On 2025-12-28 at 08:07, didier gaumet wrote: > Le 28/12/2025 à 13:31, The Wanderer a écrit : > >> Plenty of them. >> >> To pick just one example: in current testing, tellico and >> tellico-doc are both available at version 4.1.4-2, but when I ran >> >> $ apt-get install tellico >> >> (or rather, a command line listing multiple packages all already >> marked as manually installed, with tellico being one of them), >> tellico was upgraded to that version but tellico-doc was left >> behind at version 4.1.3-1. > > Have you tried several times? Perhaps when you tried tellico-doc was > not yet updated in the repo?
Every upgrade attempt is after running 'apt-get update'.
If I instead run 'apt-get dist-upgrade', it shows tellico-doc as one of
the packages that will be upgraded. (I don't want to use that as the
actual upgrade method, however, for the reasons that I explained in the
original post.)
> After an apt update, what does apt policy tellico-doc tells?
$ apt-cache policy tellico{,-doc}
tellico:
Installed: 4.1.4-2
Candidate: 4.1.4-2
Version table:
*** 4.1.4-2 900
900 http://ftp.us.debian.org/debian testing/main amd64 Packages
100 /var/lib/dpkg/status
4.1.1-1 800
800 http://ftp.us.debian.org/debian stable/main amd64 Packages
tellico-doc:
Installed: 4.1.3-1
Candidate: 4.1.4-2
Version table:
4.1.4-2 900
900 http://ftp.us.debian.org/debian testing/main amd64 Packages
900 http://ftp.us.debian.org/debian testing/main i386 Packages
*** 4.1.3-1 100
100 /var/lib/dpkg/status
4.1.1-1 800
800 http://ftp.us.debian.org/debian stable/main amd64 Packages
800 http://ftp.us.debian.org/debian stable/main i386 Packages
# apt-get install tellico
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
tellico is already the newest version (4.1.4-2).
Solving dependencies... Done
0 upgraded, 0 newly installed, 0 to remove and 1388 not upgraded.
This is after having already run the equivalent of that second command
line while tellico was still at version 4.1.3-1, and seen it get
upgraded, while seeing tellico-doc not get upgraded.
> And you have Testing, not Stable, so bugs are more often to be
> expected. Tellico-doc does not seem to have bugs reported in testing
This is not a tellico-specific, or even package-specific, issue. As I
already said, this is just one example, and there are plenty of others.
After my last 'apt-get update', I ran through a series of command lines like
# apt-get install $(apt-mark showmanual | grep ^[abcde])
to upgrade marked-as-manually-installed packages and their dependencies;
that is my normal practice, for the reasons I gave in the original post.
After getting to a point where as many packages as possible had been
upgraded that way (the exceptions being packages that would be listed as
"kept back" by an 'apt-get dist-upgrade' command), if I run 'apt-get
dist-upgrade', it reports in part:
1373 upgraded, 43 newly installed, 2 to remove and 15 not upgraded.
*Every single one of* those 1,373 packages is a manifestation of the
issue I am trying to find a solution for. (I could provide the list of
the packages, if you want, but I doubt that would be helpful.)
There is absolutely zero possibility that anything specific to a
particular package-in-need-of-upgrade can be the cause of the issue. The
issue is that apt handles dependency-chain following differently
(whether with regard to Recommends: or with regard to package versions)
on upgrade installs as compared against initial installs.
I am asking whether there is a way to tell apt to behave differently in
that regard, either in general or (better) for a specific invocation.
> and I let you verify if your problem is known as an apt (package
> containing apt-get) bug in testing:
> https://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=apt;dist=testing
This behavior has existed for years upon years. If it is considered a
bug, it is not a new one, and it is unlikely to be fixed any time
remotely soon (unless someone, e.g. me, steps up to write a patch and
convince the maintainers to include it).
I might still file a bug report about the matter, but I would first
rather check whether there is already a way (that I've missed noticing)
to make apt do what I want; that's exactly what I'm doing by creating
this thread.
Even if I do file a bug report about it, that bug report would be a
wishlist-level feature request, not a report of the program misbehaving.
(I am currently checking - via reportbug, not via the URL you gave,
although they should produce similar or identical results - for open
bugs against apt which might be requesting that feature already.
Unfortunately, however, the BTS seems to be exceedingly slow at the
moment; I've seen individual bug reports take a minute or more to even
start loading (w3m sits waiting on an "Opening socket..." message), and
reportbug just timed out twice in a row on trying to connect at the SSL
handshake stage. I don't think it'll be practical to check for existing
bugs on this until whatever issue is present there gets fixed.)
--
The Wanderer
The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man. -- George Bernard Shaw
signature.asc
Description: OpenPGP digital signature
