[platform-releng-dev] MUST READ Re: A note about "signing" that requires

[Thomas Watson <tjwatson@xxxxxxxxxx>]

Please do not add any jarprocessor directives to your eclipse.inf files just yet.  This will cause the build to break.  Please wait until bug  https://bugs.eclipse.org/bugs/show_bug.cgi?id=412850 is sorted out.  We now have two commits that I think will cause the build to break tonight in PDE [1] and JDT [2]

Tom

[1] http://git.eclipse.org/c/pde/eclipse.pde.ui.git/commit/?id=8d6e22a453f6445a204bd432e46e2a385a51f1fa
[2] http://git.eclipse.org/c/jdt/eclipse.jdt.ui.git/commit/?id=e3eb92020c616353daf86752d252d5533bbe8473


Thomas Watson---07/15/2013 09:07:20 AM---I am pretty sure this commit [1] is going to cause build breaks similar to how my osgi.tests [2] one

From: Thomas Watson/Austin/IBM@IBMUS
To: "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
Date: 07/15/2013 09:07 AM
Subject: Re: [platform-releng-dev] A note about "signing" that requires your attention and maybe action
Sent by: platform-releng-dev-bounces@xxxxxxxxxxx

---

I am pretty sure this commit [1] is going to cause build breaks similar to how my osgi.tests [2] one did in the nightly builds.  See bug https://bugs.eclipse.org/bugs/show_bug.cgi?id=412850 for more details.  It appears Thanh has a suggested fix to also include the following in the eclipse.inf:

jarprocessor.exclude.pack=true

[1] http://git.eclipse.org/c/jdt/eclipse.jdt.ui.git/commit/?id=e3eb92020c616353daf86752d252d5533bbe8473
[2] http://git.eclipse.org/c/equinox/rt.equinox.framework.git/commit/?id=bd4a14fca9213e39df20c9c810ad2178582b216d

Tom

Daniel Megert ---07/15/2013 07:52:32 AM---Fixed the JDT UI test that failed due to the new signing stuff. Dani

From: Daniel Megert <daniel_megert@xxxxxxxxxx>
To: "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>, 
Date: 07/15/2013 07:52 AM
Subject: Re: [platform-releng-dev] A note about "signing" that requires your attention and maybe action
Sent by: platform-releng-dev-bounces@xxxxxxxxxxx

---

Fixed the JDT UI test that failed due to the new signing stuff. 

Dani 


From:        Thomas Watson <tjwatson@xxxxxxxxxx> 
To:        "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx> 
Date:        12.07.2013 14:52 
Subject:        Re: [platform-releng-dev] A note about "signing" that requires        your        attention and maybe action 
Sent by:        platform-releng-dev-bounces@xxxxxxxxxxx

---

I think the signing if the binary jars we have checked into git for the tests is causing a lot of failures in the nighty builds now.  I suspect the following need to add the eclipse.inf that David mentions below:

equinox.p2.tests - https://bugs.eclipse.org/bugs/show_bug.cgi?id=412848
osgi.tests - https://bugs.eclipse.org/bugs/show_bug.cgi?id=412850
pde.ui.tests - https://bugs.eclipse.org/bugs/show_bug.cgi?id=412851

Tom



David M Williams---07/11/2013 03:38:36 PM---Thanh, et. al., have added two important functions for us to improve our  "signing story": 1) We'll

From: David M Williams/Raleigh/IBM@IBMUS
To: "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>, 
Date: 07/11/2013 03:38 PM
Subject: [platform-releng-dev] A note about "signing" that requires your attention and maybe action
Sent by: platform-releng-dev-bounces@xxxxxxxxxxx

---

Thanh, et. al., have added two important functions for us to improve our "signing story": 1) We'll start to sign Windows and Mac executables! [1] 2) We will start to sign "inner jars" [2]. The second one is less "news worthy" (being required for only a few special cases) but it is the one that MIGHT require committers to make 1 or more changes. 

Unlike the PDE builds, where we skipped signing "test bundles", with CBI builds we now (correctly) sign all bundles. BUT, some of the "test bundles" have embedded in them large jars consisting of "data" or "test cases".  So, now, even those test bundle inner jars will get signed. Ordinarily, this should not hurt anything, but might add a few minutes to the build. So, if you have a test bundle, with lots (or large) "inner jars" then you might want to edit or add an eclipse.inf file saying to skip signing children. Especially if you see  test failures caused by the signing (which itself might be interesting .... but, might not be, either). 

If you don't have an eclipse.inf file, as a reminder it goes in META-INF directory: 
META-INF/eclipse.inf 
and should suffice to have one line: 
jarprocessor.exclude.children=true 

Second point, ordinarily we do not "sign" nightly builds, but I've temporarily changed that bit of logic in our production builds so that nightly builds will be signed (probably until next I-build) just to help flush out any problems or issues with these new signing functions. So a) keep you eye on nightly tests, so see if any new failures due to signed inner jars; and b) you might want to try out a Nightly to make sure the Windows and Mac versions "run" without any "gatekeepers" turned off. If everything works as planned, next week we'll resume to the "don't sign nightlies" settings to save a little wear and tear on the infrastructure. 

And, just to complicate matters, for Nightlies, where qualifiers always change, we'd expect all "inner jars" to start getting signed tonight (with the temporary change in settings). But, for I and M builds, if you have a case where you WANT your inner jars signed you may have to "touch" the bundle to make sure its qualifier increases or else will be replaced during "comparator" operation by an "older bundle" without the inner jars signed. 


Thanks, 

[1] https://bugs.eclipse.org/bugs/show_bug.cgi?id=388878 
[2] https://bugs.eclipse.org/bugs/show_bug.cgi?id=401141_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
[attachment "graycol.gif" deleted by Daniel Megert/Zurich/IBM] _______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev
_______________________________________________
platform-releng-dev mailing list
platform-releng-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/platform-releng-dev