FYI: Fix PR632: patches/security/20110215/6878713.patch breaks shark zero build

[Mark Wielaard]

Hi Andrew,

On Thu, 2011-02-17 at 17:09 +0000, Dr Andrew John Hughes wrote:
> On 22:50 Wed 16 Feb     , Mark Wielaard wrote:
> > On Wed, 2011-02-16 at 21:47 +0000, Dr Andrew John Hughes wrote:
> > > On 22:28 Wed 16 Feb     , Mark Wielaard wrote:
> > > > 2011-02-16  Xerxes Ranby  <xerxes at zafena.se>
> > > >             Mark Wielaard  <mark at klomp.org>
> > > > 
> > > >     PR632: 6878713.patch breaks shark zero build
> > > >     * patches/icedtea-stdc-limit-macros.patch
> > > >     (openjdk/hotspot/src/share/vm/utilities/globalDefinitions_gcc.hpp):
> > > >     Only define __STDC_LIMIT_MACROS if undefined.
> > > >     * Makefile.am (ICEDTEA_PATCHES): Add new patch.
> > > >     * NEWS: Updated.
> > > > 
> > > Did you test this with both versions of HotSpot?
> > 
> > No, but I don't see how it would fail, the patch is right after the
> > patch that broke it. It is currently going through all the buildbot
> > builders. If you have configurations you want to have regularly tested
> > please setup a buildslave for it.
> > 
>
> Please test HotSpot patches with both available HotSpot builds to avoid
> breakage.

No. I am happy to also test the default configuration, which I did, if I
am fixing issues on a non-default configuration  that I care about. But
I am not going to test all random variants that might happen.

If you care about non-default configurations please either make sure
that it is in the default build/test setup. Or provide a buildslave for
it, so that it is regularly tested on all commits. That is what Xerxes
and I did for the non-default configurations that we care about (for
x86_64 zero/shark isn't enabled by default, and not everybody has an arm
setup).

> I'd prefer not to have yet more work on my plate because people don't do full
> testing of patches and commit them without any review.

It is hard to get a good feeling for tone in an email, so apologies for
misinterpreting. But I really don't appreciate this nitpicking. You
committed a patch, that you didn't discuss on the list, which broke the
zero/shark configuration that Xerxes and I care about. And that broke
the autobuilders that we rely on to catch issues (they did send you
email about the breakage). We quickly filed a bug report, analyzed the
root cause, discussed some patch variants and fixed it in a way that
minimized the impact and was as localized to the change that broke
things as possible.

Instead of a Thank You for making sure things were fixed ASAP, we get a
lecture on all things we should have done, but that you failed to do in
the first place with your original commit that broke things.

Cheers,

Mark