[cross-project-issues-dev] Re: [HeliosAggregation] Failed for build 2010

[David M Williams <david_williams@xxxxxxxxxx>]

Hi Christian. Glad you got it working,
even if took some "manual" intervention. 

So, you manually "... did
a -repack of the original cononical.jar, then a signing, then pack200 "
and all is well. 
Just to be sure, that is what you do each time, right?
For every jar? Every build?  Those are indeed the correct, required
3 steps: -repack (conditioning), then sign, then pack200. (all with Java
5). 
And, I assume you do this for all your jars, and all others
work ok with no noise? 

Do you do anything with the specifiable properties to
control the pack process? I know some use -E4 since that was recommend
by Sun in some early work to avoid some particular problem with corrupt
files. So, if you don't , you might use -E4 to see if it helps .... and.
... my initial point .... if you use parameters at all, you must use exactly
the same ones for the repack step, and the pack200 step. It'd be hard not
to, using all the standard stuff ... but, I bet you do not use all the
standard stuff? :)

For the -E4 recommendation, see  "workaround
to downgrade pack200 effort to -E4"
described in https://bugs.eclipse.org/bugs/show_bug.cgi?id=154069

Do you happen to have jars inside this jar? Embedded
jars? That can raise other complications ... which I'm too tired to recall
at the moment. 

And, last of all, there are some few cases of jars where
pack200 does not work ... particular patterns of class file code that the
tools do not quite handle correctly. They are rare, but I think TPTP had
one of those. So, in a case like that, you need to mark it "do not
pack", but still have it signed. 

But in those cases, it always failed. It kind of sounds
like you have intermittent results, which is very odd (i.e. don't recall
hearing that, before). 

So, I'll send this to cross-project also ... there's many
experts on that list that  could maybe know right away what the problem
might be. 

Good luck next week. 


Thanks,

From:	Christian Campo <Christian.Campo@xxxxxxxxxxxx>
To:	David M Williams/Raleigh/IBM@IBMUS
Date:	05/20/2010 06:00 AM
Subject:	Re: [HeliosAggregation] Failed for build 2010-05-19_23-11-20 (FIXED)

---

hi David,

thanks for the hints you gave me. I managed to twiggle
around with the riena.tests....pack.gz and make it work in the helios aggregator
build. You now have a blue build (for the first time for quite some time).

About the "twiggeling":
I was able to verify that its not a java5/java6 problem
we run everything with java5. I also was was able to unpack200 the original
riena.tests pack.gz file. However when trying to pack200 that resulting
jar again I got the same error that the MessageBoxRidgetTest$2.class had
an invalid SHA1. So I manually (using JDKs tools) did a -repack of the
original .jar (that came out of our build process when we did the RC1),
then a signing, pack200 and only replaced the pack.gz file for riena.tests
in our update site and that did the trick.

So I know a workaround, still dont understand what is
causing this problem that is only happening in one bundle (which is particularly
large)

christian


Am 20.05.2010 um 07:36 schrieb David M Williams:


Yes, you can start it. I'm about to send a note to cross-project now, setting
a hard deadline of 10 am eastern, for Thursday. 

Thanks,

From:	Christian Campo <Christian.Campo@xxxxxxxxxxxx>
To:	David M Williams/Raleigh/IBM@IBMUS
Date:	05/20/2010 01:26 AM
Subject:	Re: [HeliosAggregation] Failed for build 2010-05-19_23-11-20

---

that sounds like something I can figure out......(the java6 thing is quite
possible)......Can I start the helios.agregator hudson build job when I
am confident that its fixed ? (since you probably are about to go to sleep,
and it will take me another hour to get to the office) ??? 

christian 

Am 20.05.2010 um 07:10 schrieb David M Williams: 


It comes from trying to unpack200 the org.eclipse.riena.tests...jar.pack.gz
file. 
You can try it your self, on your pack.gz file. 
If it passes with unpack200 from Java 6 JDK, but fails with unpack200 from
a Java 5 JDK, then I could guess at the problem: 
That you pack200 using Java 6. 
In some cases, this is fine ... but, in several cases, it will create a
pack.gz file which is incompatible with Java 5 
(and we intentionally use Java 5 to unpack, and do this verification).

Other than that, it might indicate some problem with "conditioning"
(preparing for pack200). I don't recall this specific error for those types
of problems ... but, only other thing I an think of. 

Why don't you try unpack200 yourself, and report back what you find. 

Thanks,

From:	Christian Campo <Christian.Campo@xxxxxxxxxxxx>
To:	David M Williams/Raleigh/IBM@IBMUS
Date:	05/20/2010 12:54 AM
Subject:	Re: [HeliosAggregation] Failed for build 2010-05-19_23-11-20

---

hi david,

do you have any hint what I can do to fix this.......I mean I didnt tamper
with that specific bundle or change it after signing it. Is it the pack200
file that it finds the error or the .jar file ?

christian

Am 20.05.2010 um 06:46 schrieb David Williams:

> The following error occured when building Helios:
> 
> org.eclipse.core.runtime.CoreException: Unable to unpack artifact
osgi.bundle,org.eclipse.riena.tests,2.0.0.v20100517_2_0_0_RC1 in repository
file:/shared/helios/aggregation/final/aggregate: Invalid content:org/eclipse/riena/internal/ui/ridgets/swt/MessageBoxRidgetTest$2.class
> Caused by: org.eclipse.osgi.signedcontent.InvalidContentException:
The file "org/eclipse/riena/internal/ui/ridgets/swt/MessageBoxRidgetTest$2.class"
in the jar "/tmp/signatureFile25702.jar" has been tampered!
> 
> Check the log file for more information: https://build.eclipse.org/hudson/view/Repository%20Aggregation/job/helios.runAggregator/36/console
> 


-------------------------------------------------------------
compeople AG
Untermainanlage 8
60329 Frankfurt/Main
fon: +49 (0) 69 / 27 22 18 0
fax: +49 (0) 69 / 27 22 18 22
web: http://www.compeople.de/

Vorstand: Jürgen Wiesmaier
Aufsichtsratsvorsitzender: Christian Glanz

Sitz der Gesellschaft: Frankfurt/Main
Handelsregister Frankfurt HRB 56759
Ust-Ident.-Nr: DE207665352
-------------------------------------------------------------