This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: werror fallout for cross-builds (was: Re: [BUILDROBOT][PATCH] Fix mmix (unused variable))

On Tue, 22 Jul 2014, Mike Stump wrote:
> On Jul 22, 2014, at 1:40 PM, Hans-Peter Nilsson <> wrote:
> >
> > *Developers* (or rather, people cross-building non-released gcc
> > source in their usual setup) don't use the fairly old or even
> > broken host gcc versions that can be expected in use in the
> > general public (well, the users that still want to build gcc
> > from releases and not use pre-built binaries).

(Hey, I proved that false myself and stated as much, see the
"But, the above..." rebuttal half-way through my post!)

> :-)  Speak for yourself.  I do cross, I deliver cross, and we
> just use the same old /bin/gcc that everyone else uses.  And, we
> might well deliver on OSes other than the one released last
> week.  In my case, /bin/gcc is 4.4.7.  Though, I usually  develop
> on 4.6.3 and 4.8.2.  So, what I want is software that builds and
> works.  I object to any patch that causes gcc to not build.

Mike, you miss the point of my post, and the patch too.  Maybe I
was unclear.  There seems to be violent agreement...

First, about the effect on the patch, regarding code deliveries
like your case above, you don't deliver DEV-PHASE = experimental
code (hopefully, with all the default redundant internal testing
it does).  More likely, you deliver releases, in which this
developer-phase testing wouldn't be enabled.

The intent of the patch was to help avoiding the *GCC developer*
situation where a person patches a lot of targets but in his
sanity-build misses out on introducing valid warnings about a
typo-level warning, exactly like the commit from which this
thread started.

The patch worked as intended, but as I mentioned (apparently
ambiguously enough), that intent was based on a false pretense
that most targets *do* work with -Werror.  The fallout is
actually (still) overwhelming.  You definitely wanted to make
sure I didn't miss that last point.  You don't have to worry
about that, never needed.

(I did mention breaking host gcc versions overlapping those you
mention - including quotes of identical breakages!)

I posted the patch on the off-chance that there actually *is* a
later version in which all invalid warnings are gone.  Note that
I didn't actually ask for approval.  I did ask for a host gcc
version where builds with --enable-werror-always work!

brgds, H-P

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]