[Fwd: Boot cycle builds]

[Andrew John Hughes]

2009/10/27 Jonathan Gibbons <Jonathan.Gibbons at sun.com>:
> Andrew John Hughes wrote:
>
> 2009/10/26 Jonathan Gibbons <Jonathan.Gibbons at sun.com>:
>
>
> Andrew John Hughes wrote:
>
> 2009/10/26 Andrew John Hughes <gnu_andrew at member.fsf.org>:
>
>
> 2009/10/26 Kelly O'Hair <Kelly.Ohair at sun.com>:
>
>
> FYI...
>
> -------- Original Message --------
> Subject: Boot cycle builds
> Date: Mon, 26 Oct 2009 11:21:22 -0700
> From: Kelly O'Hair <Kelly.Ohair at Sun.COM>
> To: jdk7-gk at openjdk.java.net
>
>
> Heads up...
>
> As soon as we can do a jdk7 boot cycle build again, I plan on changing
> the default build to include a boot cycle build
> (where the jdk7 image is built, then used as the boot jdk to build
> it again).
>
> I consider it critical that boot cycle builds work, and I see
> no other course than requiring it as part of a default jdk7 build.
>
> It can be turned off with SKIP_BOOT_CYCLE=true, but I hope the
> integrators and everyone will be careful when it is turned off.
>
> Depending on your build situation, this could add 20-30 minutes
> (maybe more) to your jdk7 build, which is painful I know, but
> so is finding out that the jdk you just built doesn't work.
>
> Comments are welcome.
>
> -kto
>
>
>
>
> IcedTea does just this for a full bootstrap (the default) and it would
> be good to see the Sun developers applying the same level of testing
> to their builds.
>
> I would ask that we wait until the end of the current milestone (this
> week I believe) before adding this though.
> --
> Andrew :-)
>
> Free Java Software Engineer
> Red Hat, Inc. (http://www.redhat.com)
>
> Support Free Java!
> Contribute to GNU Classpath and the OpenJDK
> http://www.gnu.org/software/classpath
> http://openjdk.java.net
>
> PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
> Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8
>
>
>
> To add to that, I think just having RE doing a full bootstrap would
> have prevented the issues we've just hit with b74 (for the
> uninitiated, changes to both javah and sun.awt.FontManager broke the
> build).
>
>
> Andrew,
>
> The full bootstrap would not have uncovered the issues with the changes to
> javah. I did full bootstrap builds. The issue arose with fastdebug builds.
>
> -- Jon
>
>
>
> Strange; I did see the error when bootstrapping via IcedTea.
>
>
> Andrew,
>
> The fault shows up if you use the javac -g option. That is not the case in
> normal JDK builds, with or without the bootstrap build.
>
> -- Jon
>

Aaaah!  DEBUG_CLASSFILES="true" also turns this on and is standard for
all IcedTea builds.  It was introduced in an OpenJDK patch from Andrew
Haley, but prior to that we were manually patching in -g options.  We
don't ship stuff without debugging information.
-- 
Andrew :-)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and the OpenJDK
http://www.gnu.org/software/classpath
http://openjdk.java.net

PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8