Re: Debian Stable Updates Announcement SUA 197-1
On Tue 23 Mar 2021 at 07:52:52 (-0400), The Wanderer wrote:
> On 2021-03-23 at 07:43, Greg Wooledge wrote:
>
> > I think the request was really "Please tell us how to use this
> > buster-proposed-updates thing, which I've never heard of before."
Actually I thought David wanted to do what I did, which is *not* to add
buster-proposed-updates to his sources.list, but just to poke around.
With good reason: ISTR that David runs his farm of machines on stretch.
I didn't think anyone would miss the pattern in sources.list's buster,
buster-updates, buster-proposed-updates.
However, I was unprepared for David's response today, of going back
to square one.
> > Here's a wiki page: <https://wiki.debian.org/StableProposedUpdates>
> > However, I would not follow its advice literally. You *never* want
> > to put the word "stable" in your sources.list. You want to put the
> > actual name of a release instead.
>
> That depends on what your update patterns are. It's probably generally
> good advice, but it's certainly not universal.
Agreed, it's not universal practice, but it might be worth having
as a default for advice.
> For myself, I have stable and testing in my sources.liist, and usually
> update at least weekly if not daily. This is effectively updating
> against testing, but keeps packages in stable which have been removed
> from testing available to be installed if I want them.
>
> When testing is released as stable and a new testing becomes available,
> I lose access to the packages which were previously in stable (now
> oldstable), which are old enough that it's probably not reasonable to
> want to install them without specifically knowing that that's what
> you're doing; I retain access to the packages previously in testing
> (because they're now in stable, which is already in my sources.list); I
> gain access to the new testing, to continue the pattern; and I do it all
> without having to update sources.list for the purpose.
>
> I don't see how that's unsafe or unreasonable; certainly I've been doing
> it for years, I believe for over a decade, without encountering issues
> which I think could even potentially be traced to it.
>
> > So, wherever the wiki says to use stable-proposed-updates, please
> > use buster-proposed-updates instead.
>
> This is generally reasonable, however. My update pattern is unlikely to
> be typical, and might be considered a special use-case.
To be fair, once you've added testing, then you're in the category of
a rolling-update Debian distribution, which is not the same as the
target of that webpage, a rolling-update Debian Stable distribution.
The problem that putting "stable" into sources.list causes is, of
course, the shock of Release Day to people who might be quite
unprepared for it. It's difficult to envisage that you're unprepared
for the effect of that day on testing.
Cheers,
David.
Reply to: