On 26/01/14 03:38, Maxime Gauduin wrote: > Add a proto for bzr, git, hg and svn PKGBUILDs and remove the > corresponding protos from abs. >
Commit message needs updated. > Signed-off-by: Maxime Gauduin <[email protected]> > --- > proto/PKGBUILD-vcs.proto | 95 > ++++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 95 insertions(+) > create mode 100644 proto/PKGBUILD-vcs.proto > > diff --git a/proto/PKGBUILD-vcs.proto b/proto/PKGBUILD-vcs.proto > new file mode 100644 > index 0000000..3aa7890 > --- /dev/null > +++ b/proto/PKGBUILD-vcs.proto > @@ -0,0 +1,95 @@ > +# This is an example PKGBUILD file. Use this as a start to creating your own, > +# and remove these comments. For more information, see 'man PKGBUILD'. > +# NOTE: Please fill out the license field for your package! If it is unknown, > +# then please put 'unknown'. > + > +# The following guidelines are specific to BZR, GIT, HG and SVN packages. > +# CVS and DARCS are not natively supported in pacman yet, please refer to the > +# dedicated prototypes instead. > + > +# Maintainer: Your Name <[email protected]> > +pkgname=NAME-VCS # '-bzr', '-git', '-hg' or '-svn' > +pkgver=VERSION > +pkgrel=1 epoch= > +pkgdesc="" > +arch=() > +url="" > +license=('GPL') > +groups=() > +depends=() > +makedepends=('VCS_PACKAGE') # 'bzr', 'git', 'mercurial' or 'subversion' checkdepends=() optdepends=() > +provides=("${pkgname%-VCS}") > +conflicts=("${pkgname%-VCS}") > +replaces=() > +backup=() > +options=() > +install= changelog= > +source=('DIR_NAME::VCS+REPO_URL#FRAGMENT') > +noextract=() > +md5sums=('SKIP') > + > +################################################################################ > + Can we just refer to the "Using VCS Sources" in the PKGBUILD man page here? > +# DIR_NAME: Use to change the source directory name if needed > + > +# VCS: Use to specify the VCS type, not needed when REPO_URL is explicit (for > +# example git://URL, svn://URL or lp:REPO_NAME) > + > +# FRAGMENT: Use to pull a specific branch, commit/revision or tag > +# Bazaar accepts the 'revision' keyword > +# Git accepts the 'branch', 'commit' and 'tag' keywords > +# Mercurial accepts the 'branch', 'revision' and 'tag' keywords > +# Subversion accepts the 'revision' keyword > + > +# Examples: > +# source=("lp:${pkgname%-bzr}#revision=42") > +# > source=("git+https://github.com/author/${pkgname%-git}.git#branch=unstable";) > +# source=('hg+https://bitbucket.org/author/project#tag=2.0') > +# source=("${pkgname%-svn}::svn://svn.code.sf.net/p/project/code/trunk"} > + > +################################################################################ > + > +pkgver() { > + cd "$srcdir/${pkgname%-VCS}" > + > +# The examples below are not absolute and need to be adapted to each repo. > The > +# general idea is to have VERSION='VER_NUM.rREV_NUM.HASH' or any subset in > case > +# VER_NUM or HASH are not available. Why are we using "r" everywhere? What is its advantage - especially in the non-git cases? And why adding "." between the segments instead of an underscore? I have seen a lot of software release version A.B followed by A.B.C, so separating with a "." will break. > +# Bazaar > + printf "r%s" "$(bzr revno)" > + > +# Git, tags available > + printf "%s" "$(git describe --long | sed 's/\([^-]*-\)g/r\1/;s/-/./g')" > + > +# Git, no tags available > + printf "r%s.%s" "$(git rev-list --count HEAD)" "$(git rev-parse --short > HEAD)" > + > +# Mercurial > + printf "r%s.%s" "$(hg identify -n)" "$(hg identify -i)" > + > +# Subversion > + printf "r%s" "$(svnversion | tr -d 'A-z')" > +} > + > +prepare() { > + cd "$srcdir/${pkgname%-VCS}" > + patch -p1 -i "$srcdir/${pkgname%-VCS}.patch" > +} > + > +build() { > + cd "$srcdir/${pkgname%-VCS}" > + ./autogen.sh > + ./configure --prefix=/usr > + make > +} > + > +check() { > + cd "$srcdir/${pkgname%-VCS}" > + make -k check > +} > + > +package() { > + cd "$srcdir/${pkgname%-VCS}" > + make DESTDIR="$pkgdir/" install > +} >
