Important change to the policy governing the 10.2 branch.

[Rick Hillegas <Ri...@Sun.COM>]

I have wrapped up the last mega-merge from the trunk to the 10.2 branch. 
At this point the branch has changed state from beta to release 
candidacy. I would like to tighten up the branch control policy now. 
 From this time onward, I would like us to focus on keeping the branch 
stable. From now on, please hand-port your fixes from the trunk to 10.2. 
Until we vet a release candidate, please be extra cautious about porting 
fixes which might destabilize the branch.

Thanks,
-Rick

Re: Important change to the policy governing the 10.2 branch.

[Rick Hillegas <Ri...@Sun.COM>]

Daniel John Debrunner wrote:

> Rick Hillegas wrote:
>
>> I have wrapped up the last mega-merge from the trunk to the 10.2 branch. 
>
>
> Thanks Rick, for do all the merges, saves everyone else time.
> From the information in DERBY-1725, it seems the ending points
> (inclusive) were:
>
> Code - 448900 - 2006-09-22 05:18:15 -0700
> Docs - 448987 - 2006-09-22 09:46:58 -0700
>
> It's a little strange that the end points are different in time, there 
> are two additional commits to the code in those intervening four and a 
> half hours. In this case I guess there's not an major issue because 
> the latest doc and code changes are independent. Obviously if the doc 
> changes were related to recent code changes or vice-versa the ending 
> merge points would need to be in sync.

I agree, it's odd. It's too bad that the repository branches into doc 
and code lines above the release fork. It would be nice to branch a 
release in one transaction with a single copy command rather than with 
the two required today.

>
> DERBY-1862 is one of the missing changes (448949 & 448948), it can 
> show a 30-40% improvement in performance for some queries, maybe 
> especially when running with hibernate. Any chance of adding these 
> last two changes for the release candidate?

Fortunately, you caught me just before I bumped the version id! I have 
merged in this change and am rebuilding the release candidate.

Regards,
-Rick

>
> Dan.
>

Re: Important change to the policy governing the 10.2 branch.

["Jean T. Anderson" <jt...@bristowhill.com>]

Daniel John Debrunner wrote:
> Rick Hillegas wrote:
> 
>> I have wrapped up the last mega-merge from the trunk to the 10.2 branch. 
> 
> Thanks Rick, for do all the merges, saves everyone else time.
> From the information in DERBY-1725, it seems the ending points
> (inclusive) were:
> 
> Code - 448900 - 2006-09-22 05:18:15 -0700
> Docs - 448987 - 2006-09-22 09:46:58 -0700
> 
> It's a little strange that the end points are different in time, there
> are two additional commits to the code in those intervening four and a
> half hours. In this case I guess there's not an major issue because the
> latest doc and code changes are independent. Obviously if the doc
> changes were related to recent code changes or vice-versa the ending
> merge points would need to be in sync.

That last doc commit (revision 448987) was for DERBY-1839: Updates to
the Reference Guide based on the 10.2 doc review.

If code and doc commits were ever synchronized (i.e., finished docs
provided at the same time the code is committed), I think it'd be
something of a miracle -- and one I'd be happy to spend extra time
syncing in the end.

 -jean


> DERBY-1862 is one of the missing changes (448949 & 448948), it can show
> a 30-40% improvement in performance for some queries, maybe especially
> when running with hibernate. Any chance of adding these last two changes
> for the release candidate?
> 
> Dan.
> 

Re: Important change to the policy governing the 10.2 branch.

[Daniel John Debrunner <dj...@apache.org>]

Rick Hillegas wrote:
> I have wrapped up the last mega-merge from the trunk to the 10.2 branch. 

Thanks Rick, for do all the merges, saves everyone else time.
 From the information in DERBY-1725, it seems the ending points
(inclusive) were:

Code - 448900 - 2006-09-22 05:18:15 -0700
Docs - 448987 - 2006-09-22 09:46:58 -0700

It's a little strange that the end points are different in time, there 
are two additional commits to the code in those intervening four and a 
half hours. In this case I guess there's not an major issue because the 
latest doc and code changes are independent. Obviously if the doc 
changes were related to recent code changes or vice-versa the ending 
merge points would need to be in sync.

DERBY-1862 is one of the missing changes (448949 & 448948), it can show 
a 30-40% improvement in performance for some queries, maybe especially 
when running with hibernate. Any chance of adding these last two changes 
for the release candidate?

Dan.