lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steven A Rowe <>
Subject RE: Robert Muir thinks we should stop supporting Sun/Oracle JDK 1.5 on branch_3x
Date Fri, 15 Apr 2011 15:19:13 GMT
Robert, did you read my email at all?

Your javadoc comment has a link to a non-existent class.  That’s not “valid” under anybody’s
jdk, regardless of version.  Try running ‘ant javadoc’ under your favorite jre on branch_3x.
 No worky, dude.

The fix is to remove “.util” from the package name.  Doing so will not violate your principles.


From: Robert Muir []
Sent: Friday, April 15, 2011 10:52 AM
Subject: RE: Robert Muir thinks we should stop supporting Sun/Oracle JDK 1.5 on branch_3x


my code is 100% valid java 5. if this oracle bug bothers you so much, perhaps you should switch
to ibm's j9.

this is an oracle specific bug, again there is nothing wrong with my code technically, thus
no justification for revert.

remember, working around bugs in particular jre implementations is optional, I've helped with
this before but we need to make sure everyone understands this is not mandatory.

I code to the spec
On Apr 15, 2011 9:55 AM, "Steven A Rowe" <<>>
> Turns out that the problem is that ReusableAnalyzerBase is not in the same package in
branch_3x as on trunk, so the class reference is just wrong on branch_3x, regardless of the
JDK one uses to generate javadocs. (Oracle 1.6.0_21 JDK triggers the same failure on branch_3x.)
> Robert, I trust you will see your way clear to fixing this, regardless of your lack of
interest in supporting Oracle's buggy implementations.
> But I still think the issue should be resolved: Robert thinks he doesn't have to play
by the rules (JDK 1.5 support for Lucene, javadocs warnings fail the build, failed builds
are not allowed), and I think that's unacceptable. I think the proper course of action is
calling a vote to change the rules, not just blithely ignoring them because you don't feel
like following them.
> Unhappily,
> Steve
>> -----Original Message-----
>> From: DM Smith [<>]
>> Sent: Friday, April 15, 2011 9:31 AM
>> To:<>
>> Subject: Re: Robert Muir thinks we should stop supporting Sun/Oracle JDK
>> 1.5 on branch_3x
>> What is the bug? I have an interest in this component and am willing to
>> see about fixing it. It appears that it is a Javadoc bug??? Why would we
>> keep good code out for that?
>> -- DM
>> On 04/15/2011 09:15 AM, Steven A Rowe wrote:
>> > As a result of Robert Muir's r1092398 commit on branch_3x (the Latvian
>> analysis stuff), the build is now broken under Sun/Oracle JDK 1.5:
>> >
>> > [javadoc]
>> C:\svn\lucene\dev\branches\branch_3x\lucene\contrib\analyzers\common\src\
>> java\org\apache\lucene\analysis\lv\ warning -
>> Tag @link: reference not found:
>> org.apache.lucene.analysis.util.ReusableAnalyzerBase.TokenStreamComponent
>> s
>> > [...]
>> > [javadoc] 5 warnings
>> >
>> > C:\svn\lucene\dev\branches\branch_3x\lucene\build.xml:217: The
>> following error occurred while executing this line:
>> > C:\svn\lucene\dev\branches\branch_3x\lucene\common-build.xml:813:
>> Javadocs warnings were found!
>> >
>> > Robert said on #lucene IRC:
>> >> (9:02:53 AM) rmuir: its a bug in the EOL'ed java5
>> >> (9:02:57 AM) rmuir: im not fixing it
>> >> (9:03:21 AM) rmuir: build isnt broken
>> >> (9:04:07 AM) rmuir: jre bug = not my problem
>> >> (9:04:32 AM) rmuir: complain to oracle, having hit another one of
>> their bugs at 3am last night, ive had it with their crap
>> >> (9:04:42 AM) rmuir: i didnt break the build
>> >> (9:04:43 AM) rmuir: its a bug in your jre
>> > Needless to say, I disagree vehemently with Robert.
>> >
>> > I think Robert should fix this or his commit should be reverted.
>> >
>> > I'm interested in hearing other opinions on this.
>> >
>> > Steve
View raw message