As a user these points would be  MAJOR concerns
1. external HTTP insecure URLs are now blocked by default

2. your builds may fail when using this new Maven release.

I would say go for version 5.0 suggesting a fresh start. A clear separation.

Leaving you enough versions to fix 3.6.3 bugs where existing project are
still compatible.

Just floating an indea.




On Sun, 28 Mar 2021, 11:07 Hervé BOUTEMY, <[email protected]> wrote:

> thank you Romain for your view
>
> current reasoning behind 3.8.0 choice is written in release notes [1]
>
> -  Why not 3.6.4?
> This is not just a bugfix as it contains three features that cause a
> change of default behavior (external HTTP insecure URLs are now blocked by
> default): your builds may fail when using this new Maven release, if you
> use now blocked repositories. Please check and eventually fix before
> upgrading.
>
> - Why not 3.7.0?
> Apache Maven 3.7.0 has been advertised in the past that it would be the
> first release where you could optionally activate the build/consumer
> feature: the version containing this feature has been renamed to 4.0.0.
> Reusing 3.7.0 might lead to confusion, hence we picked the next available
> minor version.
>
>
> I personally have a strong feeling against 3.6.4: it's not just a bugfix,
> it would cause surprises to users upgrading with full confidence.
>
> On 3.7 vs 3.8, reasoning is fully written. We skipped versions in the
> past, it's not a big deal.
>
> tm me, 3.8.0 is the best choice for users (and if they have questions why
> this version, they have 2 little answers in the release notes)
>
> Regards,
>
> Hervé
>
>
> [1]
> https://maven.apache.org/docs/3.8.0/release-notes.html#why-does-this-version-have-the-value-3-8-0
>
> Le dimanche 28 mars 2021, 11:47:11 CEST Romain Manni-Bucau a écrit :
> > Hi all,
> >
> > Before we reroll the failed 3.8.0 I'd like we discuss openly the next
> > versioning since it seems we didn't reach a consensus yet and trying to
> not
> > create too much friction for users and in the community.
> >
> > As a reminder the only feature the release will get is to prevent HTTP
> repo
> > (in favor of HTTPS ones). In that regard it is a breaking change if users
> > rely on HTTP repo but a security fix (I don't come back on the HTTP ->
> > HTTPS move IT ecosystem got recently here).
> >
> > So it seems there are multiple versioning options:
> >
> > 1. 3.6.4: seems natural since it is a security fix, enables companies to
> > get this fix respecting a project versioning policy without having to
> > upgrade and avoids us to have to maintain 3.6 + 3.7/3.8 and soon 4.x.
> > Indeed it requires a very well documented paragraph about this change and
> > how to workaround it (local proxy/mirror is a trivial one for example)
> but
> > it will be the case whatever version we pick anyway IMHO.
> > 2. 3.7.0: since it is a breaking change it can seem natural too (but has
> > the pitfall to likely require a backport in 3.6 anyway, due to the
> > versioning policies which can prevent some users to upgrade to a 3.7)
> > 3. 3.8.0: was the vote, seems the rational was that originally we
> > targetting mvnw in 3.7 and since we didn't make it 3.8 was used. Have to
> > admit I'm not sure of this reasoning more than that (cause for me if we
> > don't have a planned feature we can either try to push/wait for it or
> > postpone it but not skip a version due to that) so if anyone wants to
> > complete the reasoning here it would be great.
> >
> > Indeed my preference is for 3.6.4 which has the most advantages for
> > everyone and no additional drawbacks compared to 3.7 or 3.8 options until
> > we try to push to get mvnw in which would mean 3.7 becomes more natural
> > (and likely imply a 3.6.x maintenance version).
> >
> > Goal of this thread is to feel the overall trend and see if we can refine
> > the proposals (for example: can we drop 3.8 one and only keep 3.7 and 3.6
> > or - best - can we refine it to a single version after some exchanges).
> > If we keep a few proposals after some days, what about a vote where the
> > majority wins - we would just need to define how we count,
> > bindings/committers/all (my preference being last one indeed)?
> >
> > Romain Manni-Bucau
> > @rmannibucau <https://twitter.com/rmannibucau> |  Blog
> > <https://rmannibucau.metawerx.net/> | Old Blog
> > <http://rmannibucau.wordpress.com> | Github <
> https://github.com/rmannibucau>
> > | LinkedIn <https://www.linkedin.com/in/rmannibucau> | Book
> > <
> https://www.packtpub.com/application-development/java-ee-8-high-performance
> > >
>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>

Reply via email to