On Sun, Aug 15, 2010 at 11:21 AM, John Blinka <john.bli...@gmail.com> wrote:
> Attached is a 3.1.2 ebuild.  It allowed me to build and install
> amanda-3.1.2 and to complete a successful backup.  I know nothing
> about writing ebuilds, so this is nothing more than a very crude and
> shameless and uninformed hack of the 2.6.1_p2 ebuild.  There's very
> little testing behind it.  But it worked for me.

Here are the diffs for the ebuild you posted:

--- /usr/portage/app-backup/amanda/amanda-2.6.1_p2.ebuild
2010-05-08 15:08:17.000000000 -0500
+++ /home/dustin/Downloads/amanda-3.1.2.ebuild  2010-08-15
12:07:43.000000000 -0500
@@ -10,7 +10,7 @@
 SRC_URI="mirror://sourceforge/amanda/${MY_P}.tar.gz"
 LICENSE="as-is"
 SLOT="0"
-KEYWORDS="~amd64 ~ppc ~ppc64 ~sparc ~x86"
+KEYWORDS="~amd64 ~ppc ~ppc64 ~sparc x86"
 RDEPEND="sys-libs/readline
        virtual/inetd
        sys-apps/gawk
@@ -18,7 +18,7 @@
        >=dev-lang/perl-5.6
        app-arch/dump
        net-misc/openssh
-       >=dev-libs/glib-2.2
+       >=dev-libs/glib-2.24.0
        nls? ( virtual/libintl )
        s3? ( >=net-misc/curl-7.10.0 )
        samba? ( net-fs/samba )
@@ -118,10 +118,10 @@

        # Fix a fun race condition if you use encryption.
        # This is one of the reasons you should test your recovery
procedures often.
-       epatch "${FILESDIR}"/${PN}-2.6.0p2-amcrypt-ossl-asym-race-fix.patch
+       #epatch "${FILESDIR}"/${PN}-2.6.0p2-amcrypt-ossl-asym-race-fix.patch

        # gentoo bug 248838, check /sbin stuff before /bin
-       epatch "${FILESDIR}"/${PN}-2.6.0_p2-syslocpath.patch
+       #epatch "${FILESDIR}"/${PN}-2.6.0_p2-syslocpath.patch

        eautoreconf

The change in glib dependency is not required - Amanda will work
against older glibs without any problem (we test this).  Bug 331111 is
a problem with *older* (<=2.6.1) Amanda and *newer* glib (>=2.24.0).

Both of the patches you've commented out should still apply.  If I
understand the amcrypt-ossl-asym patch correctly, the underlying bug
in Amanda has already been fixed (server discards data sent before
headers), but the patch doesn't hurt.  The syslocpath patch is
probably still required.  On the other hand, I notice that the ebuild
Jeroen attached for 3.1.1 makes a *lot* of changes, including
commenting out these patches.

Dustin

-- 
Open Source Storage Engineer
http://www.zmanda.com

Reply via email to