You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@bigtop.apache.org by Konstantin Boudnik <co...@apache.org> on 2016/05/04 00:12:15 UTC
Moving away from JDK7
Fellas,
as a part of 1.2 release it would be a good idea to move away from JDK/JRE 7
as far as possible. It is an EOL'ed version, which means there's no point of
beating the dead horse any further.
I thought we might start with our provisioner, which will effectively forces
us to update the build and deploy images and, once we certain that everything
still works, we'll just remove Java7 from the toolchain.
Thoughts?
Cos
Re: Moving away from JDK7
Posted by Andrew Purtell <an...@gmail.com>.
Agreed, for deployment it's fine. Everybody should be using 8 based runtimes IMHO.
> On May 3, 2016, at 6:40 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
>> On Tue, May 03, 2016 at 06:37PM, Andrew Purtell wrote:
>> As long as we maintain an option and support for building with Java 7. If
>> the plan is to move to 8 only then I'm -1
>
> Well, at least for the deployment. For instance, Ubuntu-14.04 doesn't even
> provide jdk7 anymore. So, this old stuff is really, for the lack of better
> polite word, is outdated.
>
> Cos
>
>>> On May 3, 2016, at 6:26 PM, 김영우 (Youngwoo Kim) <wa...@gmail.com> wrote:
>>>
>>> +1 for java8
>>>
>>> We are using java8 for building and running Bigtop components in production
>>> but, all of the Bigtop apps are not used. Let's find out what is needed for
>>> moving to 8
>>>
>>> Thanks,
>>> Youngwoo
>>>
>>> 2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
>>>
>>>> Fellas,
>>>>
>>>> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
>>>> 7
>>>> as far as possible. It is an EOL'ed version, which means there's no point
>>>> of
>>>> beating the dead horse any further.
>>>>
>>>> I thought we might start with our provisioner, which will effectively
>>>> forces
>>>> us to update the build and deploy images and, once we certain that
>>>> everything
>>>> still works, we'll just remove Java7 from the toolchain.
>>>>
>>>> Thoughts?
>>>> Cos
>>>>
Re: Moving away from JDK7
Posted by Olaf Flebbe <of...@oflebbe.de>.
Sorry, it was only the two latest Fedoras that only provide JDK8 .
Olaf
> Am 04.05.2016 um 18:59 schrieb Olaf Flebbe <of...@oflebbe.de>:
>
> Hi,
>
> We need jdk8 compile support (at least optional) since at least Fedora23 and opensuse42.1 does only provide jdk8 .
>
> Roman has issues with openjdk7 on arm aarch64.
>
> But we have may have blockers: See https://issues.apache.org/jira/browse/HADOOP-11090 for instance.
>
> Olaf
>
>
>> Am 04.05.2016 um 03:40 schrieb Konstantin Boudnik <co...@apache.org>:
>>
>> On Tue, May 03, 2016 at 06:37PM, Andrew Purtell wrote:
>>> As long as we maintain an option and support for building with Java 7. If
>>> the plan is to move to 8 only then I'm -1
>>
>> Well, at least for the deployment. For instance, Ubuntu-14.04 doesn't even
>> provide jdk7 anymore. So, this old stuff is really, for the lack of better
>> polite word, is outdated.
>>
>> Cos
>>
>>>> On May 3, 2016, at 6:26 PM, 김영우 (Youngwoo Kim) <wa...@gmail.com> wrote:
>>>>
>>>> +1 for java8
>>>>
>>>> We are using java8 for building and running Bigtop components in production
>>>> but, all of the Bigtop apps are not used. Let's find out what is needed for
>>>> moving to 8
>>>>
>>>> Thanks,
>>>> Youngwoo
>>>>
>>>> 2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
>>>>
>>>>> Fellas,
>>>>>
>>>>> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
>>>>> 7
>>>>> as far as possible. It is an EOL'ed version, which means there's no point
>>>>> of
>>>>> beating the dead horse any further.
>>>>>
>>>>> I thought we might start with our provisioner, which will effectively
>>>>> forces
>>>>> us to update the build and deploy images and, once we certain that
>>>>> everything
>>>>> still works, we'll just remove Java7 from the toolchain.
>>>>>
>>>>> Thoughts?
>>>>> Cos
>>>>>
>
Re: Moving away from JDK7
Posted by Olaf Flebbe <of...@oflebbe.de>.
Hi,
We need jdk8 compile support (at least optional) since at least Fedora23 and opensuse42.1 does only provide jdk8 .
Roman has issues with openjdk7 on arm aarch64.
But we have may have blockers: See https://issues.apache.org/jira/browse/HADOOP-11090 for instance.
Olaf
> Am 04.05.2016 um 03:40 schrieb Konstantin Boudnik <co...@apache.org>:
>
> On Tue, May 03, 2016 at 06:37PM, Andrew Purtell wrote:
>> As long as we maintain an option and support for building with Java 7. If
>> the plan is to move to 8 only then I'm -1
>
> Well, at least for the deployment. For instance, Ubuntu-14.04 doesn't even
> provide jdk7 anymore. So, this old stuff is really, for the lack of better
> polite word, is outdated.
>
> Cos
>
>>> On May 3, 2016, at 6:26 PM, 김영우 (Youngwoo Kim) <wa...@gmail.com> wrote:
>>>
>>> +1 for java8
>>>
>>> We are using java8 for building and running Bigtop components in production
>>> but, all of the Bigtop apps are not used. Let's find out what is needed for
>>> moving to 8
>>>
>>> Thanks,
>>> Youngwoo
>>>
>>> 2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
>>>
>>>> Fellas,
>>>>
>>>> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
>>>> 7
>>>> as far as possible. It is an EOL'ed version, which means there's no point
>>>> of
>>>> beating the dead horse any further.
>>>>
>>>> I thought we might start with our provisioner, which will effectively
>>>> forces
>>>> us to update the build and deploy images and, once we certain that
>>>> everything
>>>> still works, we'll just remove Java7 from the toolchain.
>>>>
>>>> Thoughts?
>>>> Cos
>>>>
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 06:37PM, Andrew Purtell wrote:
> As long as we maintain an option and support for building with Java 7. If
> the plan is to move to 8 only then I'm -1
Well, at least for the deployment. For instance, Ubuntu-14.04 doesn't even
provide jdk7 anymore. So, this old stuff is really, for the lack of better
polite word, is outdated.
Cos
> > On May 3, 2016, at 6:26 PM, 김영우 (Youngwoo Kim) <wa...@gmail.com> wrote:
> >
> > +1 for java8
> >
> > We are using java8 for building and running Bigtop components in production
> > but, all of the Bigtop apps are not used. Let's find out what is needed for
> > moving to 8
> >
> > Thanks,
> > Youngwoo
> >
> > 2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
> >
> >> Fellas,
> >>
> >> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
> >> 7
> >> as far as possible. It is an EOL'ed version, which means there's no point
> >> of
> >> beating the dead horse any further.
> >>
> >> I thought we might start with our provisioner, which will effectively
> >> forces
> >> us to update the build and deploy images and, once we certain that
> >> everything
> >> still works, we'll just remove Java7 from the toolchain.
> >>
> >> Thoughts?
> >> Cos
> >>
Re: Moving away from JDK7
Posted by Andrew Purtell <an...@gmail.com>.
As long as we maintain an option and support for building with Java 7. If the plan is to move to 8 only then I'm -1
> On May 3, 2016, at 6:26 PM, 김영우 (Youngwoo Kim) <wa...@gmail.com> wrote:
>
> +1 for java8
>
> We are using java8 for building and running Bigtop components in production
> but, all of the Bigtop apps are not used. Let's find out what is needed for
> moving to 8
>
> Thanks,
> Youngwoo
>
> 2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
>
>> Fellas,
>>
>> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
>> 7
>> as far as possible. It is an EOL'ed version, which means there's no point
>> of
>> beating the dead horse any further.
>>
>> I thought we might start with our provisioner, which will effectively
>> forces
>> us to update the build and deploy images and, once we certain that
>> everything
>> still works, we'll just remove Java7 from the toolchain.
>>
>> Thoughts?
>> Cos
>>
Re: Moving away from JDK7
Posted by "김영우 (Youngwoo Kim)" <wa...@gmail.com>.
+1 for java8
We are using java8 for building and running Bigtop components in production
but, all of the Bigtop apps are not used. Let's find out what is needed for
moving to 8
Thanks,
Youngwoo
2016년 5월 4일 수요일, Konstantin Boudnik<co...@apache.org>님이 작성한 메시지:
> Fellas,
>
> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
> 7
> as far as possible. It is an EOL'ed version, which means there's no point
> of
> beating the dead horse any further.
>
> I thought we might start with our provisioner, which will effectively
> forces
> us to update the build and deploy images and, once we certain that
> everything
> still works, we'll just remove Java7 from the toolchain.
>
> Thoughts?
> Cos
>
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 05:08PM, Andrew Purtell wrote:
> I think you also have to set the bootclasspath. Even so:
>
> "Additionally, even when the bootclasspath and -source/-target are all set
> appropriately for cross-compilation, compiler-internal contracts, such as
> how anonymous inner classes are compiled, may differ" (
> https://blogs.oracle.com/darcy/entry/how_to_cross_compile_for)
>
> Although, granted, that is old information. Maybe Oracle is using a
> different and better compatibility contract these days.
>
> Basically, use the oldest JDK to build the software for the oldest JRE you
> intend to support.
Yeah, I agree - it might be interesting is a pure mind-stretching exercise,
but would be too much of a pain in practice. Oh, well we digressed...
Let's at least see if we can move to Java8 in the deployment, unless there are
objections ;)
Cos
> On Tue, May 3, 2016 at 5:04 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
> > On Tue, May 03, 2016 at 04:58PM, Andrew Purtell wrote:
> > > Same.
> >
> > While I agree it won't fool the enforcer, unfortunately, but, in theory,
> > that
> > what should help to avoid the linkage error you're referring to. Well, at
> > least this was the design idea for the feature.
> >
> > Cos
> >
> > > On Tue, May 3, 2016 at 4:57 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > >
> > > > On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> > > > > No the Maven enforcer will fail the build if JDK 8 is used when JDK
> > 7 is
> > > > > required. Source level is only one component of Java cross version
> > > > > compatibility. You also have to build against the correct version of
> > the
> > > > JRE
> > > > > or even if you built with JDK 8 at source level 7 if you try to run
> > the
> > > > > software on a 7 JRE you'll see linkage errors. Especially in the
> > > > concurrency
> > > > > library.
> > > >
> > > > Sorry, I meant to say target=1.7 rather
> > > >
> > > > > > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > > > > >
> > > > > >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> > > > > >> Yep.
> > > > > >>
> > > > > >> Deployment is different though. It's my experience Java 8 is a
> > much
> > > > > >> improved version of the Java 7 runtime, if you want to look at it
> > > > that way.
> > > > > >> Build with 7, run with 8.
> > > > > >
> > > > > > Technically, speaking everything needed JDK7 could be build with
> > JDK8
> > > > if
> > > > > > source=1.7 is set, right?
> > > > > >
> > > > > > I agree, Java8 is a big step ahead and, at least, in the
> > deployment we
> > > > need to
> > > > > > switch to Java8, as you pointed out.
> > > > > >
> > > > > > Any objections on that part?
> > > > > > Cos
> > > > > >
> > > > > >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <
> > cos@apache.org>
> > > > wrote:
> > > > > >>>
> > > > > >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > > > > >>>> Some components run the Maven enforcer plugin and demand Java 7
> > as
> > > > the
> > > > > >>>> build version. Are we to patch their POMs? Otherwise we will
> > have
> > > > build
> > > > > >>>> failures with pretty core things.
> > > > > >>>
> > > > > >>> Damn.... So, I guess we are stuck with this old dusty piece until
> > > > the most
> > > > > >>> orthodox of the upstreams are married to it, right?
> > > > > >>>
> > > > > >>> Cos
> > > > > >>>
> > > > > >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <
> > cos@apache.org
> > > > >
> > > > > >>>> wrote:
> > > > > >>>>
> > > > > >>>>> Fellas,
> > > > > >>>>>
> > > > > >>>>> as a part of 1.2 release it would be a good idea to move away
> > from
> > > > > >>> JDK/JRE
> > > > > >>>>> 7
> > > > > >>>>> as far as possible. It is an EOL'ed version, which means
> > there's no
> > > > > >>> point
> > > > > >>>>> of
> > > > > >>>>> beating the dead horse any further.
> > > > > >>>>>
> > > > > >>>>> I thought we might start with our provisioner, which will
> > > > effectively
> > > > > >>>>> forces
> > > > > >>>>> us to update the build and deploy images and, once we certain
> > that
> > > > > >>>>> everything
> > > > > >>>>> still works, we'll just remove Java7 from the toolchain.
> > > > > >>>>>
> > > > > >>>>> Thoughts?
> > > > > >>>>> Cos
> > > > > >>>>
> > > > > >>>>
> > > > > >>>>
> > > > > >>>> --
> > > > > >>>> Best regards,
> > > > > >>>>
> > > > > >>>> - Andy
> > > > > >>>>
> > > > > >>>> Problems worthy of attack prove their worth by hitting back. -
> > Piet
> > > > Hein
> > > > > >>>> (via Tom White)
> > > > > >>
> > > > > >>
> > > > > >>
> > > > > >> --
> > > > > >> Best regards,
> > > > > >>
> > > > > >> - Andy
> > > > > >>
> > > > > >> Problems worthy of attack prove their worth by hitting back. -
> > Piet
> > > > Hein
> > > > > >> (via Tom White)
> > > >
> > >
> > >
> > >
> > > --
> > > Best regards,
> > >
> > > - Andy
> > >
> > > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > > (via Tom White)
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
Re: Moving away from JDK7
Posted by Jonathan Kelly <jo...@gmail.com>.
Andrew,
Which Bigtop apps are affected by this Maven enforcer plugin problem that
forces JDK 7? I wasn't aware of that, but I haven't yet tried building
Bigtop with JDK 8. I've had success doing the "build with 7, run with 8"
approach though, but of course I haven't tried this with all of the
Bigtop-supported apps yet.
~ Jonathan
On Tue, May 3, 2016 at 5:08 PM Andrew Purtell <ap...@apache.org> wrote:
> I think you also have to set the bootclasspath. Even so:
>
> "Additionally, even when the bootclasspath and -source/-target are all set
> appropriately for cross-compilation, compiler-internal contracts, such as
> how anonymous inner classes are compiled, may differ" (
> https://blogs.oracle.com/darcy/entry/how_to_cross_compile_for)
>
> Although, granted, that is old information. Maybe Oracle is using a
> different and better compatibility contract these days.
>
> Basically, use the oldest JDK to build the software for the oldest JRE you
> intend to support.
>
>
>
> On Tue, May 3, 2016 at 5:04 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
> > On Tue, May 03, 2016 at 04:58PM, Andrew Purtell wrote:
> > > Same.
> >
> > While I agree it won't fool the enforcer, unfortunately, but, in theory,
> > that
> > what should help to avoid the linkage error you're referring to. Well, at
> > least this was the design idea for the feature.
> >
> > Cos
> >
> > > On Tue, May 3, 2016 at 4:57 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > >
> > > > On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> > > > > No the Maven enforcer will fail the build if JDK 8 is used when JDK
> > 7 is
> > > > > required. Source level is only one component of Java cross version
> > > > > compatibility. You also have to build against the correct version
> of
> > the
> > > > JRE
> > > > > or even if you built with JDK 8 at source level 7 if you try to run
> > the
> > > > > software on a 7 JRE you'll see linkage errors. Especially in the
> > > > concurrency
> > > > > library.
> > > >
> > > > Sorry, I meant to say target=1.7 rather
> > > >
> > > > > > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > > > > >
> > > > > >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> > > > > >> Yep.
> > > > > >>
> > > > > >> Deployment is different though. It's my experience Java 8 is a
> > much
> > > > > >> improved version of the Java 7 runtime, if you want to look at
> it
> > > > that way.
> > > > > >> Build with 7, run with 8.
> > > > > >
> > > > > > Technically, speaking everything needed JDK7 could be build with
> > JDK8
> > > > if
> > > > > > source=1.7 is set, right?
> > > > > >
> > > > > > I agree, Java8 is a big step ahead and, at least, in the
> > deployment we
> > > > need to
> > > > > > switch to Java8, as you pointed out.
> > > > > >
> > > > > > Any objections on that part?
> > > > > > Cos
> > > > > >
> > > > > >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <
> > cos@apache.org>
> > > > wrote:
> > > > > >>>
> > > > > >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > > > > >>>> Some components run the Maven enforcer plugin and demand Java
> 7
> > as
> > > > the
> > > > > >>>> build version. Are we to patch their POMs? Otherwise we will
> > have
> > > > build
> > > > > >>>> failures with pretty core things.
> > > > > >>>
> > > > > >>> Damn.... So, I guess we are stuck with this old dusty piece
> until
> > > > the most
> > > > > >>> orthodox of the upstreams are married to it, right?
> > > > > >>>
> > > > > >>> Cos
> > > > > >>>
> > > > > >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <
> > cos@apache.org
> > > > >
> > > > > >>>> wrote:
> > > > > >>>>
> > > > > >>>>> Fellas,
> > > > > >>>>>
> > > > > >>>>> as a part of 1.2 release it would be a good idea to move away
> > from
> > > > > >>> JDK/JRE
> > > > > >>>>> 7
> > > > > >>>>> as far as possible. It is an EOL'ed version, which means
> > there's no
> > > > > >>> point
> > > > > >>>>> of
> > > > > >>>>> beating the dead horse any further.
> > > > > >>>>>
> > > > > >>>>> I thought we might start with our provisioner, which will
> > > > effectively
> > > > > >>>>> forces
> > > > > >>>>> us to update the build and deploy images and, once we certain
> > that
> > > > > >>>>> everything
> > > > > >>>>> still works, we'll just remove Java7 from the toolchain.
> > > > > >>>>>
> > > > > >>>>> Thoughts?
> > > > > >>>>> Cos
> > > > > >>>>
> > > > > >>>>
> > > > > >>>>
> > > > > >>>> --
> > > > > >>>> Best regards,
> > > > > >>>>
> > > > > >>>> - Andy
> > > > > >>>>
> > > > > >>>> Problems worthy of attack prove their worth by hitting back. -
> > Piet
> > > > Hein
> > > > > >>>> (via Tom White)
> > > > > >>
> > > > > >>
> > > > > >>
> > > > > >> --
> > > > > >> Best regards,
> > > > > >>
> > > > > >> - Andy
> > > > > >>
> > > > > >> Problems worthy of attack prove their worth by hitting back. -
> > Piet
> > > > Hein
> > > > > >> (via Tom White)
> > > >
> > >
> > >
> > >
> > > --
> > > Best regards,
> > >
> > > - Andy
> > >
> > > Problems worthy of attack prove their worth by hitting back. - Piet
> Hein
> > > (via Tom White)
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
>
Re: Moving away from JDK7
Posted by Andrew Purtell <ap...@apache.org>.
I think you also have to set the bootclasspath. Even so:
"Additionally, even when the bootclasspath and -source/-target are all set
appropriately for cross-compilation, compiler-internal contracts, such as
how anonymous inner classes are compiled, may differ" (
https://blogs.oracle.com/darcy/entry/how_to_cross_compile_for)
Although, granted, that is old information. Maybe Oracle is using a
different and better compatibility contract these days.
Basically, use the oldest JDK to build the software for the oldest JRE you
intend to support.
On Tue, May 3, 2016 at 5:04 PM, Konstantin Boudnik <co...@apache.org> wrote:
> On Tue, May 03, 2016 at 04:58PM, Andrew Purtell wrote:
> > Same.
>
> While I agree it won't fool the enforcer, unfortunately, but, in theory,
> that
> what should help to avoid the linkage error you're referring to. Well, at
> least this was the design idea for the feature.
>
> Cos
>
> > On Tue, May 3, 2016 at 4:57 PM, Konstantin Boudnik <co...@apache.org>
> wrote:
> >
> > > On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> > > > No the Maven enforcer will fail the build if JDK 8 is used when JDK
> 7 is
> > > > required. Source level is only one component of Java cross version
> > > > compatibility. You also have to build against the correct version of
> the
> > > JRE
> > > > or even if you built with JDK 8 at source level 7 if you try to run
> the
> > > > software on a 7 JRE you'll see linkage errors. Especially in the
> > > concurrency
> > > > library.
> > >
> > > Sorry, I meant to say target=1.7 rather
> > >
> > > > > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org>
> wrote:
> > > > >
> > > > >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> > > > >> Yep.
> > > > >>
> > > > >> Deployment is different though. It's my experience Java 8 is a
> much
> > > > >> improved version of the Java 7 runtime, if you want to look at it
> > > that way.
> > > > >> Build with 7, run with 8.
> > > > >
> > > > > Technically, speaking everything needed JDK7 could be build with
> JDK8
> > > if
> > > > > source=1.7 is set, right?
> > > > >
> > > > > I agree, Java8 is a big step ahead and, at least, in the
> deployment we
> > > need to
> > > > > switch to Java8, as you pointed out.
> > > > >
> > > > > Any objections on that part?
> > > > > Cos
> > > > >
> > > > >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <
> cos@apache.org>
> > > wrote:
> > > > >>>
> > > > >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > > > >>>> Some components run the Maven enforcer plugin and demand Java 7
> as
> > > the
> > > > >>>> build version. Are we to patch their POMs? Otherwise we will
> have
> > > build
> > > > >>>> failures with pretty core things.
> > > > >>>
> > > > >>> Damn.... So, I guess we are stuck with this old dusty piece until
> > > the most
> > > > >>> orthodox of the upstreams are married to it, right?
> > > > >>>
> > > > >>> Cos
> > > > >>>
> > > > >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <
> cos@apache.org
> > > >
> > > > >>>> wrote:
> > > > >>>>
> > > > >>>>> Fellas,
> > > > >>>>>
> > > > >>>>> as a part of 1.2 release it would be a good idea to move away
> from
> > > > >>> JDK/JRE
> > > > >>>>> 7
> > > > >>>>> as far as possible. It is an EOL'ed version, which means
> there's no
> > > > >>> point
> > > > >>>>> of
> > > > >>>>> beating the dead horse any further.
> > > > >>>>>
> > > > >>>>> I thought we might start with our provisioner, which will
> > > effectively
> > > > >>>>> forces
> > > > >>>>> us to update the build and deploy images and, once we certain
> that
> > > > >>>>> everything
> > > > >>>>> still works, we'll just remove Java7 from the toolchain.
> > > > >>>>>
> > > > >>>>> Thoughts?
> > > > >>>>> Cos
> > > > >>>>
> > > > >>>>
> > > > >>>>
> > > > >>>> --
> > > > >>>> Best regards,
> > > > >>>>
> > > > >>>> - Andy
> > > > >>>>
> > > > >>>> Problems worthy of attack prove their worth by hitting back. -
> Piet
> > > Hein
> > > > >>>> (via Tom White)
> > > > >>
> > > > >>
> > > > >>
> > > > >> --
> > > > >> Best regards,
> > > > >>
> > > > >> - Andy
> > > > >>
> > > > >> Problems worthy of attack prove their worth by hitting back. -
> Piet
> > > Hein
> > > > >> (via Tom White)
> > >
> >
> >
> >
> > --
> > Best regards,
> >
> > - Andy
> >
> > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > (via Tom White)
>
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 04:58PM, Andrew Purtell wrote:
> Same.
While I agree it won't fool the enforcer, unfortunately, but, in theory, that
what should help to avoid the linkage error you're referring to. Well, at
least this was the design idea for the feature.
Cos
> On Tue, May 3, 2016 at 4:57 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
> > On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> > > No the Maven enforcer will fail the build if JDK 8 is used when JDK 7 is
> > > required. Source level is only one component of Java cross version
> > > compatibility. You also have to build against the correct version of the
> > JRE
> > > or even if you built with JDK 8 at source level 7 if you try to run the
> > > software on a 7 JRE you'll see linkage errors. Especially in the
> > concurrency
> > > library.
> >
> > Sorry, I meant to say target=1.7 rather
> >
> > > > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org> wrote:
> > > >
> > > >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> > > >> Yep.
> > > >>
> > > >> Deployment is different though. It's my experience Java 8 is a much
> > > >> improved version of the Java 7 runtime, if you want to look at it
> > that way.
> > > >> Build with 7, run with 8.
> > > >
> > > > Technically, speaking everything needed JDK7 could be build with JDK8
> > if
> > > > source=1.7 is set, right?
> > > >
> > > > I agree, Java8 is a big step ahead and, at least, in the deployment we
> > need to
> > > > switch to Java8, as you pointed out.
> > > >
> > > > Any objections on that part?
> > > > Cos
> > > >
> > > >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > > >>>
> > > >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > > >>>> Some components run the Maven enforcer plugin and demand Java 7 as
> > the
> > > >>>> build version. Are we to patch their POMs? Otherwise we will have
> > build
> > > >>>> failures with pretty core things.
> > > >>>
> > > >>> Damn.... So, I guess we are stuck with this old dusty piece until
> > the most
> > > >>> orthodox of the upstreams are married to it, right?
> > > >>>
> > > >>> Cos
> > > >>>
> > > >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <cos@apache.org
> > >
> > > >>>> wrote:
> > > >>>>
> > > >>>>> Fellas,
> > > >>>>>
> > > >>>>> as a part of 1.2 release it would be a good idea to move away from
> > > >>> JDK/JRE
> > > >>>>> 7
> > > >>>>> as far as possible. It is an EOL'ed version, which means there's no
> > > >>> point
> > > >>>>> of
> > > >>>>> beating the dead horse any further.
> > > >>>>>
> > > >>>>> I thought we might start with our provisioner, which will
> > effectively
> > > >>>>> forces
> > > >>>>> us to update the build and deploy images and, once we certain that
> > > >>>>> everything
> > > >>>>> still works, we'll just remove Java7 from the toolchain.
> > > >>>>>
> > > >>>>> Thoughts?
> > > >>>>> Cos
> > > >>>>
> > > >>>>
> > > >>>>
> > > >>>> --
> > > >>>> Best regards,
> > > >>>>
> > > >>>> - Andy
> > > >>>>
> > > >>>> Problems worthy of attack prove their worth by hitting back. - Piet
> > Hein
> > > >>>> (via Tom White)
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> Best regards,
> > > >>
> > > >> - Andy
> > > >>
> > > >> Problems worthy of attack prove their worth by hitting back. - Piet
> > Hein
> > > >> (via Tom White)
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
Re: Moving away from JDK7
Posted by Andrew Purtell <ap...@apache.org>.
Same.
On Tue, May 3, 2016 at 4:57 PM, Konstantin Boudnik <co...@apache.org> wrote:
> On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> > No the Maven enforcer will fail the build if JDK 8 is used when JDK 7 is
> > required. Source level is only one component of Java cross version
> > compatibility. You also have to build against the correct version of the
> JRE
> > or even if you built with JDK 8 at source level 7 if you try to run the
> > software on a 7 JRE you'll see linkage errors. Especially in the
> concurrency
> > library.
>
> Sorry, I meant to say target=1.7 rather
>
> > > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org> wrote:
> > >
> > >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> > >> Yep.
> > >>
> > >> Deployment is different though. It's my experience Java 8 is a much
> > >> improved version of the Java 7 runtime, if you want to look at it
> that way.
> > >> Build with 7, run with 8.
> > >
> > > Technically, speaking everything needed JDK7 could be build with JDK8
> if
> > > source=1.7 is set, right?
> > >
> > > I agree, Java8 is a big step ahead and, at least, in the deployment we
> need to
> > > switch to Java8, as you pointed out.
> > >
> > > Any objections on that part?
> > > Cos
> > >
> > >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org>
> wrote:
> > >>>
> > >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > >>>> Some components run the Maven enforcer plugin and demand Java 7 as
> the
> > >>>> build version. Are we to patch their POMs? Otherwise we will have
> build
> > >>>> failures with pretty core things.
> > >>>
> > >>> Damn.... So, I guess we are stuck with this old dusty piece until
> the most
> > >>> orthodox of the upstreams are married to it, right?
> > >>>
> > >>> Cos
> > >>>
> > >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <cos@apache.org
> >
> > >>>> wrote:
> > >>>>
> > >>>>> Fellas,
> > >>>>>
> > >>>>> as a part of 1.2 release it would be a good idea to move away from
> > >>> JDK/JRE
> > >>>>> 7
> > >>>>> as far as possible. It is an EOL'ed version, which means there's no
> > >>> point
> > >>>>> of
> > >>>>> beating the dead horse any further.
> > >>>>>
> > >>>>> I thought we might start with our provisioner, which will
> effectively
> > >>>>> forces
> > >>>>> us to update the build and deploy images and, once we certain that
> > >>>>> everything
> > >>>>> still works, we'll just remove Java7 from the toolchain.
> > >>>>>
> > >>>>> Thoughts?
> > >>>>> Cos
> > >>>>
> > >>>>
> > >>>>
> > >>>> --
> > >>>> Best regards,
> > >>>>
> > >>>> - Andy
> > >>>>
> > >>>> Problems worthy of attack prove their worth by hitting back. - Piet
> Hein
> > >>>> (via Tom White)
> > >>
> > >>
> > >>
> > >> --
> > >> Best regards,
> > >>
> > >> - Andy
> > >>
> > >> Problems worthy of attack prove their worth by hitting back. - Piet
> Hein
> > >> (via Tom White)
>
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 04:50PM, Andrew Purtell wrote:
> No the Maven enforcer will fail the build if JDK 8 is used when JDK 7 is
> required. Source level is only one component of Java cross version
> compatibility. You also have to build against the correct version of the JRE
> or even if you built with JDK 8 at source level 7 if you try to run the
> software on a 7 JRE you'll see linkage errors. Especially in the concurrency
> library.
Sorry, I meant to say target=1.7 rather
> > On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org> wrote:
> >
> >> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> >> Yep.
> >>
> >> Deployment is different though. It's my experience Java 8 is a much
> >> improved version of the Java 7 runtime, if you want to look at it that way.
> >> Build with 7, run with 8.
> >
> > Technically, speaking everything needed JDK7 could be build with JDK8 if
> > source=1.7 is set, right?
> >
> > I agree, Java8 is a big step ahead and, at least, in the deployment we need to
> > switch to Java8, as you pointed out.
> >
> > Any objections on that part?
> > Cos
> >
> >>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org> wrote:
> >>>
> >>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> >>>> Some components run the Maven enforcer plugin and demand Java 7 as the
> >>>> build version. Are we to patch their POMs? Otherwise we will have build
> >>>> failures with pretty core things.
> >>>
> >>> Damn.... So, I guess we are stuck with this old dusty piece until the most
> >>> orthodox of the upstreams are married to it, right?
> >>>
> >>> Cos
> >>>
> >>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org>
> >>>> wrote:
> >>>>
> >>>>> Fellas,
> >>>>>
> >>>>> as a part of 1.2 release it would be a good idea to move away from
> >>> JDK/JRE
> >>>>> 7
> >>>>> as far as possible. It is an EOL'ed version, which means there's no
> >>> point
> >>>>> of
> >>>>> beating the dead horse any further.
> >>>>>
> >>>>> I thought we might start with our provisioner, which will effectively
> >>>>> forces
> >>>>> us to update the build and deploy images and, once we certain that
> >>>>> everything
> >>>>> still works, we'll just remove Java7 from the toolchain.
> >>>>>
> >>>>> Thoughts?
> >>>>> Cos
> >>>>
> >>>>
> >>>>
> >>>> --
> >>>> Best regards,
> >>>>
> >>>> - Andy
> >>>>
> >>>> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> >>>> (via Tom White)
> >>
> >>
> >>
> >> --
> >> Best regards,
> >>
> >> - Andy
> >>
> >> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> >> (via Tom White)
Re: Moving away from JDK7
Posted by Andrew Purtell <an...@gmail.com>.
No the Maven enforcer will fail the build if JDK 8 is used when JDK 7 is required. Source level is only one component of Java cross version compatibility. You also have to build against the correct version of the JRE or even if you built with JDK 8 at source level 7 if you try to run the software on a 7 JRE you'll see linkage errors. Especially in the concurrency library.
> On May 3, 2016, at 4:40 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
>> On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
>> Yep.
>>
>> Deployment is different though. It's my experience Java 8 is a much
>> improved version of the Java 7 runtime, if you want to look at it that way.
>> Build with 7, run with 8.
>
> Technically, speaking everything needed JDK7 could be build with JDK8 if
> source=1.7 is set, right?
>
> I agree, Java8 is a big step ahead and, at least, in the deployment we need to
> switch to Java8, as you pointed out.
>
> Any objections on that part?
> Cos
>
>>> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org> wrote:
>>>
>>>> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
>>>> Some components run the Maven enforcer plugin and demand Java 7 as the
>>>> build version. Are we to patch their POMs? Otherwise we will have build
>>>> failures with pretty core things.
>>>
>>> Damn.... So, I guess we are stuck with this old dusty piece until the most
>>> orthodox of the upstreams are married to it, right?
>>>
>>> Cos
>>>
>>>>> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org>
>>>> wrote:
>>>>
>>>>> Fellas,
>>>>>
>>>>> as a part of 1.2 release it would be a good idea to move away from
>>> JDK/JRE
>>>>> 7
>>>>> as far as possible. It is an EOL'ed version, which means there's no
>>> point
>>>>> of
>>>>> beating the dead horse any further.
>>>>>
>>>>> I thought we might start with our provisioner, which will effectively
>>>>> forces
>>>>> us to update the build and deploy images and, once we certain that
>>>>> everything
>>>>> still works, we'll just remove Java7 from the toolchain.
>>>>>
>>>>> Thoughts?
>>>>> Cos
>>>>
>>>>
>>>>
>>>> --
>>>> Best regards,
>>>>
>>>> - Andy
>>>>
>>>> Problems worthy of attack prove their worth by hitting back. - Piet Hein
>>>> (via Tom White)
>>
>>
>>
>> --
>> Best regards,
>>
>> - Andy
>>
>> Problems worthy of attack prove their worth by hitting back. - Piet Hein
>> (via Tom White)
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 04:37PM, Andrew Purtell wrote:
> Yep.
>
> Deployment is different though. It's my experience Java 8 is a much
> improved version of the Java 7 runtime, if you want to look at it that way.
> Build with 7, run with 8.
Technically, speaking everything needed JDK7 could be build with JDK8 if
source=1.7 is set, right?
I agree, Java8 is a big step ahead and, at least, in the deployment we need to
switch to Java8, as you pointed out.
Any objections on that part?
Cos
> On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
> > On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > > Some components run the Maven enforcer plugin and demand Java 7 as the
> > > build version. Are we to patch their POMs? Otherwise we will have build
> > > failures with pretty core things.
> >
> > Damn.... So, I guess we are stuck with this old dusty piece until the most
> > orthodox of the upstreams are married to it, right?
> >
> > Cos
> >
> > > On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org>
> > wrote:
> > >
> > > > Fellas,
> > > >
> > > > as a part of 1.2 release it would be a good idea to move away from
> > JDK/JRE
> > > > 7
> > > > as far as possible. It is an EOL'ed version, which means there's no
> > point
> > > > of
> > > > beating the dead horse any further.
> > > >
> > > > I thought we might start with our provisioner, which will effectively
> > > > forces
> > > > us to update the build and deploy images and, once we certain that
> > > > everything
> > > > still works, we'll just remove Java7 from the toolchain.
> > > >
> > > > Thoughts?
> > > > Cos
> > > >
> > >
> > >
> > >
> > > --
> > > Best regards,
> > >
> > > - Andy
> > >
> > > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > > (via Tom White)
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
Re: Moving away from JDK7
Posted by Andrew Purtell <ap...@apache.org>.
Yep.
Deployment is different though. It's my experience Java 8 is a much
improved version of the Java 7 runtime, if you want to look at it that way.
Build with 7, run with 8.
On Tue, May 3, 2016 at 4:36 PM, Konstantin Boudnik <co...@apache.org> wrote:
> On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> > Some components run the Maven enforcer plugin and demand Java 7 as the
> > build version. Are we to patch their POMs? Otherwise we will have build
> > failures with pretty core things.
>
> Damn.... So, I guess we are stuck with this old dusty piece until the most
> orthodox of the upstreams are married to it, right?
>
> Cos
>
> > On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org>
> wrote:
> >
> > > Fellas,
> > >
> > > as a part of 1.2 release it would be a good idea to move away from
> JDK/JRE
> > > 7
> > > as far as possible. It is an EOL'ed version, which means there's no
> point
> > > of
> > > beating the dead horse any further.
> > >
> > > I thought we might start with our provisioner, which will effectively
> > > forces
> > > us to update the build and deploy images and, once we certain that
> > > everything
> > > still works, we'll just remove Java7 from the toolchain.
> > >
> > > Thoughts?
> > > Cos
> > >
> >
> >
> >
> > --
> > Best regards,
> >
> > - Andy
> >
> > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > (via Tom White)
>
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)
Re: Moving away from JDK7
Posted by Konstantin Boudnik <co...@apache.org>.
On Tue, May 03, 2016 at 04:30PM, Andrew Purtell wrote:
> Some components run the Maven enforcer plugin and demand Java 7 as the
> build version. Are we to patch their POMs? Otherwise we will have build
> failures with pretty core things.
Damn.... So, I guess we are stuck with this old dusty piece until the most
orthodox of the upstreams are married to it, right?
Cos
> On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org> wrote:
>
> > Fellas,
> >
> > as a part of 1.2 release it would be a good idea to move away from JDK/JRE
> > 7
> > as far as possible. It is an EOL'ed version, which means there's no point
> > of
> > beating the dead horse any further.
> >
> > I thought we might start with our provisioner, which will effectively
> > forces
> > us to update the build and deploy images and, once we certain that
> > everything
> > still works, we'll just remove Java7 from the toolchain.
> >
> > Thoughts?
> > Cos
> >
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)
Re: Moving away from JDK7
Posted by Andrew Purtell <ap...@apache.org>.
Some components run the Maven enforcer plugin and demand Java 7 as the
build version. Are we to patch their POMs? Otherwise we will have build
failures with pretty core things.
On Tue, May 3, 2016 at 3:12 PM, Konstantin Boudnik <co...@apache.org> wrote:
> Fellas,
>
> as a part of 1.2 release it would be a good idea to move away from JDK/JRE
> 7
> as far as possible. It is an EOL'ed version, which means there's no point
> of
> beating the dead horse any further.
>
> I thought we might start with our provisioner, which will effectively
> forces
> us to update the build and deploy images and, once we certain that
> everything
> still works, we'll just remove Java7 from the toolchain.
>
> Thoughts?
> Cos
>
--
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)