Bug#1118939: rocr-runtime: Makes pymia to FTBFS due to unmet build-depends
On 2025-10-25 23:36, Santiago Vila wrote:
> On Sat, Oct 25, 2025 at 11:17:02PM +0200, Christian Kastner wrote:
>> Second: even with the removed Breaks, this doesn't fix Santiago's issue
>> because Santiago's build is still trying to install libamdhip64-5, which
>> is no longer available in unstable.
>
> Hmm, I triggered a rebuild in Salsa CI and the package built ok:
>
> https://salsa.debian.org/med-team/pymia/-/jobs/8501122
>
> so I went ahead and did the upload of pymia.
>
> So I don't fully understand this Second part.
Hm, that only seems to use libamdhip64-5, which naturally will still
linger as long as something references it, as in this case here.
I was assuming libamdhip64-dev as a B-D. As this -dev lacks a SOVER,
only one can exist in unstable, and the current one depends on
libamdhip64-6.
I was too strict in that regard. It's of course entirely possible for
some other package to rebuild and work successfully, without change,
with the newer libamdhip64-dev; say because it doesn't touch
functionality which effected the SOVER bump.
But this still needs to be evaluated as part of a transition, and bugs
filed where appropriate.
Best,
Christian
Reply to: