geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Struberg <strub...@yahoo.de>
Subject Re: [DISCUSS] implement jakarta spec apis
Date Sun, 05 May 2019 19:24:08 GMT
Btw, the limit so far has been hit whenever some internal JDK class did return some javax class.
It appears that the JDK ships many old EE parts. E.g. old common-annotations which still miss
@Priority.
Or even worse an ancient javax.transaction package. And then some javax.sql classes from the
JDK return some transaction stuff :(
Here subclassing would make perfect sense as you noted on Twitter. So we could leave the existing
javax.* classes and if we need some change then we introduce a subclass in a corresponding
 jakarta.*. This might work in many cases. But we need to check if it als works for hierarchies.
You cannot add something in a root class in jakarta as the original,classes would not extend
you. Need. A bit more time to think about that.

LieGrue,
Strub

> Am 05.05.2019 um 21:03 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:
> 
> I see, makes sense.
> 
> Personally i strongly think we dont need a strong toggle and that future compat can rely
on javax, this is what javaee was about after all.
> 
> But no issue testing things, we can even use sandbox/ for that.
> 
>> Le dim. 5 mai 2019 à 20:49, Mark Struberg <struberg@yahoo.de> a écrit :
>> Of course there will be no releases until the EF has a common understanding how to
proceed on their side. After all the main goal is compatibility amongst vendors.
>> 
>> I'd actually even would avoid to push snapshots to our repository.apache.org ...
>> 
>> This is mainly for understanding how far we come, what the limits are and what other
options we have.
>> 
>> LieGrue,
>> strub
>> 
>> 
>> > Am 05.05.2019 um 19:35 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:
>> > 
>> > Ok.
>> > 
>> > Can we agree to take this discussion back and hold any release - no issue with
snaps - until it is clarified?
>> > 
>> > Le dim. 5 mai 2019 à 18:45, Mark Struberg <struberg@yahoo.de> a écrit
:
>> > I'm not even sure whether they yet got all the necessary IP to release anything.
>> > 
>> > LieGrue,
>> > strub
>> > 
>> > 
>> > > Am 05.05.2019 um 18:39 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:
>> > > 
>> > > 
>> > > 
>> > > Le dim. 5 mai 2019 à 18:30, Mark Struberg <struberg@yahoo.de> a
écrit :
>> > > I'm not mandating jakarta in the groupId, but it should something else
than the current one.
>> > > Because otw we would have them completely mixed up in the same folder.
That's not nice.
>> > > 
>> > > Depends, that said happy to just replace specs by jakarta if it works for
you better (org.apache.geronimo.jakarta). I just dont want jakarta-specs or _spec-xxx as before,
always looked fishy and almost wrong even if I get where it comes from.
>> > > 
>> > > Btw, what is our status on having eclipse releasing api under asf2 license?
>> > > 
>> > > I dont want us to invest in something we drop like in 2 weeks and sounds
it can be for most of specs. Any page tracking that?
>> > > 
>> > > 
>> > > LieGrue
>> > > strub
>> > > 
>> > > > Am 05.05.2019 um 18:20 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:
>> > > > 
>> > > > We dont need jakarta in the gav at all.
>> > > > 
>> > > > Why not org.apache.geronimo.spec:servlet:4.0.1?
>> > > > 
>> > > > As a reminder specs means jakarta already and there id jo ambiguity
between jakarta and javaee thanks the version. 
>> > > > 
>> > > > That said if we move to git it id even physically clearer.
>> > > > 
>> > > > Finally servlet is a bad example cause owned at tomcat for apache
i think. We should absolutely stop duplicating them, it pollutes user land for no gain IMHO.
>> > > > 
>> > > > 
>> > > > 
>> > > > 
>> > > > Le dim. 5 mai 2019 à 16:28, Mark Struberg <struberg@yahoo.de>
a écrit :
>> > > > Eclipse itself probably doesn't yet have all the IP themselves. This
first needs to be clarified. Since all those legal questions have been dealt with behind closed
doors we simply have no idea.
>> > > > 
>> > > > But we do have clean-room implemented APIs under ALv2 over here at
Geronimo.
>> > > > And we can move this ourselves without having to wait for anybody.
>> > > > 
>> > > > LieGrue,
>> > > > strub
>> > > > 
>> > > > 
>> > > > > Am 05.05.2019 um 16:12 schrieb Bernd Eckenfels <ecki@zusammenkunft.net>:
>> > > > > 
>> > > > > I wonder if you need that going forward for Jakarta Specs, they
could just be distributed by Eclipse directly? Having said that, if this is not the case I
would at least remove „geronimo-“ from the artifact Id?
>> > > > > 
>> > > > > 
>> > > > > --
>> > > > > http://bernd.eckenfels.net
>> > > > >  
>> > > > > Von: Mark Struberg <struberg@yahoo.de>
>> > > > > Gesendet: Sonntag, Mai 5, 2019 4:09 PM
>> > > > > An: geronimo-dev
>> > > > > Betreff: Re: [DISCUSS] implement jakarta spec apis
>> > > > >  
>> > > > > For now I've used the following patterns: 
>> > > > > 
>> > > > > <groupId>org.apache.geronimo.jakarta-specs</groupId>

>> > > > > because specs and jakarta-specs should be in a clearly separated
folder. 
>> > > > > 
>> > > > > <artifactId>geronimo-jakarta-servlet_spec</artifactId>

>> > > > > because 'jakarta' should be in the jar name 
>> > > > > 
>> > > > > <version>4.0_1-SNAPSHOT</version> 
>> > > > > 4.0 is for servlet-4.0, 1 is the patch level. 
>> > > > > 
>> > > > > I'd NOT do a release or push to our snapshots repo until in about
2 weeks when the modus operandi is clear within the Jakarta community. 
>> > > > > 
>> > > > > LieGrue, 
>> > > > > strub 
>> > > > > 
>> > > > > 
>> > > > > 
>> > > > > > Am 05.05.2019 um 08:55 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:

>> > > > > > 
>> > > > > > Do we also want to clean our gav? Artifact=spec, major.minor
version =spec version 
>> > > > > > 
>> > > > > > Ex: org.apache.geronimo.specs:jsp:2.1.1 
>> > > > > > 
>> > > > > > Le sam. 4 mai 2019 à 21:49, Romain Manni-Bucau <rmannibucau@gmail.com>
a écrit : 
>> > > > > > 
>> > > > > > 
>> > > > > > Le sam. 4 mai 2019 à 21:44, Mark Struberg <struberg@yahoo.de>
a écrit : 
>> > > > > > The problem is that in a git repo you can only release all
at once. That means we would need to have a single git repo for each and every spec. That
will be quite many... 
>> > > > > > 
>> > > > > > No, maven plugins was a monorepo for years and then they
split. 
>> > > > > > 
>> > > > > > That said i proposed that exactly for that. At the end the
release process is more on jira dev etc, one or N repos does not compress that time. Release
prepare/perform is very fast on these repo so one or 100 is likely the same for release manager
and seems it will also enable better osgi support and probably - hopefully - enable servicemix
to stop forking the fork ;). 
>> > > > > > 
>> > > > > > I also see svn as legacy now gitbox is mainstream and people
contributing like to see their name in - I expect maybe some help for new spec as we got for
each new version. 
>> > > > > > Fixed are generally trivial there and a good reason to use
github. 
>> > > > > > 
>> > > > > > 
>> > > > > > LieGrue, 
>> > > > > > strub 
>> > > > > > 
>> > > > > > > Am 04.05.2019 um 21:35 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:

>> > > > > > > 
>> > > > > > > AFAIK we dont have limitations there and can do share
stuff outside with jgit - but it is very rare - so probably sane to unify all repo to git.
In particular since we will not do all specs probably. Cxf already moved to jakarta spec so
we dont need jaxrs stack for instance, same for cdi, bval,... So we wil reduce a lot what
we fork IMHO. 
>> > > > > > > 
>> > > > > > > Le sam. 4 mai 2019 à 21:12, Mark Struberg <struberg@yahoo.de>
a écrit : 
>> > > > > > > I’d keep that in svn because of the tons of modules.

>> > > > > > > 
>> > > > > > > Lg, 
>> > > > > > > Strub 
>> > > > > > > 
>> > > > > > > Am 04.05.2019 um 19:28 schrieb Romain Manni-Bucau <rmannibucau@gmail.com>:

>> > > > > > > 
>> > > > > > >> We mainly fork for legal reasons and defaults so
name is probably not critical while we respect module names. 
>> > > > > > >> 
>> > > > > > >> Btw do we do it in gitbox? Svn had some limitations
by the past for contributions. 
>> > > > > > >> 
>> > > > > > >> Le sam. 4 mai 2019 à 17:47, Raymond Auge <raymond.auge@liferay.com>
a écrit : 
>> > > > > > >> One thing to consider is there may be cases where
it is desirable to retain the javax API alongside some extra jakarta packages & types.

>> > > > > > >> 
>> > > > > > >> For example, for JAX-RS you may wish to add some
newly defined jakarta types (part of a new spec) which interact over the original javax API.

>> > > > > > >> 
>> > > > > > >> The result might be that "Jakarta EE REST" (a fictitious
name for next JAX-RS) might contain a subset of packages which, in combination with JAXRS
v2.1, also qualifies as "Jakarata EE Rest". 
>> > > > > > >> 
>> > > > > > >> - Ray 
>> > > > > > >> 
>> > > > > > >> On Sat, May 4, 2019 at 11:26 AM Raymond Auge <raymond.auge@liferay.com>
wrote: 
>> > > > > > >> so is this a matter of forking all the current
specs into the new namespace? Or is the intention to completely change the packages in-place?

>> > > > > > >> 
>> > > > > > >> - Ray 
>> > > > > > >> 
>> > > > > > >> On Fri, May 3, 2019 at 1:58 PM Romain Manni-Bucau
<rmannibucau@gmail.com> wrote: 
>> > > > > > >> Hmm 
>> > > > > > >> 
>> > > > > > >> My understanding was it was getting under eclipse
license as well and was fully donated but can have missed some details. 
>> > > > > > >> 
>> > > > > > >> If we cant reuse them let's just create new ones
and fix module name for others. 
>> > > > > > >> 
>> > > > > > >> specs/ is fine since it is the same for us IMHO

>> > > > > > >> 
>> > > > > > >> Le ven. 3 mai 2019 à 18:24, Mark Struberg <struberg@yahoo.de>
a écrit : 
>> > > > > > >> No, it is not the same. microprofile specs are
licensed under ALv2 and we know all the legal details. 
>> > > > > > >> For the EE specs this is by far not the same. We
don't even know exactly what parts did yet get donated by Oracle to the EF. 
>> > > > > > >> 
>> > > > > > >> LieGrue, 
>> > > > > > >> strub 
>> > > > > > >> 
>> > > > > > >> 
>> > > > > > >> > Am 03.05.2019 um 18:12 schrieb Romain Manni-Bucau
<rmannibucau@gmail.com>: 
>> > > > > > >> > 
>> > > > > > >> > Hi 
>> > > > > > >> > 
>> > > > > > >> > Idnt it the exact same as for microprofile?
So we dont do? 
>> > > > > > >> > 
>> > > > > > >> > Le ven. 3 mai 2019 à 16:21, Mark Struberg
<struberg@yahoo.de> a écrit : 
>> > > > > > >> > I've started tinkering something under specs/branches/jakarta.

>> > > > > > >> > It's wip but have to rush out for a few hours
now. 
>> > > > > > >> > Will continue later today. 
>> > > > > > >> > 
>> > > > > > >> > LieGrue, 
>> > > > > > >> > strub 
>> > > > > > >> > 
>> > > > > > >> > 
>> > > > > > >> > > Am 03.05.2019 um 15:50 schrieb Mark Struberg
<struberg@yahoo.de>: 
>> > > > > > >> > > 
>> > > > > > >> > > hi folks! 
>> > > > > > >> > > 
>> > > > > > >> > > You might have read todays post from
Mike Milinkovich. 
>> > > > > > >> > > 
>> > > > > > >> > > https://eclipse-foundation.blog/2019/05/03/jakarta-ee-java-trademarks/

>> > > > > > >> > > 
>> > > > > > >> > > It basically says that Jakarta will not
be able to change a single bit in the current spec apis under the javax.* package. 
>> > > > > > >> > > Any change has to be done in a different
package. 
>> > > > > > >> > > The Jakarta people over at Eclipse already
did some voting and the new package name will be jakarta.* 
>> > > > > > >> > > 
>> > > > > > >> > > Thus I would like to recommend to use
our IP clean geronimo-specs to setup a new project for the EE8 specs under the jakarta.* package
name. 
>> > > > > > >> > > 
>> > > > > > >> > > I'll go forward and create a branch starting
with the most important specs. 
>> > > > > > >> > > 
>> > > > > > >> > > Any feedback and help is welcome! 
>> > > > > > >> > > 
>> > > > > > >> > > LieGrue, 
>> > > > > > >> > > strub 
>> > > > > > >> > > 
>> > > > > > >> > 
>> > > > > > >> 
>> > > > > > >> 
>> > > > > > >> 
>> > > > > > >> -- 
>> > > > > > >> Raymond Augé (@rotty3000) 
>> > > > > > >> Senior Software Architect Liferay, Inc. (@Liferay)

>> > > > > > >> Board Member & EEG Co-Chair, OSGi Alliance
(@OSGiAlliance) 
>> > > > > > >> 
>> > > > > > >> 
>> > > > > > >> -- 
>> > > > > > >> Raymond Augé (@rotty3000) 
>> > > > > > >> Senior Software Architect Liferay, Inc. (@Liferay)

>> > > > > > >> Board Member & EEG Co-Chair, OSGi Alliance
(@OSGiAlliance) 
>> > > > > > 
>> > > > > 
>> > > > 
>> > > 
>> > 
>> 

Mime
View raw message