Hi,
For this point "Camel v4 core and component JARs will no longer generate
OSGi MANIFEST.MF" I'm not sure that removing the generation from the
core Camel is a good thing...
regards,
François
On 30/11/2022 10:44, Claus Ibsen wrote:
Hi
On Mon, Nov 28, 2022 at 10:40 AM Jean-Baptiste Onofré <j...@nanthrax.net>
wrote:
Hi guys,
I understand that Karaf/OSGi is not in the Camel community target
anymore, and it makes sense.
I proposed a time ago to refactor the approach of Camel components for
Karaf, using special packaging (embedded the deps as private to avoid
to have bunch of SMX bundles deps), etc.
Even at Karaf, there are discussions about the next step in the
project decoupled from OSGi (see Minho).
I would split the discussion in two parts:
- In terms of the roadmap/Camel future, I don't think it's worth it
for Camel community to maintain OSGi/Karaf support anymore. It's
always possible to wrap Camel routes in an uber jar and deploy in
Karaf.
- In terms of community/maintenance, I think camel-karaf could be part
of the Karaf community. We had a similar discussion about jclouds: the
jclouds community didn't want to maintain jclouds-karaf anymore (for
the same reasons as the Camel community). The jclouds community asked
the karaf community if they were interested in maintaining and
managing jclouds-karaf. So we can do the same for camel-karaf: the
karaf community can take the lead there and maintain it.
Thoughts ?
Yes i Agree on this JB.
- Move camel-karaf to Apache Karaf as a new karaf-camel sub-project, and
let the community and committers in that project take over maintaining and
releasing this.
- For Camel v4 onwards then camel-karaf will no longer be part of Apache
Camel.
- Karaf Camel is released under a new GAV - org.apache.karaf.camel, and no
longer org.apache.camel.karaf.
- Camel v4 core and component JARs will no longer generate OSGi MANIFEST.MF
as Karaf Camel will be responsible for this (if even needed); such as JB
talks about a new way of packing and running things in Karaf.
- For Camel v3 we keep as-is and maintain and release camel-karaf until
Camel 3 is EOL
Regards
JB
On Sat, Nov 26, 2022 at 9:51 AM Andrea Cosentino <anco...@gmail.com>
wrote:
Hello
I'll come back for other questions. Let me just say that camel-karaf is
too
hard to maintain today. If it won't be released because of misalignments,
it should be aligned by some volunteers or community member and released
later. Active contributors are not really focused on Karaf runtime and we
cannot do everything. This doesn't mean we won't release camel Karaf
anymore. It only means it could be released later on. Even after the core
camel and SB part have been released.
In more than one situation aligning OSGi stuff have been really hard.
Less
and less community members are helping on the Karaf side and releasing
sometimes have been slow down by these troubles. Also OSGi have been drop
in a lot of 3rd party libraries.
So I'm +1 with this approach.
I'll continue the discussion in the next days for the other points.
Cheers
Il ven 25 nov 2022, 15:06 Nicolas Filotto <nfilo...@talend.com> ha
scritto:
Hi Claus,
That sounds like a good plan, here are the first questions that I have
in
mind:
* Why do we need to keep on releasing new LTS versions of Camel 3?
* Why not simply consider 3.20 as the last LTS version of Camel 3
and
only maintain it?
* What kind of features/improvements do you want to add to Camel 3
after releasing 3.20?
* If camel-karaf is released independently, when will it be
released?
My fear is that it will be desynchronized with Camel very quickly.
*
With 2 LTS of Camel 3 and 2 LTS of Camel 4 per year, it would mean 4
LTS
versions to support at the same time, don't you think that it is too
many?
I'm wondering if it is not a good opportunity to rethink our LTS
version
policy. Could you please remind me why we decided to have this policy
(2
LTS versions per year supported for one year)?
I would personally prefer to have:
* only one LTS version per year (or 2 if we keep on releasing LTS
versions of Camel 3) but supported for at least 2 years instead of one
otherwise Camel users are less inclined to migrate to the latest LTS
version because 1 year of support doesn't really worth the migration
effort, especially for big projects where the migration takes several
months.
* only propose milestone versions or equivalent between 2 LTS
versions
for early adopters to add more clarity on which versions are LTS.
Indeed,
for example, the next LTS version will be 3.20 while we could expect
3.22
to be the next one after 3.14 and 3.18. With this logic, instead of
releasing 3.19 and 3.20, we could have released 3.19 M1 and 3.19, it
would
then be obvious to the Camel users that only 3.19 is an LTS version as
all
final versions would then be LTS versions.
What do you think of it?
Regards,
Nicolas
________________________________
From: Claus Ibsen <claus.ib...@gmail.com>
Sent: Friday, November 25, 2022 11:42
To: dev <dev@camel.apache.org>
Subject: Camel 4 roadmap and affect on Camel 3
Hi
This is a proposal for a plan for Apache Camel 4 and how this can
affect
Camel 3.
Summary
=======
The overall scope is that the leap from Camel 3 to 4 is a lot less than
going from Camel 2 to 3.
And that we have a timebox approach where we aim for a 6 month period
of
work.
The need for Camel v4 is mainly driven by Java open source projects
migrating to jakarta APIs,
and to keep up with popular runtimes a la Spring Boot and Quarkus, and
to
jump to the next major Java version.
Goals
=====
a) Primary Goals
1) Migrate from javax -> jakarta (JEE 10)
2) Java 17 as base line
3) Spring Framework 6
4) Spring Boot 3
5) Quarkus 3
b) Release Goals
6) Release only what is ready (JEE10 / Java17 etc)
This means that Camel components that are not ready (yet) will be
dropped in a release until they are ready.
7) Release core + spring boot together
8) Release camel-karaf independently (like we do for other Camel
projects)
c) Major Goals
9) Support Java 17 features such as records, multiline strings, and
what
else
10) EIP model without JAXB dependency
11) Endpoint URI parsing (do not use java.net.URI)
12) Deprecate message.getIn()
use getMessage() instead
13) Deprecate camel-cdi
14) Deprecate/Remove MDC logging (complex and buggy and does not fit
modern
app development)
d) Minor Goals
15) Remove MEP InOptionalOut (not in use)
16) Remove JUnit 4 support
Timeline
=======
The timelines are ESTIMATES and the number of releases can vary
depending
on need and how far we are in the process
Feb 2023: Camel 4.0 milestone 1
Mar 2023: Camel 4.0 milestone 2
Apr 2023: Camel 4.0 RC1
May 2023: Camel 4.0
Aug 2023: Camel 4.1 LTS
Oct 2023: Camel 4.2
Dec 2023: Camel 4.3 LTS
The plan is to start working on Camel 4 after the next Camel 3 LTS
release,
e.g. 3.20 which is planned for next month (December 2022).
For Camel 3 then we slow down in releases and provide 2 LTS releases
per
year.
For example a scheduled could look as follows:
Dec 2022: Camel 3.20 LTS
Jun 2023: Camel 3.21 LTS
Dec 2023: Camel 3.22 LTS (last Camel v3 release, supported until Dec
2024)
???
Jun 2024: Camel 3.23 LTS (last Camel v3 release, supported until Dec
2025)
????
Each Camel 3 LTS release will likely also contain less new features and
improvements as previously, as our focus and work shifts to Camel v4
instead.
As a recipient of an email from Talend, your contact personal data
will be
on our systems. Please see our privacy notice. <
https://www.talend.com/privacy/>
--
--
François
