<AWT Dev> Urgent Request for review: 7035109 Regression: awt SplashScreen/test18.sh fails - missing mapfile entry

David Holmes David.Holmes at oracle.com
Sun Apr 10 15:22:46 PDT 2011

Phil Race said the following on 04/11/11 02:39:
> I think that there can be times when something is extremely cross-area
> and that makes it more of a pain if it can't easily be divided up.

The embedded changes were very cross-area and we would have had to put 
back through three different team repos, meaning we wouldn't actually 
get a full working copy until everything propagated through the master. 
Hence everything was pushed through TL/jdk (with approval from build and 
AWT folk)

> But I completely agree that running the right tests is a vital part of 
> making
> sure there are no problems. I don't know if that would have happened any 
> faster in this case.

Yes the testing was lacking - mea culpa. This was the last piece to push 
through and the AWT folk insisted that we re-work parts of it compared 
to how it was done in 6u23. The change that introduced the native method 
wasn't specifically tested.

> But so is the review. Code changes that cross areas should also be 
> posted for
> review by the relevant teams. Some times that might save pain down
> the line. So if you change awt code, send the review to the AWT team (etc).
> I expect core libs would like to know if I decided to change something 
> in java.util :-)

Just so there's no misunderstanding these changes were all approved by AWT.

All of the SE Embedded integration work was done in full cooperation 
with the TL, Build and AWT folk and I've immensely grateful for their 
assistance in getting this all pushed through.

This fix has been pushed into the PIT respin for b138.


> -phil.
> On 4/9/2011 8:08 AM, Kumar Srinivasan wrote:
>> Approved!. I agree with Alan if a change is to be made in
>> a component, it is best that it is pushed to that component's
>> forest/repo, where all the necessary/appropriate tests will be
>> performed on a nightly basis.
>> Kumar
>> ----- Alan.Bateman at oracle.com wrote:
>>> From: Alan.Bateman at oracle.com
>>> To: David.Holmes at oracle.com
>>> Cc: awt-dev at openjdk.java.net, core-libs-dev at openjdk.java.net
>>> Sent: Saturday, April 9, 2011 1:59:17 AM GMT -08:00 US/Canada Pacific
>>> Subject: Re: Urgent Request for review: 7035109 Regression: awt 
>>> SplashScreen/test18.sh fails - missing mapfile entry
>>> David Holmes wrote:
>>>> Very simple review - the mapfile was missing an entry for a new
>>> native
>>>> method added in 7030063 and caused an UnsatisfiedLinkError
>>>> http://cr.openjdk.java.net/~dholmes/jdk7-clone/webrev-7035109/
>>>> Failing test now passes.
>>>> Due to the urgency this will get pushed directly into the TL PIT jdk
>>>> repo so that it will be paired with 7030063 and present in b138.
>>> There is always a bit of risk pushing AWT or other client area changes
>>> to the TL forest as probably very few of us run those tests. The
>>> update
>>> the map file looks good to me.
>>> -Alan.

More information about the awt-dev mailing list