Request for review: perl-BerkeleyDB

[Ralf Corsepius]

On Tue, 2005-08-02 at 11:07 -0500, Tom 'spot' Callaway wrote:
> On Tue, 2005-08-02 at 17:57 +0200, Ralf Corsepius wrote:
> > On Tue, 2005-08-02 at 10:45 -0500, Tom 'spot' Callaway wrote:
> > > On Mon, 2005-08-01 at 16:43 -0500, Steven Pritchard wrote:
> > > 
> > > > > Optflags missing from %build:
> > > > > %{__make} %{?_smp_mflags} OPTIMIZE="$RPM_OPT_FLAGS"
> > > > 
> > > > I'm still not convinced that it is a good idea to allow modules to
> > > > build with different optimization flags than perl itself, but I seem
> > > > to be the only one, so I've added it to this package (and to
> > > > cpanspec).
> > > 
> > > Again, perl is being build with $RPM_OPT_FLAGS. So, unless you build
> > > with it, you're not matching perl's optimization flags.
> > It depends on the Makefile's internals. ExtUtils::MakeMaker
> > Makefile.pl's use the OPTIMIZE-flags that had been used when building
> > perl. Passing OPTIMIZE to make in most cases is superfluous.
> 
> Nevertheless, the flags used to build perl were $RPM_OPT_FLAGS. So we're
> not changing anything, even if we are being a little redundant. If
> nothing else, we're ensuring that we're using the same OPTIMIZE-flags
> that perl used. :)
Nope, all you are doing is passing RPM_OPT_FLAGS redundantly for a third
time (Once through MakeMaker ("The default value"), a second time when
running perl Makefile.pl ("The configuration time value"), and a third
time when running make ("The build-time value").

This is similar redundant as mandating 
./configure CFLAGS=RPM_OPT_FLAGS
make CFLAGS=$RPM_OPT_FLAGS
for autoconf scripts.

Ralf