In message <20030312203712.GH1710@desk06.rd.bbc.co.uk>
	on Wed, 12 Mar 2003 20:37:12 +0000,
	Jonathan Perkin <sketch@rd.bbc.co.uk> wrote:
> Need some sanity checking:
> 
> GNU's config.sub mangles a --host=LOWER_ARCH-sun-solaris into
> LOWER_ARCH-sun-solaris2 to avoid clashes with old solaris1 stuff, yet we
> define:
> 
> LOWER_VENDOR?=          sun
> LOWER_OPSYS?=           solaris
Current pkgsrc dosen't seems to care about SunOS 4.X stuff very much
and I think that supporting SunOS 4.X is wasted effort since it is too
old and can't handle these day's security problem any more.

> in bsd.prefs.mk.  This breaks anything which assumes a PLIST of
> ${MACHINE_GNU_PLATFORM}/* (e.g. ruby which I'm upgrading to 1.6.8).
Ruby package itself could solve this problem another way.  I'll update
ruby packages to 1.6.8 based in a few days.

> I'd suggest
> 
>   RCS file: /cvsroot/pkgsrc/mk/bsd.prefs.mk,v
>   retrieving revision 1.106
>   diff -u -r1.106 bsd.prefs.mk
>   --- bsd.prefs.mk        2003/01/25 22:39:56     1.106
>   +++ bsd.prefs.mk        2003/03/12 20:32:54
>   @@ -84,7 +84,7 @@
>    MAKEFLAGS+=            LOWER_ARCH=${LOWER_ARCH}
>    .  endif
>    LOWER_VENDOR?=         sun
>   -LOWER_OPSYS?=          solaris
>   +LOWER_OPSYS?=          solaris2
>    
>    .elif ${OPSYS} == "Linux"
>    LOWER_OPSYS?=          linux
>
> rather than fudging each config.sub (and pkgsrc/mk/gnu-config/config.sub),
I think that patching config.sub is better.

o GNU configure itself generate fuzzy way system platform.  I've
  checked on SunOS 5.8,

  (1) No argument configure results "sparc-sun-solaris2.8".
  (2) Giving -host=sparc-sun-solaris results "sparc-sun-solaris2".
      (Yes, this is config.sub's result.)

o I believe that current pkgsrc use "Solaris" as SunOS 5.X based
  system and recent Solaris aren't Solaris 2.X but Solaris X.  

Thus using solaris2 isn't good name.  If we rellay support Solaris 1,
then we should use names like SunOS 4.X and SunOS 5.X.

> unless there are plans to support Solaris 1.x machines in the future,
> but I'm not confident this is the "correct" fix (indeed it doesn't
> completely fix the currently-broken ruby install on Solaris as that also
> mangles out the -sun- part) so would appreciate some clarity.
As for ruby packge, I'll finish to ruby-base package to work fine
under Solaris.

-- 
Takahiro Kambe <taca@back-street.net>