Changeset issues with jdk6/jaxp and jdk6/jdk

[Dr Andrew John Hughes]

On 11:57 Fri 18 Feb     , Kelly O'Hair wrote:
> 
> We have just discovered that a changeset made to the jdk6/jaxp  
> repository is missing a comment:
>    http://hg.openjdk.java.net/jdk6/jdk6/jaxp/rev/73a4d5be8649 (No  
> bugid or review information)
> The actual patch in the changeset is not the issue, but the comment of  
> the changeset is.
> 
> [By the way, there is a new jaxp source bundle that is needed, which  
> is a different heads up for people
>   that cache these source drop bundles. But that is a separate issue.]
> 
> This changeset should have been caught as not having a proper  
> changeset comment by
> our jcheck mechanism, but I had inadvertently turned off this check  
> with a change
> made back in b17.
>     http://hg.openjdk.java.net/jdk6/jdk6/jaxp/rev/33e6947318ae
> The comments=lax option should not be in these .jcheck/conf files.
> This is a problem with all the jdk6 repositories that I will rectify.
> 
> In the process of fixing this .jcheck/conf file, I discovered 3  
> additional changesets in the jdk6/jdk
> repository done between 5-8 months ago that have also violated the  
> changeset comment rules:
>    http://hg.openjdk.java.net/jdk6/jdk6/jdk/rev/a25f15bfd04b    
> (Missing "-by" after "Reviewed")
>    http://hg.openjdk.java.net/jdk6/jdk6/jdk/rev/bf87d5af4361  (Missing  
> ":" after the bugid)
>    http://hg.openjdk.java.net/jdk6/jdk6/jdk/rev/d77434402021 (Extra  
> "Fixed" and text that should be be there)
> All snuck by because I had added that comments=lax back in jdk6-b17  
> time frame.
> 
> So I certainly should fix .jcheck/conf, but the question remains as to  
> what to do with these
> other changesets. The jdk6/jaxp one could be rolled back and fixed  
> easily, the jdk6/jdk ones
> are a bit older and would require rebasing all the newer changesets.
> 
> I'm open to ideas here. Should we avoid repository rollbacks  
> completely? Or just rollback/fix jdk6/jaxp?
> 

Please don't rollback anything.

We should avoid rolling back changesets that have been made public and
pulled into other repositories, which includes the JAXP one.

A few missing comments is not the end of the world.  No-one seems to have
even noticed the earlier ones until now.  Messing around with the revision
history is a lot more dangerous IMHO.

Given jcheck is still some proprietary script, I don't have a clue how
it actually works.  Does it check all previous changesets?  If so, I presume
these need to be exempted as Igor says.

Note that there are other jcheck exemptions in OpenJDK6; for instance, we had
to allow repeated use of the same bug ID in HotSpot to allow merges.

> -kto
> 

-- 
Andrew :)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and IcedTea
http://www.gnu.org/software/classpath
http://icedtea.classpath.org
PGP Key: F5862A37 (https://keys.indymedia.org/)
Fingerprint = EA30 D855 D50F 90CD F54D  0698 0713 C3ED F586 2A37