contrib/packagelib.sh

perf: clear revlog caches on every iteration

cmdutil.openrevlog() may return a cached revlog instance. This /may/
be a recent "regression" due to refactoring of the manifest API. I'm
not sure.

Either way, this perf command was broken for at least manifests because
subsequent invocations of the perf function would get cache hits from
previous invocations, invalidating results. In the extreme case,
testing the last revision in the revlog resulted in near-instantanous
execution of subsequent runs (since the fulltext is cached). A time
of ~1us would be reported in this case.

# Extract version number into 4 parts, some of which may be empty:
#
# version: the numeric part of the most recent tag. Will always look like 1.3.
#
# type: if an rc build, "rc", otherwise empty
#
# distance: the distance from the nearest tag, or empty if built from a tag
#
# node: the node|short hg was built from, or empty if built from a tag
gethgversion() {
    make cleanbutpackages
    make local || make local PURE=--pure
    HG="$PWD/hg"

    $HG version > /dev/null || { echo 'abort: hg version failed!'; exit 1 ; }

    hgversion=`$HG version | sed -ne 's/.*(version \(.*\))$/\1/p'`

    if echo $hgversion | grep + > /dev/null 2>&1 ; then
        tmp=`echo $hgversion | cut -d+ -f 2`
        hgversion=`echo $hgversion | cut -d+ -f 1`
        distance=`echo $tmp | cut -d- -f 1`
        node=`echo $tmp | cut -d- -f 2`
    else
        distance=''
        node=''
    fi
    if echo $hgversion | grep -- '-' > /dev/null 2>&1; then
        version=`echo $hgversion | cut -d- -f1`
        type=`echo $hgversion | cut -d- -f2`
    else
        version=$hgversion
        type=''
    fi
}