Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 12 Jul 2002 12:59:16 +1000 (EST)
From:      Bruce Evans <bde@zeta.org.au>
To:        Mike Barcroft <mike@FreeBSD.org>
Cc:        Giorgos Keramidas <keramida@FreeBSD.org>, <current@FreeBSD.org>
Subject:   Re: sparc64 tinderbox failure
Message-ID:  <20020712123623.K26605-100000@gamplex.bde.org>
In-Reply-To: <20020711180442.G89841@espresso.q9media.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 11 Jul 2002, Mike Barcroft wrote:

> Bruce Evans <bde@zeta.org.au> writes:
> > The setting of NO_WERROR belongs in [BTX]MAKE if anywhere.  This is
> > already done for [BX]MAKE but not for TMAKE.  However, I don't like
> > turning off warnings for any of these.  Warnings for these stages may
> > be even more important and should be less likely than warnings for
> > building the final world, since it is very important for basic tools
> > to be correct and for their source to be careful about portabilty
> > issues.
>
> Well, unfortunately I don't think we can depend on older compilers
> having correct warnings.  In PR 40382, it would seem the 4.5 -> HEAD
> upgrade path is broken because of fatal warnings.  A good workaround

The problem in that PR seems to be just the current breakage of
__printf0like.

> for that problem might be specifying NO_WERROR for the entire build,
> in which case -Werror becomes useless anyway.  So we might just as
> well disable early fatal warnings and hope that developers can catch
> most of the bugs later on in the build.

It's true that the host compiler might emit spurious warnings.  But I
think it's worth checking new ones (for new hosts and especially for
new sources) carefully.  I won't argue this point strongly.  We mostly
avoid seeing problems by not enabling many warnings (we don't set WARNS
for cc, and we set it to to for binutils).

> > This also has some style bugs :).  Any setting of NO_WERROR turns it on,
> > so setting it to different spellings of boolean true is just confusing.
> > It is set correctly for for [BX]MAKE.
>
> Oh, that's a much nicer location.  :)  I think only BMAKE has
> NO_WERROR defined, not XMAKE.

XMAKE is based on BMAKE :).

Bruce


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020712123623.K26605-100000>