The new dbscripts do not need this. All info can
be gathered from the $CARCH variable in the filename.

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

This is a very very hackish way to deal with rebuilds for
right now.

The intent is the following:
    First build foobar-1.2
    makechrootpkg -i foobar-1.2-1-i686.pkg.tar.gz
    Now build packages that depend on foobar-1.2

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Sometimes makepkg fails, but a package is built - for
instance, when makepkg cannot remove deps. In this case we
will attempt to move the pkg/src files regardless.

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

It should only be run from a trunk dir containing a PKGBUILD

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Add a message that explains what to do when a merge conflict is detected and needs to be dealt with.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

The latest incarnations of unionfs have issues appending
text via shell redirection, so to work around it, we can
touch the file before-hand to move it to the RW portion
of the union.

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

I thought it was tracked by default

This prevents the copying of local files and only
copies versioned files

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

This was accidentally scrubbed, breaking 64bit uploads

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Because we need a root user to chroot, when we drop
permissions to 'nobody', we're still using HOME=/root

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

If only you could have seen the facepalming going on...

Signed-off-by: Simo Leone <simo@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>

This change will remove 4 needless sed lines from the PKGBUILD we use on
Arch for building this package and do the logic locally where it probably
makes more sense anyway. $CARCH should always be present as we fail if we
cannot find a correct makepkg.conf file in /etc/.

Signed-off-by: Dan McGee <dan@archlinux.org>

Checking /proc/filesystems for unionfs is
friendlier for kernels with unionfs builtin.
Also, a total lack of unionfs now causes
makechrootpkg to bail out.

Signed-off-by: Simo Leone <simo@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>

Copying the configs before installing stuff
caused conflicts. Running the host's pacman
uses the host's configuration anyway, so
there is no need to copy them in this case.

Signed-off-by: Simo Leone <simo@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Reported-by: Jaroslaw Swierczynski <swiergot@juvepoland.com>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Use basename when checking for files in the source array

This allows us to actually check for pre-downloaded
remote files and copy them to the build root as well.

Original-work-by: Jaroslaw Swierczynski <swiergot@juvepoland.com>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

No longer needed in pacman 3.1

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Clear when creating the dir, and additionally DO not clear it if the user is
attempting to repack the package.

Original-work-by: Jaroslaw Swierczynski <swiergot@juvepoland.com>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Make the call to scp use $(basename $pkgfile) so that we don't get oddness.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

This keeps the pkgurl variable consistently fully-qualified, so that whenever we
use that variable we know it points straight at the package file.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Now, checkpkg and extrapkg will fail if /etc/makepkg.conf doesn't exist, and will also try to load in personal user's settings in ~/.makepkg.conf

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

My recent commit adding the -c option and renaming the old -c to -r accidentally left out the new '-r' option from the getopts call.  Additionally, add ${pkgrel} to the mv line after building to be more specific about that file to move.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

This change allows all packager-defined settings (such as PACKAGER or OPTIONS) to propogate into the chroot when building packages.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

This ensures that sources and packages built inside makechrootpkg go where
the user expects them to, as defined in their /etc/makepkg.conf and
~/.makepkg.conf files.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

The old -c option, which specified a 'chroot', has been renamed to -r, since
most other scripts use -r for a different root.  A new -c option is
introduced, which cleans all files from ${chrootdir}/rw before beginning,
thus ensuring a clean chroot.

Signed-off-by: Travis Willard <travis@archlinux.org>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Fix an issue where a missing add directory causes the package to be uploaded as
a file named "add"

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Resolv.conf changes, like mtab, so we can't just trust it to copy on build only

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Allows us to run 'make install' even if the symlinks exist (helpful for testing
locally)

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>

Add the same vim modeline to all the files, as well as cleanup the newly
added scripts a bit. If you look at this diff with the -w option, you'll see
it really isn't all that significant.

Signed-off-by: Dan McGee <dan@archlinux.org>