On 2019-12-05, Brian (bex) Exelbierd <bexelbie(a)redhat.com> wrote:
--===============6343409591866461936==
Content-Type: multipart/alternative; boundary="0000000000003a3ad80598f34f04"
--0000000000003a3ad80598f34f04
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
On Wed, Nov 20, 2019 at 12:13 PM Petr Pisar <ppisar(a)redhat.com> wrote:
> On 2019-11-19, John M. Harris Jr <johnmh(a)splentity.com> wrote:
> > On Tuesday, November 19, 2019 4:42:31 AM MST Petr Pisar wrote:
> >> Manual work. Random commiters skipping them.
> >
> > If your goal is to make it so that "Random commiters" are packagers,
> that's
> > going to fall flat very quickly - as they'll just throw one version of
> the
> > package in, never think about it again. That is an untenable situation.
> >
> No. Random commiter is some one who does not know that the two
> branches have to be kept in synchronization. In other words someone else
> than the maintainer. E.g. If there is a guideline change or a SONAME
> bump then somebody touches your package.
>
While not perfect, this would be an ideal use of the README file in the
master branch. This could specify things like, where to report bugs, if a
proven packager should try to contact the customary maintainer in advance,
if possible, how branches are maintained, etc.
You don't need a README. We have Bugzilla entry for each component.
Morover nobody's going to checkout master branch if he's going to commit
into a different branch. And nobody's going to read any READMEs if he's
going change hunders or thousands of packages. I'm sorry but that does
not scale.
-- Petr