thomas
> > --
> > Thomas Petazzoni, CTO, Bootlin
> > Embedded Linux and Kernel engineering
> > https://bootlin.com
> I appreciate the compliment, however, I do get frustrated because I
> tend to submit complicated
> patches that sit for weeks, months or even years without any feedback.
> Polkit is a pretty big exception
> to this rule, and I do appreciate the fairly quick review.
> I just fear that if I spend time making a v7, the patch set will sit
> for months without any feedback, or if there is
> then I will submit a v8 which will include the fixes requested, and
> then another maintainer will chime in with
> more requested fixes even though they didn't say anything to the
> previous patch set when it has been sitting
> without feedback for months. This has been a pain point for me for
> quite some time.
> I understand we are all volunteers with families and lives, however
> this also means that more complicated patches
> get either zero reviews, sit for very long periods of time, or are
> never reviewed at all. This makes me really apprehensive
> about submitting anything other than simple fixes and version bumps
> that require nothing other than a version/hash update
> from now on.
> I will see if I can apply your requested fixes in the meantime.
As a side note: I know my patches aren't perfect, so I also understand
that not a lot of people even want to review them in the first place.
More information about the buildroot