"portmaster -o" behaves a little strangely (IMHO)

[Conrad J. Sabatier]

On Tue, 24 Jan 2012 08:57:14 -0800
Kevin Oberman <kob6558 at gmail.com> wrote:

> On Tue, Jan 24, 2012 at 7:14 AM, Conrad J. Sabatier <conrads at cox.net>
> wrote:
> > On Tue, 24 Jan 2012 09:21:05 -0500
> > Michael Scheidell <scheidell at freebsd.org> wrote:
> >>
> >> On 1/24/12 9:19 AM, Conrad J. Sabatier wrote:
> >> > # portmaster -CK -o lang/gcc46 gcc-4.4.7.20111108
> >> >
> >> what happens with:
> >>
> >> # portmaster -CK -o lang/gcc46 lang/gcc44
> >>
> >> ?
> >
> > Well, since gcc44 is no longer installed, I just tried:
> >
> > portmaster -o lang/gcc47 lang/gcc46
> >
> > And the actions that started were much more along the lines of what
> > I was expecting.  The build of gcc47 is currently underway as I
> > write this.
> >
> > So, it seems it's not a good idea to "mix and match" port origins
> > with port names in the case of this particular option.  Could this
> > be considered a "bug"?  It certainly seems to at least qualify as a
> > "quirk", I would say.  :-)
> >
> > --
> > Conrad J. Sabatier
> > conrads at cox.net
> 
> I wanted to note is that you installed lang/gcc46, the development
> version of the compiler. It gets updated very frequently and you may
> want this, but it is probably better for most people to install
> lang/gcc which is the stable version of gcc-4.6. It gets updated only
> when a new gcc-4.6 version is released, so it does not get updated
> nearly so often. Posts that depend on USE_FORTRAN or USE_GCC=4.6+ will
> both be happy with lang/gcc, though they will pull in the development
> version if no gcc-4.6 is installed.

Yes, I do recall now a lot of talk about 4.6 being subject to frequent
change.  Slipped my mind there for a bit.  :-)

I think I will back off to the other version.  I mainly was interested
in getting up to the minimum version needed to truly support my AMD
Phenom 9550 processor (-march=amdfam10) the same way Clang does.

Thanks for that info.

-- 
Conrad J. Sabatier
conrads at cox.net