libhtml-formfu-perl/2.06000-1 breaks libhtml-formfu-model-dbic-perl/2.00-2 autopkgtest in testing

[Paul Gevers]

Hi Florian,

Thanks for figuring this out.

On 15-05-18 18:25, Florian Schlichting wrote:
> On Tue, May 15, 2018 at 02:11:56PM +0100, Paul Gevers wrote:
>> tl;dr: libhtml-formfu-perl/2.06000-1 breaks libhtml-formfu-model-dbic-perl/2.00-2 autopkgtest in testing
>> see: https://ci.debian.net/packages/libh/libhtml-formfu-model-dbic-perl/testing/amd64/
> 
> and
> 
>> tl;dr: libhtml-formfu-perl/2.06000-1 breaks libcatalyst-controller-html-formfu-perl/2.02-1 autopkgtest in testing
>> see: https://ci.debian.net/packages/libc/libcatalyst-controller-html-formfu-perl/testing/amd64/
> 
> is
> 
>> * deprecation of functionality that is used in the reverse dependency
>>   and/or its test case (discussion needed)
> 
> libhtml-formfu-perl shipped a patched version of a buggy module, which
> now got fixed, so libhtml-formfu-perl/2.06000-1 dropped that module
> (MooseX::Attribute::FormFuChained) and instead added a dependency on
> libmoosex-attribute-chained-perl (>= fixed_version).
> 
> libhtml-formfu-model-dbic-perl also added that dependency, and version
> 2.03-2 passes on debci (but old version 2.00-2 does not)
> 
> I wasn't aware of libcatalyst-controller-html-formfu-perl, for which
> there is also a new version available that adds the necessary
> dependency, which I wasn't aware of.

Great. That's why we run these tests.

> I could upload the new version
> tonight.

That would be great.

> However, I'm not sure if it's enough to have a fixed set of packages in
> unstable, and wait for them to eventually migrate. Or if we need a set
> of Breaks (in libhtml-formfu-perl, for the old versions of
> libhtml-formfu-model-dbic-perl and
> libcatalyst-controller-html-formfu-perl?) or something else, for
> migration to happen and partial updates / use as backport to reflect the
> problem on the Debian package level?

Well, for unstable-to-testing migration you can just wait until the
package migrate (and you can retrigger the tests for libhtml-formfu-perl
after the others migrate to speed it up), but I think a breaks is valid
and very defend-able. So if you want the fastest solution without
persuading me or the RT team to override stuff, a breaks is the quickest
and IMHO the best way forward and should be enough.

Paul

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-perl-maintainers/attachments/20180515/2b42cac7/attachment-0001.sig>