• Bug#1061866: adns: NMU diff for 64-bit time_t transition

    From Steve Langasek@21:1/5 to Ian Jackson on Tue Jan 30 03:50:01 2024
    On Tue, Jan 30, 2024 at 12:34:50AM +0000, Ian Jackson wrote:
    To ensure that inconsistent combinations of libraries with their reverse-dependencies are never installed together, it is necessary to
    have a library transition, which is most easily done by renaming the runtime library package.

    I'm surprised at

    +Provides: ${t64:Provides}
    +Replaces: libadns1
    +Breaks: libadns1 (<< ${source:Version})

    I don't know why this isn't just a soname transition. But I think
    probably people more involved in this have thoughyt this through.
    (Perhaps this to do with making this a no-op on 64-bit arches.)

    Well first, sonames are generally considered upstream's namespace to manage
    so we would have to have Debian-specific sonames here which is a bit ugly.

    Second, changing the soname requires changes to the upstream build systems
    of each of the 1200 different packages, so that would really be incompatible with mass-NMUing.

    And third, yes, the ABI is only changing on 32-bit architectures; so maintaining compatibility with third-party packages that depend on e.g. libpam0g on amd64 has at least some value, and thus we maximize this value
    by applying compatibility rules consistently rather than trying to suss out which libraries it's worth keeping compatibility for.

    If you have any concerns about this patch, please reach out ASAP.
    Although this package will be uploaded to experimental immediately,
    there will be a period of several days before we begin uploads to
    unstable; so if information becomes available that your package should
    not be included in the transition, there is time for us to amend the planned uploads.

    If what I have written doesn't indicate that something has been
    overlooked, you should go ahead as planned. I guess I should avoid
    uploading myself.

    adns is not an "unusual" package, other than insofar as time_t being
    part of an ABI-exposed struct makes it unusual.

    Cheers,
    --
    Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developer https://www.debian.org/ slangasek@ubuntu.com vorlon@debian.org

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEErEg/aN5yj0PyIC/KVo0w8yGyEz0FAmW4YOAACgkQVo0w8yGy Ez2bZg//Wq5v0eU2P020Lrdn+i/J8xSJii3dgsnFh5faR/omMd/DxsuUntTUSlzf cfjUcxG38xGiiVuovs/lp9bp/dY5vQLvYUgo3A9PeCCeMCtTTyckUcfbqt5eX53Y dRITJNexYhKPdMY9pNyO/mxKVA/S+V0arRu5aseksiKSYST+r9KbuOlSIBriOfvC 1pXmY9igc2yfi9VKPH8cMoUiRPXYBuwvnc1VrO0KhzzfKU/wNPIlic6ni23/Polh cw+PhEAkprSwMbdjf81S1DtRxOvij09saNTnZHBruNXuYAOObAccptNs6GbKwAX5 nTwB0+wcTM+vxE0wO0iu3Qy2zOWFZnV024OJAM6sZT8IS2qjPmhohFnlh30KOZ2v NZlcrQk/g2qwpRHF2hi9yKr8tg+RsFQ56Qt49bLOj/Sfe99qqd62xnh71FJD5ROc N23x0H9wOMCwoqrPOXH1WJ/5GoHjhmp7EVhitW06QuJ8L/ye1iANY37d8MlsWgsS e7cyWOS8N84sF+ZoYqAjT90MWEOMb5GbzPfOB+X4f6g35o3FXbWPeJLkG11FAqLb 6JQZ3l8yXokZTQZbp9v8
  • From Steve Langasek@21:1/5 to All on Wed Jan 31 03:40:02 2024
    This is a multi-part MIME message sent by reportbug.


    Source: adns
    Followup-For: Bug #1061866

    Apologies, an oversight in the conversion script caused us to fail to update strict versioned dependencies on the previous package name. Please find attached a fixed patch.

    This has also now been uploaded to experimental.

    diff -Nru adns-1.6.0/debian/changelog adns-1.6.0/debian/changelog
    --- adns-1.6.0/debian/changelog 2020-08-23 20:25:53.000000000 +0000
    +++ adns-1.6.0/debian/changelog 2024-01-31 02:31:47.000000000 +0000
    @@ -1,3 +1,11 @@
    +adns (1.6.0-2.1~exp2) experimental; urgency=medium
    +
    + * Non-maintainer upload.
    + * Rename libraries for 64-bit time_t transition.
    + * Fix uninstallable packages from the previous upload.
    +
    + -- Steve Langasek <vorlon@debian.org> Wed, 31 Jan 2024 02:31:47 +0000
    +
    adns (1.6.0-2) unstable; urgency=medium

    * Don't ignore make clean failures.
    diff -Nru adns-1.6.0/debian/control adns-1.6.0/debian/control
    --- adns-1.6.0/debian/control 2020-08-23 20:25:53.000000000 +0000
    +++ adns-1.6.0/debian/control 2024-01-31 02:31:24.000000000 +0000
    @@ -10,7 +10,7 @@
    Package: libadns1-dev
    Section: libdevel
    Architecture: any
    -Depends: libadns1 (= ${binary:Version}), libc6-dev, ${misc:Depends}
    +Depends: libadns1t64 (= ${binary:Version}), libc6-dev, ${misc:Depends}
    Conflicts: libadns0
    Description: Asynchronous-capable DNS client library (develop
  • From Ian Jackson@21:1/5 to Ian Jackson on Wed Feb 7 20:10:01 2024
    Control: severity -1 important

    Ian Jackson writes ("Re: Bug#1061866: adns: NMU diff for 64-bit time_t transition"):
    I have just got an alert saying adns is now scheduled for autoremoval
    due to #1061866.

    My understanding was that you were intending to NMU to unstable after "several days". I have been holding off making an upload myself so as
    not to interfere.

    I'm not sure if I should:

    (i) wait

    (ii) apply that patch (on top of what's in experimental)
    and upload to experimental

    (iii) apply that patch on top of what's in experimental
    and upload the result to sid.

    For now I am going to downgrade this bug in the hope that the current
    answer is (i).

    Regards,
    Ian.

    --
    Ian Jackson <ijackson@chiark.greenend.org.uk> These opinions are my own.

    Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
    that is a private address which bypasses my fierce spamfilter.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)