broken build (Was: /hg/icedtea6: 6 new changesets)

[Dr Andrew John Hughes]

On 01:15 Fri 30 Jul     , Mark Wielaard wrote:
> On Thu, 2010-07-29 at 21:55 +0100, Dr Andrew John Hughes wrote:
> > On 14:34 Thu 29 Jul     , Omair Majid wrote:
> > > > That seems right, given the circumstances.  The question is how you
> > > > got into a state where there was a conflict to begin with.
> > > >
> > > 
> > > Hm... 'hg view' shows what might have happened. Apparently mjw had an 
> > > old checkout lying around (missing hg pull && hg update perhaps?).
> > > 
> > 
> > I haven't checked that.  My hg view segfaults, due to tcl not liking gcc 4.5:
> > 
> > $ hg view
> > *** buffer overflow detected ***: wish terminated
> 
> It works on my machine, and it is useful to see what seemed to have
> happened. I still don't understand how that can have be. But apparently
> I did my commit and then the merge against something really old :{
> 

Neither do I.  It's very odd.

> > > > Note that you can check what will be pushed with 'hg in'.
> > > 
> > > Shouldnt that be 'hg out'? IIRC 'in' (or incoming) checks for the 
> > > changesets that would get pulled in on hg pull
> > > 
> > 
> > Yeah, typo.  hg in is useful too though! :-)
> 
> Wish I had used that before pushing. hg outgoing is what would have
> shown me something odd was going on before a push.
> 

Yes, it's very useful in these cases!

> > > > There don't seem to be any problems building anyway.
> > > 
> > > Yup, I can confirm. Even though there are tons of files listed, they 
> > > seem to refer to stuff that was merged. The build works just fine.
> 
> Sadly it seems something did break. The auto-builder just informed me:
> http://icedtea.classpath.org/pipermail/testresults/2010-July/000672.html
> 
> It only seems to impact older setups that trigger the QNAME workaround
> patch. I cannot immediately see how my commit/merge broke that. But will
> try to fix it tomorrow. Slightly scared my fix/merging will make it
> worse though. So first some sleep to get a clear head.
> 

I see you now fixed this.  I think your autobuilder is the only one
that triggers this old QName workaround, as it still uses gcc 4.3, and
it's the workaround that was duplicated.

Maybe you were testing the fix on your laptop on some previous
occasion, due to the autobuilder failure?

> Cheers,
> 
> Mark
> 

-- 
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