Re: [RFC PATCH 3/5] maintenance: use packaged systemd units

From: Max Gautier <mg@max.gautier.name>
To: Patrick Steinhardt <ps@pks.im>
Cc: git@vger.kernel.org, "Lénaïc Huard" <lenaic@lhuard.fr>,
	"Derrick Stolee" <stolee@gmail.com>
Subject: Re: [RFC PATCH 3/5] maintenance: use packaged systemd units
Date: Thu, 21 Mar 2024 14:19:51 +0100	[thread overview]
Message-ID: <Zfwz9zs3odKcU9Dm@framework> (raw)
In-Reply-To: <ZfwqFU_b3MqLCZNR@tanuki>

On Thu, Mar 21, 2024 at 01:37:41PM +0100, Patrick Steinhardt wrote:
> On Mon, Mar 18, 2024 at 04:31:17PM +0100, Max Gautier wrote:
> > - Remove the code writing the units
> > - Simplify calling systemctl
> >   - systemctl does not error out when disabling units which aren't
> >     enabled, nor when enabling already enabled units
> >   - call systemctl only once, with all the units.
> > - Add clean-up code for leftover units in $XDG_CONFIG_HOME created by
> >   previous versions of git, taking care of preserving actual user
> >   override (by checking the start of the file).
> 
> We very much try to avoid bulleted-list-style commit messages. Once you
> have a list of changes it very often indicates that there are multiple
> semi-related things going on in the same commit. At that point we prefer
> to split the commit up into multiple commits so that each of the bullet
> items can be explained separately.
> 
> Also, as mentioned before, commit messages should explain what the
> problem is and how that problem is solved. 
> 
> Patrick
> 

I need to think a bit about patch ordering here because changing from
the "random minute" implementation to the systemd
(RandomDelaySec/FixedRandomDelay) is required to the other changes of
putting the unit out of the code and in the package.

I think I have an idea how, I'll work on that, thanks.

-- 
Max Gautier