svn commit: r228668 - head/usr.bin/netstat

[David Schultz]

On Sat, Dec 17, 2011, David Schultz wrote:
> On Sat, Dec 17, 2011, Dimitry Andric wrote:
> > On 2011-12-17 23:32, Dimitry Andric wrote:
> > > Author: dim
> > > Date: Sat Dec 17 22:32:00 2011
> > > New Revision: 228668
> > > URL: http://svn.freebsd.org/changeset/base/228668
> > > 
> > > Log:
> > >   Revert r228650, and work around the clang false positive with printf
> > >   formats in usr.bin/netstat/atalk.c by conditionally adding NO_WFORMAT to
> > >   the Makefile instead.
> > >   
> > >   MFC after:	1 week
> > 
> > Requested by:	bz
> 
> Have you been keeping track of the other hacks you've been
> sprinkling throughout the tree to work around clang bugs, e.g.,
> the one in fsdb?  It would be unfortunate if someone else has to
> waste their time later on figuring out what you did, when we could
> just as easily have waited a month for the clang bug to be fixed.
> 
> Incidentally, the "bug" you fixed in telnet/utilities.c is also a
> false positive; clang doesn't understand that an index into a
> string constant is also a string constant.
> 
> By the way, I think it's great that you've found so many actual
> bugs in the tree.  I'm just complaining about a small subset of
> the changes, which fixed non-bugs. :)

Sorry, one more: In less(1), you cast away a bunch of const qualifiers
to fix some warnings, but that seems like a step in the wrong
direction.  The warnings were complaining about genuinely bad code.
Disabling the warnings with casts doesn't make less(1) any better;
instead, it guarantees that nobody will ever fix the code.

Perhaps the larger question is whether it makes sense to fix non-bugs
in contributed code at all.  What do we get out of it?  Maybe if the
contrib software is poorly maintained we'll find a bunch of real bugs
that won't be addressed upstream.  Otherwise, the diffs are only creating
headaches for whoever imports the next version.