[wtp-releng] Yes ... we are done now: Fw: [wtp-dev] Declaring Build for

[David M Williams <david_williams@xxxxxxxxxx>]

This build had that final fix for bug
335991 and is what was contributed
to Helios SR2 repo and EPP packages (ready later on Friday). 

From here on, it will take a "catch
your computer on fire" sort of bug to consider a respin. A patch/update
site would most likely be more appropriate, even for a serious regression.


There is one week before official declare,
but that is primarily "prep" buffer ... and any respins would
introduce high levels of risk ... under the principle of "anything
can go wrong when you rebuild". 

Case in point ... you'll notice that
our final build had 3 JUnit failures, 2 are the typical intermittent errors
we see frequently [contradiction intended :) ] but one of them was different.

The comparator test failed showing that
the currently built JST level tests were all three files short. from reference
bundles. This was because the JST tests were not signed. Full story is
that the build was spinning waiting for them to be signed for an hour or
so (which is a long) ... I've seen this in the past when the zips were
actually signed, but no change detected by our build scripts ... so, I
"touched" the files, so the build scripts would see changed,
and continue on its way ... but this time, apparently, for some unknown
reason, the zips had never been picked up by the eclipse.org signing que
(though, they were not signing anything else, from what I could tell).
I've only seen this happen once every 3 months, or so, so doesn't seem
worth a bug report, at the moment ... but, does demonstrate how a simple
rebuild can go wrong. (Especially when you do your final rebuild 3 or 4
times, as we have :/ 
 
While unfortunate to break routine,
in actual fact, the way the build scripts and p2 mirroring works, is that
the OLD version of the bundles are pulled for final distribution, when
the version/qualifier is identical, so in this case the old, signed versions
are what ended up in our distribution, so everything is signed after all.
Pretty cool, eh? Kind of an unintended benefit of using p2.mirror task
(with comparator) for our distributable. 

So, rest, test, rest, and test some
more :) ... job well done on providing much improved quality here in SR2!


Thanks so much to you all, 

 


----- Forwarded by David
M Williams/Raleigh/IBM on 02/18/2011 03:40 AM -----

From:      
 David M Williams/Raleigh/IBM@IBMUS
To:      
 wtp-dev@xxxxxxxxxxx,
wtp-dev@xxxxxxxxxxx
Date:      
 02/18/2011 03:17 AM
Subject:    
   [wtp-dev] Declaring
Build for wtp-R3.2.3-M: M-3.2.3-20110217214612
Sent by:    
   wtp-dev-bounces@xxxxxxxxxxx

---

Download Page: 
http://download.eclipse.org/webtools/downloads/drops/R3.2.3/M-3.2.3-20110217214612

General Smoketest results page:
http://wiki.eclipse.org/WTP_Smoke_Test_Results

_______________________________________________
wtp-dev mailing list
wtp-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-dev