Oh, make no mistake I use aggregator POMs for that reason. But I ALSO have
many aggregators that are just containers with <modules>.
They need their own artifactId, and their own directory (technically not a
requirement, but in practice no-one wants 2 poms in one directory).
Delany

On Tue, 8 Nov 2022 at 15:23, Mantas Gridinas <mgridi...@gmail.com> wrote:

> I think the reasoning makes sense. I would rather have a strong primitive
> using which I could build my own process, rather than fight an exact
> implementation. Current structure of an aggregator being its own packaging
> type is that primitive.
>
> Why do you consider having "glue boilerplate" bad, delany? I think its a
> great place to put all thr project's verbose declarations, such as
> dependency management, repository, plugin management blocks without
> litering the actual working modules.
>
> On Tue, Nov 8, 2022, 14:00 Łukasz Dywicki <l...@code-house.org> wrote:
>
> > Hello Delany,
> > Number of combinations in which people would like to assembly their
> > software is larger than hours maven maintainers can provide to support
> > all of them equally well. I remember that a lot of initial "hate" maven
> > received was caused by assembly plugin which was not covering all edge
> > cases people had.
> > Probably there are more reasons which come from initial assumptions made
> > when designing Maven, thus maybe someone closer to the core will explain
> > why its not possible.
> >
> > Best,
> > Łukasz
> >
> > On 7.11.2022 18:52, Delany wrote:
> > > Let me answer that with another question. How often does a project
> > include
> > > a core component/module with a bunch of other components that depend on
> > it?
> > > I would guess quite often.
> > > So why cant I have a directory structure that looks like:
> > >
> > > webserver
> > > ├── feature-a
> > > ├── feature-b
> > > ├── skin-1
> > > └── skin-2
> > >
> > > Instead I have to come up with an aggregation project to ties them all
> > > together
> > >
> > > aggregation-webserver
> > > ├── actual-webserver
> > > ├── feature-a
> > > ├── feature-b
> > > ├── skin-1
> > > └── skin-2
> > >
> > > This aggregation project is just extra glue that must be built,
> deployed,
> > > downloaded.
> > >
> > > What about the declaration of the webserver dependency. In the first
> case
> > > it could be an implicit effect of having <parent> type jar, but in the
> > > second I have to write it out in the aggregation-webserver.
> > >
> > > I just don't see the point. Why have 6 projects when 5 will do?
> > >
> > > I do see the point of aggregation poms if it was 5 unrelated
> components,
> > > but there is an obvious hierarchy here.
> > > I also see the potential pitfall of plugin configuration meant only for
> > the
> > > webserver being inherited by its children. But if I make customizations
> > > inappropriate for the children then I'll abstract them into an
> > aggregation
> > > pom when the need arises. Most plugin configuration I have applies to
> all
> > > my modules.
> > >
> > > Kind regards,
> > > Delany
> > >
> > >
> > > On Fri, 4 Nov 2022 at 16:08, Thai Le <lnthai2...@gmail.com> wrote:
> > >
> > >> I'm curious about the usecase of this.
> > >>
> > >> Thai Le
> > >>
> > >> On Fri, Nov 4, 2022, 09:30 Delany <delany.middle...@gmail.com> wrote:
> > >>
> > >>> Hi. Why can I only have pom aggregator projects and not jar
> aggregator
> > >>> projects?
> > >>> Is this still the case in Maven 4?
> > >>>
> > >>> Thanks,
> > >>> Delany
> > >>>
> > >>
> > >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: users-unsubscr...@maven.apache.org
> > For additional commands, e-mail: users-h...@maven.apache.org
> >
> >
>

Reply via email to