• Bug#1105174: src:rkward: fails to migrate to testing for too long

    From Santiago Vila@21:1/5 to All on Fri Jul 25 01:30:02 2025
    Hello.

    This package should propagate to testing to fix RC bug #1103204.

    However, it has r-base-core (>= 4.5.1-1) in its Depends line, while
    r-base-core in trixie is at version 4.5.0-3, so the migration
    will not happen in a natural way.

    After asking Dirk (maintainer of R packages, in CC) for advice about
    this, it seems there are basically two ways to deal with this.

    A) Make an upload for trixie-proposed-updates. This would put the
    package directly in trixie with >= trixie-version.

    B) Hardcode (>= 4.5.0-3) (or maybe 4.5.0-3~ or just 4.5.0) in the
    control line, only for trixie, and only as a one-time thing. Dirk
    tells me that in cases like this one, this would be ok, as R does not
    really care about such minor versioning.

    Please indicate if you (maintainers of rkward) are available to fix
    this, because otherwise Dirk tells me he would be willing to NMU
    the package to handle the issue.

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Dirk Eddelbuettel@21:1/5 to Santiago Vila on Fri Jul 25 02:30:01 2025
    On 25 July 2025 at 01:18, Santiago Vila wrote:
    | Hello.
    |
    | This package should propagate to testing to fix RC bug #1103204.
    |
    | However, it has r-base-core (>= 4.5.1-1) in its Depends line, while
    | r-base-core in trixie is at version 4.5.0-3, so the migration
    | will not happen in a natural way.
    |
    | After asking Dirk (maintainer of R packages, in CC) for advice about
    | this, it seems there are basically two ways to deal with this.
    |
    | A) Make an upload for trixie-proposed-updates. This would put the
    | package directly in trixie with >= trixie-version.
    |
    | B) Hardcode (>= 4.5.0-3) (or maybe 4.5.0-3~ or just 4.5.0) in the
    | control line, only for trixie, and only as a one-time thing. Dirk
    | tells me that in cases like this one, this would be ok, as R does not
    | really care about such minor versioning.

    For added colour: I am aware of breakage with other applications built
    against R 4.4.* and linking to it now falling over of a missing symbol under
    R 4.5.* -- this happened famously with rstudio (outside of Debian).

    Changes within a minor version are known-compatible so compiling under R
    4.5.1 but allowing R 4.5.0 as a runtime will work, and I offered Santiago to help with a NMU.

    Obviously not trying to either 'jump a gun' or stepping onto somebody's toes
    so if any of you as group maintainers want to deal with it I will merrily defer. Else I am happy to step in an help with a quick one-time override to debian/control depending on 'R (>= 4.5.0)' as opposed to the automagic
    fill-in from $rvers which will get us 4.5.1 which is only in unstable and not in testing.

    Cheers, Dirk

    | Please indicate if you (maintainers of rkward) are available to fix
    | this, because otherwise Dirk tells me he would be willing to NMU
    | the package to handle the issue.
    |
    | Thanks.

    --
    dirk.eddelbuettel.com | @eddelbuettel | edd@debian.org

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