Re: py311 conflict

[Greg Troxel <gdt%lexort.com@localhost>]

Jonathan Perkin <jperkin%mnx.io@localhost> writes:

> * On 2023-08-15 at 17:38 BST, nia wrote:
>
>>On Tue, Aug 15, 2023 at 12:30:58PM -0400, Greg Troxel wrote:
>>> > => Creating binary package
>>> > /Users/bacon/Pkgsrc/pkgsrc/packages/All/py311-anytree-2.9.0.tgz
>>> > ===> Installing binary package of py311-anytree-2.9.0
>>> > pkg_add: Conflicting PLIST with py310-anytree-2.9.0: bin/ebt
>>> > pkg_add: 1 package addition failed
>>> > *** Error code 1
>>> >
>>> > Stop.
>>> > bmake[3]: stopped in /Users/bacon/Pkgsrc/pkgsrc/devel/py-anytree
>>> > *** Error code 1
>>>
>>> Yep, exactly what it says.  This is how it goes with packages that
>>> install un-namespaced binaries.
>>
>>In my opinion the lack of a version suffix on the ebt binary (and
>>corresponding PKG_ALTERNATIVES support) is a bug, enough to file
>>a PR.

Agreed.

> Certainly leaving it to PLIST duplication to cause a conflict is a
> bug, packages should set CONFICTS so that binary package users do not
> have a terrible experience.  I believe there's PYTHON_SELF_CONFLICT
> for python stuff to make it easier.

The experience of a bad error message vs a nicer one is not to terrible
IMHO, but it is better to catch it earlier.

This was fixed minutes later in pkgsrc (PYTHON_SELF_CONFLICT), so all
that's left is to get upstream to avoid the issue or that locally.