Mailing-List: contact dev-help@spark.apache.org; run by ezmlm
Precedence: bulk
MIME-Version: 1.0
References: <CAAswR-6iT9ZVakdC+PdLAc92=YWp8Mbv+bgoUdsgpZivF3=Brg@mail.gmail.com>
 <CAJ5_7ELDP4q_1KNuqAAy=MMbxn0LvcCyhLJwjqQJFwR_H05=7g@mail.gmail.com>
 <CAJLcJd9e5CvfM5fgz1F+T420gWF4WUTN8rAK6kOk5UXJwxU9Zw@mail.gmail.com>
 <DM2PR0401MB14215728A646089D1A36A12E91090@DM2PR0401MB1421.namprd04.prod.outlook.com>
 <CAAswR-75iu1YFdsZWmQ4nXuJ+NKiFw4LBCXeZOSMoO=QOVTgPA@mail.gmail.com>
 <CAJLcJd-6xPEtSJBNBr-WfSNCicPuo90qjvS04xLei7GWgN4bnQ@mail.gmail.com>
 <CAJLcJd9B6E_Y1_zt-=2VMjADBGNG+a22KWSmzrqQKZi86b9qQg@mail.gmail.com>
 <CAJLcJd8BgPu00Fi4M0hAGJLDC0FH4oy9n6ciOfaN2TYa_LXNXg@mail.gmail.com>
 <CAEYYnxZHNR3wuqc6Xd6HWRZ8pHeHCu1T+edS2tc_GXJjvvjXtw@mail.gmail.com>
 <CAO4re1mwNmP2vZzNSkXtUtiFnySqRmAyT7wusgBHMxdrsrLw6Q@mail.gmail.com>
 <CAEYYnxbX7kYTamgn3yvp4TGtpuqRL-DZJUwU7yRx1dxZtpbc4A@mail.gmail.com>
 <CAJLcJd_S3oiXZdF6DBcWEDhtVF5tiSH+8WvAJpHOahUNHOkRBQ@mail.gmail.com>
 <CAAswR-5+HQCQtwpt035s1VVHqV2kCB52m3bKK_YmYAZ96hHc7w@mail.gmail.com>
 <CAJLcJd_6qB5vuzDBFAL0QxLV5xJ7jfSNvLBL2ZgPsEP8QTRXiQ@mail.gmail.com>
 <CAJLcJd9nQAzoxPTUJdKzP+-4aCfnABksjWZScZk_8R0LA25Hvg@mail.gmail.com>
 <CAJLcJd_JxfDDYkpXYpQX-pzwYDnQV89qfhjoVbBFuaw-KsoQXw@mail.gmail.com> <CAO4re1ms6=kKM+nAUDQSeuMna0dB7d0Mzrb66=BoeBK4ieGx3A@mail.gmail.com>
In-Reply-To: <CAO4re1ms6=kKM+nAUDQSeuMna0dB7d0Mzrb66=BoeBK4ieGx3A@mail.gmail.com>
From: Sean Owen <sowen@cloudera.com>
Date: Sat, 15 Apr 2017 09:33:32 +0000
Message-ID: <CAMAsSdLJt2-CrgJanq7N4hvzoiKFKS0kxaShE4NvMJ5wh-ZCSg@mail.gmail.com>
Subject: Re: [VOTE] Apache Spark 2.1.1 (RC2)
To: Ryan Blue <rblue@netflix.com.invalid>, Holden Karau <holden@pigscanfly.ca>
Cc: Michael Armbrust <michael@databricks.com>, Pedro Rodriguez <ski.rodriguez@gmail.com>,
	DB Tsai <dbtsai@dbtsai.com>, Felix Cheung <felixcheung_m@hotmail.com>,
	"dev@spark.apache.org" <dev@spark.apache.org>
Content-Type: multipart/alternative; boundary=f4030437a6f40b48c7054d314143
archived-at: Sat, 15 Apr 2017 09:33:49 -0000

--f4030437a6f40b48c7054d314143
Content-Type: text/plain; charset=UTF-8

I don't think this is an example of Java 8 javadoc being more strict; it is
not finding classes, not complaining about syntax.
(Hyukjin cleaned up all of the javadoc 8 errors in master, and they're
different and much more extensive!)

It wouldn't necessarily break anything to build with Java 8 because it'll
still emit Java 7 bytecode, etc.

That said, it may very well be that it is somehow due to Java 7 vs 8, and
is probably best to stick to 1.7 in the release build.

On Sat, Apr 15, 2017 at 1:38 AM Ryan Blue <rblue@netflix.com.invalid> wrote:

> I've hit this before, where Javadoc for 1.8 is much more strict than 1.7.
>
> I think we should definitely use Java 1.7 for the release if we used it
> for the previous releases in the 2.1 line. We don't want to break java 1.7
> users in a patch release.
>
> rb
>
> On Fri, Apr 14, 2017 at 5:21 PM, Holden Karau <holden@pigscanfly.ca>
> wrote:
>
>> Ok and with a bit more digging between RC2 and RC3 we apparently switched
>> which JVM we are building the docs with.
>>
>> The relevant side by side diff of the build logs (
>> https://amplab.cs.berkeley.edu/jenkins/view/Spark%20Release/job/spark-release-docs/60/consoleFull
>>
>> https://amplab.cs.berkeley.edu/jenkins/view/Spark%20Release/job/spark-release-docs/59/consoleFull
>> ):
>>
>> HEAD is now at 2ed19cf... Preparing Spark release v2.1.1-rc3  | HEAD is
>> now at 02b165d... Preparing Spark release v2.1.1-rc2
>> Checked out Spark git hash 2ed19cf                            | Checked
>> out Spark git hash 02b165d
>> Building Spark docs                                             Building
>> Spark docs
>> Configuration file: /home/jenkins/workspace/spark-release-doc
>> Configuration file: /home/jenkins/workspace/spark-release-doc
>> Moving to project root and building API docs.                   Moving to
>> project root and building API docs.
>> Running 'build/sbt -Pkinesis-asl clean compile unidoc' from /   Running
>> 'build/sbt -Pkinesis-asl clean compile unidoc' from /
>> Using /usr/java/jdk1.8.0_60 as default JAVA_HOME.             | Using
>> /usr/java/jdk1.7.0_79 as default JAVA_HOME.
>> Note, this will be overridden by -java-home if it is set.       Note,
>> this will be overridden by -java-home if it is set.
>>
>> There have been some known issues with building the docs with JDK8 and I
>> believe those fixes are in mainline, and we could cherry pick these changes
>> in -- but I think it might be more reasonable to just build the 2.1 docs
>> with JDK7.
>>
>> What do people think?
>>
>>

--f4030437a6f40b48c7054d314143
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">I don&#39;t think this is an example of Java 8 javadoc bei=
ng more strict; it is not finding classes, not complaining about syntax.<di=
v>(Hyukjin cleaned up all of the javadoc 8 errors in master, and they&#39;r=
e different and much more extensive!)</div><div><br></div><div>It wouldn=
9;t necessarily break anything to build with Java 8 because it&#39;ll still=
 emit Java 7 bytecode, etc.<br><div><br><div><div>That said, it may very we=
ll be that it is somehow due to Java 7 vs 8, and is probably best to stick =
to 1.7 in the release build.=C2=A0</div></div></div></div><br><div class=3D=
"gmail_quote"><div dir=3D"ltr">On Sat, Apr 15, 2017 at 1:38 AM Ryan Blue &l=
t;rblue@netflix.com.invalid&gt; wrote:<br></div><blockquote class=3D"gmail_=
quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1=
ex"><div dir=3D"ltr">I&#39;ve hit this before, where Javadoc for 1.8 is muc=
h more strict than 1.7.<div><br></div><div>I think we should definitely use=
 Java 1.7 for the release if we used it for the previous releases in the 2.=
1 line. We don&#39;t want to break java 1.7 users in a patch release.</div>=
</div><div dir=3D"ltr"><div><br></div><div>rb</div></div><div class=3D"gmai=
l_extra"><br><div class=3D"gmail_quote">On Fri, Apr 14, 2017 at 5:21 PM, Ho=
lden Karau <span dir=3D"ltr">&lt;<a href=3D"mailto:holden@pigscanfly.ca" ta=
rget=3D"_blank">holden@pigscanfly.ca</a>&gt;</span> wrote:<br><blockquote c=
lass=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;=
padding-left:1ex"><div dir=3D"ltr">Ok and with a bit more digging between R=
C2 and RC3 we apparently switched which JVM we are building the docs with.<=
div><br></div><div>The relevant side by side diff of the build logs (<a hre=
f=3D"https://amplab.cs.berkeley.edu/jenkins/view/Spark%20Release/job/spark-=
release-docs/60/consoleFull" target=3D"_blank">https://amplab.cs.berkeley.e=
du/jenkins/view/Spark%20Release/job/spark-release-docs/60/consoleFull</a>=
=C2=A0<a href=3D"https://amplab.cs.berkeley.edu/jenkins/view/Spark%20Releas=
e/job/spark-release-docs/59/consoleFull" target=3D"_blank">https://amplab.c=
s.berkeley.edu/jenkins/view/Spark%20Release/job/spark-release-docs/59/conso=
leFull</a> ):</div><div><br><div><div>HEAD is now at 2ed19cf... Preparing S=
park release v2.1.1-rc3 =C2=A0| HEAD is now at 02b165d... Preparing Spark r=
elease v2.1.1-rc2</div><div>Checked out Spark git hash 2ed19cf =C2=A0 =C2=
=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =
=C2=A0 =C2=A0| Checked out Spark git hash 02b165d</div><div>Building Spark =
docs =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=
=A0 =C2=A0 Building Spark docs</div><div>Configuration file: /home/jenkins/=
workspace/spark-release-doc =C2=A0 Configuration file: /home/jenkins/worksp=
ace/spark-release-doc</div><div>Moving to project root and building API doc=
s. =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Moving to=
 project root and building API docs.</div><div>Running &#39;build/sbt -Pkin=
esis-asl clean compile unidoc&#39; from / =C2=A0 Running &#39;build/sbt -Pk=
inesis-asl clean compile unidoc&#39; from /</div><div>Using /usr/java/jdk1.=
8.0_60 as default JAVA_HOME. =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 | Us=
ing /usr/java/jdk1.7.0_79 as default JAVA_HOME.</div><div>Note, this will b=
e overridden by -java-home if it is set. =C2=A0 =C2=A0 =C2=A0 Note, this wi=
ll be overridden by -java-home if it is set.</div></div><div><br></div><div=
>There have been some known issues with building the docs with JDK8 and I b=
elieve those fixes are in mainline, and we could cherry pick these changes =
in -- but I think it might be more reasonable to just build the 2.1 docs wi=
th JDK7.</div><div><br></div><div>What do people think?</div><div><br></div=
></div></div></blockquote></div>
</div>
</blockquote></div></div>

--f4030437a6f40b48c7054d314143--