Indirect dependencies on libsystemd-dev

[The Wanderer]

(I haven't received a copy of the message I'm replying to via the list -
i.e., with the necessary headers for e.g. "reply to list" to function;
the only one I've received was sent to me directly, and is timestamped
close to three hours ago.

Is the list copy just delayed somehow, or is something in the middle
filtering it out, or is the list software somehow actually intentionally
configured to detect when list members are in the addressee list and
skip sending the list copy? I would consider that latter behavior to be
actively undesirable; although I dislike receiving multiple copies on a
routine reply, I would far rather receive multiple copies than not
receive the copy sent via the list, so that I get those headers and any
other modifications that the list may make.

I seem to recall that I once before took part in a discussion on this
list that also involved a bug-number address and didn't receive all the
copies I expected to, but I thought that was because of the bug-tracker
system and not because of the mailing list.)

On 2022-07-17 at 12:35, Thorsten Glaser wrote:

> On Sun, 17 Jul 2022, The Wanderer wrote:
> 
>> The current version of libfluidsynth-dev (2.2.7-1) depends on,
>> among other packages, libsystemd-dev. (As I haven't had
>> libfluidsynth-dev
> 
> Introduced in 2.0.5-1 and probably legit. It’s a -dev package, and if
> libfluidsynth*.so.* links against libsystemd0, this is expected.

Assuming it's built that way, yes (as ldd seems to confirm that it is,
although the package containing that file doesn't list a dependency on
libsystemd0, so there's something else weird there). The question is
whether there's an alternate way to get things set up so that whatever
they want from this can be achieved, while also not introducing this
conflicting dependency.

(I find it hard to imagine what a MIDI-synthesizer library might need
which it would want to get from libsystemd0, especially given that it
didn't need to depend on these various other libraries before. My best
guess is something involving hardware access, but I'd intuitively expect
that to involve libudev rather than libsystemd...)

> You’ll normally only install -dev packages in a buildd/chroot.

My reflexive reaction is that this is blatantly inaccurate. I have to
stop and think for a moment to recognize the context in which it would
probably hold true: that of compiling for inclusion in a Debian package.

I do not remotely consider that to be the normal, much less the only
normal, use case for installing -dev packages. To the extent that Debian
considers it to be so, I consider that to be a flaw in Debian, and
potentially a regressive one (in that I'm reasonably sure it was not
always the case).

I routinely compile things on the host machine for use on the host
machine, without a package as intermediary; some of them don't have
packaging files already created, or at least don't have them shipped
with the source, since I get the source from the upstream development
repository. (And I wouldn't want to have to keep packaging files around
alongside the upstream source, and update them whenever the upstream
source changes; at that point I might as well be maintaining the
package.)

I *very definitely* would *not* want to be limited to working in a
chroot or the like, or required to write package metadata files, when
*writing and build-testing my own software*. It's unlikely that I"ll be
writing software that uses libfluidsynth at any point, but that doesn't
affect the fact of the principle.


I interpret your response as boiling down to one or both of "no, there
isn't likely to be any practical way to get this dependency chain
broken" and "there's no point in raising this as an issue with the
appropriate maintainers, so it should just be treated as a loss and left
alone". That's disappointing/depressing, in that it's a demonstration of
how it's still getting harder to avoid systemd, but not an entirely
surprising outcome.

-- 
   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

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://www.chiark.greenend.org.uk/pipermail/debian-init-diversity/attachments/20220717/d13c6e49/attachment.sig>