Re: next draft of rpm guidelines are here

[Chipzz <chipzz ULYSSIS Org>]

On 3 May 2001, Dean Scott wrote:

> From: Dean Scott <drscott rochester rr com>
> Subject: Re: next draft of rpm guidelines are here
>
>
> > Welcome aboard!
> >
> > > I've been doing a lot of packaging for a custom RPM
> > > distro lately and would like to help out wherever I can. Onto the
> > > comments ..
> > > In case you haven't caught it already, there's a typo in the BuildRoot
> > > (tmppath not tmmpath). Also, most spec files i've seen pass the rpm
> > > flags when configure is run, not when make is run (I dont suppose it
> > > makes a difference, however this is how the %configure macro does it
> > > also).
> >
> > This generally does not make a difference because the entire %build
> > section is converted to a shell script and executed.  Reguardless of
> Agreed, I guess thats just sort of my personal preference.
> > where you set the CFLAGS/CXXFLAGS the compiler will see them.  However
> > if a configure script needs them (for example you have a -I option) then
> > putting it before the ./configure helps.
>
> After looking it over one more time however, it would be nice to use
> CFLAGS="$CFLAGS $RPM_OPT_FLAGS" so that the packager can pass custom
> flags if they desire to do so.

Which brings up a quite painfull point: c++.
Standard rpm uses RPM_OPT_FLAGS for both c _and_ c++. Which may screw up if
-fno-exceptions is in the CFLAGS (which is not an uncommon optimization).

Do the *mm packages actually use exceptions? Never worked with them, so I
have no idea.

Just a small thought...

Greetings,

Chipzz AKA
Jan Van Buggenhout
-- 

--------------------------------------------------------------------------
                  UNIX isn't dead - It just smells funny
                            Chipzz ULYSSIS Org
--------------------------------------------------------------------------