Re: pkglint & bulk-test-*

[David Brownlee <abs%absd.org@localhost>]

On Tue, 29 Oct 2024 at 18:16, Roland Illig <roland.illig%gmx.de@localhost> wrote:
>
> Am 29.10.2024 um 17:49 schrieb David Brownlee:
> > I'm wondering about guiding people towards the meta-pkgs/bulk-test-*
> > packages when updating a package.
>
> What exact rule are you thinking about? For example, when updating
> editors/abiword, would there be any benefit building
> meta-pkgs/bulk-test-boost?

No, I had envisioned it as only for updates to the key package
(devel/boost-libs in this case). Updating editors/abiword and breaking
a dependency is a much lower risk than doing the same for
devel/boost-libs

> > Possibly a special variable set in the Makefile for the base package a
> > bulk-test-foo package is intended to test, with pkglint
> > unconditionally showing a message strongly suggesting building the
> > relevant bulk-test-foo before committing?
>
> What packages would be affected by this? I'd think the people who dare
> updating the packages mentioned directly in meta-pkgs/bulk-test-*/DESCR
> are only a handful, and I assume that they already know what they are doing.

This would be to try to remind people who should already know, and
help less familiar people avoid accidents.

> > We could just always put a comment in the relevant Makefile, but I was
> > hoping for something a little more "in your face" for someone updating
> > a package and running pkglint - similar to the way MAINTAINER is now
>
> pkglint also looks at the comments in a package Makefile (for finding
> rationales and commented variable assignments), so that would work as well.

Thanks

David