Question regarding workaround for gnutar 1.20 on 10.5 (and related question about gcc)

[Tabitha McNerney]

Hello all --

I would like to install the gnutar port version 1.2.0 on an Xserve running
Leopard Server (10.5.x) and noticed that when attempting to install, the
install failed. There is indeed a trouble ticket for gnutar version 1.19
(ticket # 13462 regarding Leopard problems).

Consulting the latest revision of the gnutar 1.20 Portfile (updated "1 month
ago"), I saw this interesting note:

http://trac.macports.org/browser/trunk/dports/archivers/gnutar/Portfile

version 1.20 with workaround for broken gcc-4.0 on 10.5: *beta gcc 4.2 needs
> to be installed*
>

and then the reference to gcc 4.2 at the tail of the Portfile:

48     platform darwin 9 {
> 49             # gcc 4.0 fails to compile gnutar on 10.5 (probably will get
> fixed with XCode 3.1)
> 50             configure.compiler      gcc-4.2
> 51     }
>

Leaving it to one's imagination, its not too difficult to fathom that some
people who are Apple Developer Connection members might have early access to
XCode 3.1 pre-release which may or may not included gcc 4.2. So, I looked at
the MacPorts database and found two port options regarding gcc 4.2:

1.) The port named: *apple-gcc42*, version 5531 whose Portifle is at:

http://trac.macports.org/browser/trunk/dports/lang/apple-gcc42/Portfile

and I did noticed that the apple-gcc42's Portfile lists a Master site of:

http://www.opensource.apple.com/darwinsource/tarballs/other/

Indeed, a the URL above, I found the resource defined in the apple-gcc42
Portfile named:

*gcc_42-5531.tar.gz*

So I downloaded it and looked at the file named "README.Apple" which has
this header, excerpted:

APPLE LOCAL file documentation
>
> This file describes Apple's version of GCC 4.x modified for Darwin /
> Mac OS X.  Although Apple's stated policy is to contribute all of its
> GCC work to the FSF GCC mainstream, at any given moment there will be
> changes that are permanently unacceptable for FSF GCC, in need of
> rework before acceptance, or that we simply aren't ready to send in.
> This version of GCC contains all those changes.
>
> In keeping with provision 2a of the GPL, each Apple change is marked
> with a comment saying "APPLE LOCAL", followed by optional words "begin",
> "end", or "file", followed by a short phrase describing the change
> generally ("AltiVec" for instance, if the change is related to AltiVec
> support).
>
> BUILDING, THE APPLE WAY
>

And it goes on from there.

2.) The port named: *gcc42*, version 4.2.4 whose Portifle is at:

http://trac.macports.org/browser/trunk/dports/lang/gcc42/Portfile

Now taking the two options above: (1) apple-gcc42 or (2) gcc42, does it
really matter which of these ports satisfies the gnutar's compiler
configuration option for gcc 4.2 on darwin version 9.x (Mac OS X 10.5.x)? Is
the apple-gcc42 source separate from what might be released by XCode 3.1? I
guess that's really Apple's business as to what they release with XCode 3.1
so it may be a moot point, so long as the resource is made separately
available by Apple to the open source communities at large under GPL via
opensource.apple.com

What is the *best* *practice* suggested by the MacPorts community with
respect to compilers? In other words, is it not more or less a
chicken-and-egg situation -- before MacPorts (base) can be created, gcc must
exist and if I am understanding correctly, the gcc version(s) that ship with
XCode (3.0 being the current release and best suggestion). Granted, I could
see that if someone was really paranoid they could download the gcc 4.2
source (prior to MacPorts base installation) and make it separate from what
comes with Xcode true? It wasn't until today, when I investigated this
situation, that I first became aware that there is a port named apple-gcc42.
I wonder if this port is indeed the suggested port to use for building
gnutar 1.20 on a Leopard 10.5.x system?

Thank you,

Tabitha
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.macosforge.org/pipermail/macports-users/attachments/20080622/93fa8af6/attachment.htm