On Sun, Dec 14, 2025 at 09:41:09PM +0100, Paul Gevers wrote:
> Hi,
> 
> On 9/25/25 10:39, Adrian Bunk wrote:
> > For future reference, below is a paste how excuses for octave look like
> > right now due to the transition (before the binNMUs were scheduled).
> > 
> > I am looking at the following pattern:
> > ∙ ∙ migrating octave/10.2.0-2/amd64 to testing makes dynare/6.4-3/amd64 
> > uninstallable
> > ∙ ∙ autopkgtest for dynare/6.4-3: amd64: Regression ♻ (reference ♻)
> > 
> > It does not make much sense to attempt running the migration autopkgtest 
> > when
> > it is already known that at least one of the binary packages from the 
> > package
> > to be tested becomes uninstallable.
> 
> 
> I wonder if commit 8801f97 improved this. I'm not very sure though, probably
> worth a test case.

It hasn't fixed it (at least not in all cases):

Migration status for hypre (2.33.0-3 to 3.0.0-4): BLOCKED: Rejected/violates 
migration policy/introduces a regression
Issues preventing migration:
∙ ∙ migrating libhypre-dev/3.0.0-4/amd64 to testing makes 
libmfem-dev/4.9+ds-2/amd64 uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/arm64 to testing makes 
libmfem-dev/4.9+ds-2/arm64 uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/armhf to testing makes 
libmfem-dev/4.9+ds-2/armhf uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/i386 to testing makes 
libmfem-dev/4.9+ds-2/i386 uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/ppc64el to testing makes 
libmfem-dev/4.9+ds-2/ppc64el uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/riscv64 to testing makes 
libmfem-dev/4.9+ds-2/riscv64 uninstallable
∙ ∙ migrating libhypre-dev/3.0.0-4/s390x to testing makes 
libmfem-dev/4.9+ds-2/s390x uninstallable
∙ ∙ Autopkgtest for hypre/3.0.0-4: amd64: Pass, arm64: Pass, i386: Pass, 
ppc64el: Pass, riscv64: Pass, s390x: Pass
∙ ∙ Autopkgtest for mfem/4.9+ds-2: amd64: Regression ♻ (reference ♻), arm64: 
Regression ♻ (reference ♻), i386: Regression ♻ (reference ♻), ppc64el: 
Regression ♻ (reference ♻), riscv64: Regression ♻ (reference ♻), s390x: 
Regression ♻ (reference ♻)


> Paul

cu
Adrian

Reply via email to