IcedTea bootstrap and OpenJDK build broken with ecj-3.8/4.2
gnu.andrew at redhat.com
Mon Apr 15 14:27:21 PDT 2013
----- Original Message -----
> > I don't see what Fedora's ecj packages have to do with IcedTea at all.
> > IcedTea's bootstrap works fine both for me and our autobuilders. Stop
> > trying
> > to use
> > something that isn't yet supported and it'll work fine for you too.
> > We can look at a newer ecj in the 2.5 timeframe.
> As I've been doing most of the packaging of ecj in Fedora of late,
> I must double-check my understanding of something: this issue, do
> you believe it to be caused by some fedora-specific patch? My
> assumption so far is that this bootstrapping issue is from changes
> to ecj upstream. If this is correct, I might suggest for clarity
> on this list to stop referring to "fedora's ecj" and instead refer
> to the upstream ecj version that introduces the issue. If this is
> not correct, I would be very interested in any feedback about what
> patch is guilty.
I think part of what Matthias is complaining about is:
He still hasn't posted a config.log to explain what the other issue is
with the diamond operator.
I suspect, as I noted to you about the IcedTea6 issue, that ecj has changed
what classes get dragged in during the pre-bootstrap stage and it isn't prebuilding
as many OpenJDK classes. I don't think it's so much a bug as we need to be more
explicit. If someone with the IcedTea6 issue wants to try HEAD and see if the
SSLParameters issue has been resolved, I can know whether it's worth backporting
As noted, it still bootstraps fine here and a release branch is not the place to
be trying to support new versions, especially not when we have a security release
taking place in the next 24-48 hours.
Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)
PGP Key: 248BDC07 (https://keys.indymedia.org/)
Fingerprint = EC5A 1F5E C0AD 1D15 8F1F 8F91 3B96 A578 248B DC07
More information about the distro-pkg-dev