1. Forum
  2. Usenet
  3. LINUX.DEBIAN.BUGS.DIST

  • Bug#1104638: python3-icalendar: Depends on a transitional package

    From Santiago Vila@21:1/5 to All on Mon Jul 21 22:30:01 2025
    Hi.

    I've noticed that "todoman" does not build in unstable.
    A debbisect tells me this:

    bisection finished successfully
    last good timestamp: 20250712T233653Z
    first bad timestamp: 20250713T022514Z
    only one package differs: python3-icalendar 6.0.1-2 -> 6.0.1-3

    Are you sure that the changes in 6.0.1-3 are correct?

    Usually when one replaces a dummy package in depends or build-depends
    by the package actually having the functionality, no new FTBFS issues
    should be expected.

    This is a list of packages which FTBFS in unstable today but they were
    never reported yet, and I fear that python3-icalendar is also to blame
    for some of them:

    ecbuild
    glance
    mistral
    python-idasen-ha
    python-oslo.concurrency
    python-oslo.messaging
    python-recurring-ical-events
    python-x-wr-timezone
    resvg
    todoman

    Did not have time to look in detail, but at least two of them have
    failing autopkgtests here:

    https://qa.debian.org/excuses.php?package=python-icalendar

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Santiago Vila@21:1/5 to All on Mon Jul 21 22:40:01 2025
    I think the problem is in commit [6718d1e].

    python3-tz was dropped from Depends, but the proper fix would have
    been to use python3-pytz instead, as some packages which build-depend
    on such binary package rely on such dependency to be there.

    If nobody objects, I will readd the missing Depends via team upload.

    Not that we will ask for this package to be unblocked but at least
    we will start un-breaking unstable.

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Colin Watson@21:1/5 to Santiago Vila on Tue Jul 22 01:00:01 2025
    On Mon, Jul 21, 2025 at 11:30:37PM +0200, Santiago Vila wrote:
    On Mon, Jul 21, 2025 at 10:05:52PM +0100, Colin Watson wrote:
    I think the python-icalendar change was correct: python-icalendar doesn't
    really need pytz any more (it just has some affordances for the case where >> pytz is installed), and the cases that broke all imported pytz directly.

    Ok, I trust you, but I am a little bit puzzled that we are breaking
    things at this point of the release cycle.

    Yes, sorry, I probably should have waited but I think there's no great
    harm done: I don't expect any of this to migrate to trixie.

    --
    Colin Watson (he/him) [[email protected]]

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Santiago Vila@21:1/5 to Colin Watson on Tue Jul 22 11:30:01 2025
    On Mon, Jul 21, 2025 at 11:48:31PM +0100, Colin Watson wrote:
    On Mon, Jul 21, 2025 at 11:30:37PM +0200, Santiago Vila wrote:
    On Mon, Jul 21, 2025 at 10:05:52PM +0100, Colin Watson wrote:
    I think the python-icalendar change was correct: python-icalendar doesn't really need pytz any more (it just has some affordances for the case where
    pytz is installed), and the cases that broke all imported pytz directly.

    Ok, I trust you, but I am a little bit puzzled that we are breaking
    things at this point of the release cycle.

    Yes, sorry, I probably should have waited but I think there's no great harm done: I don't expect any of this to migrate to trixie.

    Well, you managed to fix the side effects, and very quickly, so we can
    be happy that the issue is solved now, and I thank you a lot. ❤️

    btw: I found the root cause for four of the FTBFS issues in the list I
    posted yesterday:

    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1109700

    The four affected packages are from Openstack, so they will be able to
    choose if they want to fix python-eventlet, or the four individual
    affected packages instead.

    Thanks.

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