[PATCH v6 3/4] remove sysconfig files

Andrea Bolognani abologna at redhat.com
Mon Jan 3 11:18:11 UTC 2022


On Mon, Jan 03, 2022 at 11:31:20AM +0100, Olaf Hering wrote:
> Thu, 23 Dec 2021 10:29:33 -0800 Andrea Bolognani <abologna at redhat.com>:
> > On Tue, Dec 21, 2021 at 12:22:44PM +0100, Olaf Hering wrote:
> > > -# If systemd socket activation is disabled, then the following
> > > -# can be used to listen on TCP/TLS sockets
> > > -#LIBVIRTD_ARGS="--listen"
> >
> > This bit about passing --listen to libvirtd got lost during the move
> > to the unit file. Please make sure you preserve it.
>
> It was wrong from day one to have this fragment here.
> Documentation belongs to the doc directory.

The fact that we still QEMU_AUDIO_DRV and SDL_AUDIODRIVER in the
service file even after your changes goes against this principle. But
I don't necessarily disagree, given that --listen is documented both
in the manual page for libvirtd and elsewhere.

> > After this change, libvirt-guests' configuration becomes entirely
> > opaque: the only way the admin can learn how to configure the service
> > is by somehow realizing that it's a shell script as opposed to a
> > binary and looking inside it.
> >
> > Can we do better?
>
> libvirt-guests is undocumented. The file type is easy to guess, based on the filename in libvirt-guests.service.
> The Documentation= setting in this file is wrong, libvirtd(8) says nothing about this functionality.
>
> Initial documentation has to be provided as a separate change, by creating a new docs/whatever.rst.

This sounds very good in theory, but in practice we had some sort of
documentation in a place where the admin could reasonably be expected
to see it, and with your change we're now asking them to go poke
inside a script to find that same information, with no clear
indication that doing so is necessary. I believe that's a significant
downgrade in user experience which is not justified by the desire to
have a perfect separation between configuration files and
documentation.

Please either adopt the mitigation I suggested in my previous message
(add a short note to the service file and make sure configuration
variables are documented near the very top of the script) or
introduce proper documentation for libvirt-guests before making the
one we currently have much harder to locate.

-- 
Andrea Bolognani / Red Hat / Virtualization




More information about the libvir-list mailing list