Re: [PATCH 01/12] test-scripts: test migration scripts

[Dave Chinner <david@fromorbit.com>]

On Fri, Nov 13, 2015 at 01:36:50PM -0800, Darrick J. Wong wrote:
> Add two scripts: "nextid" finds the next available test ID number in a
> group, and "mvtest" relocates a test, fixes the golden output, and
> moves the group entry for that test.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  mvtest |   58 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  nextid |   35 +++++++++++++++++++++++++++++++++++

These should be placed in the "tools" directory.

>  create mode 100755 mvtest
>  create mode 100755 nextid
> 
> 
> diff --git a/mvtest b/mvtest
> new file mode 100755
> index 0000000..b5406d1
> --- /dev/null
> +++ b/mvtest
> @@ -0,0 +1,58 @@
> +#!/bin/sh
> +
> +# Renumber a test
> +
> +if [ -z "$1" ] || [ "$1" = "--help" ]; then
> +	echo "Usage: $0 path_to_test new_path_to_test"
> +	exit 1
> +fi
> +
> +src="$1"
> +dest="$2"
> +
> +die() {
> +	echo "$@"
> +	exit 1
> +}
> +
> +nsort() {
> +	sort -g < "$1" > "$2"
> +}
> +
> +append() {
> +	out="$1"
> +	shift
> +	echo "$@" >> "${out}"
> +}
> +
> +test "${src}" != "${dest}" || die "Test \"${src}\" is the same as dest."
> +test -e "tests/${src}" || die "Test \"${src}\" does not exist."
> +test ! -e "tests/${dest}" || die "Test \"${src}\" already exists."
> +
> +sid="$(basename "${src}")"
> +did="$(basename "${dest}")"
> +
> +sgroup="$(basename "$(dirname "tests/${src}")")"
> +dgroup="$(basename "$(dirname "tests/${dest}")")"
> +
> +sgroupfile="tests/${sgroup}/group"
> +dgroupfile="tests/${sgroup}/group"
> +
> +$DBG git mv "tests/${src}" "tests/${dest}"

$DBG?

....
> +$DBG sed -e "/^${sid}.*$/d" -i "${sgroupfile}"
> +$DBG cp "${dgroupfile}" "${dgroupfile}.new"
> +$DBG append "${dgroupfile}.new" "${newgrpline}"
> +$DBG nsort "${dgroupfile}.new" "${dgroupfile}"

What does this do to comments in the group file?

...

> diff --git a/nextid b/nextid
> new file mode 100755
> index 0000000..285b549
> --- /dev/null
> +++ b/nextid
> @@ -0,0 +1,35 @@
> +#!/bin/sh
> +
> +# Given a group name, find the next available test number.
> +
> +if [ -z "$1" ] || [ "$1" = "--help" ]; then
> +	echo "Usage: $0 groupname[/start_looking_at_this_number]"
> +	exit 1
> +fi
> +
> +die() {
> +	echo "$@"
> +	exit 1
> +}
> +
> +if [ "$(basename "$1")" != "$1" ]; then
> +	group="$(dirname "$1")"
> +	id="$(basename "$1")"
> +else
> +	group="$1"
> +	id=1
> +fi
> +test -e "tests/${group}/group" || die "Unknown group \"${group}\"."
> +
> +while test "${id}" -lt 1000; do
> +	name="$(printf "%.03d" "${id}")"
> +	if [ ! -e "tests/${group}/${name}" ]; then
> +		echo "${group}/${name}"
> +		exit 0
> +	fi
> +	id=$((id + 1))
> +done
> +
> +echo "No free IDs less than ${id} in group \"${group}\"."

So the "new" script does this differently, by reading the group
file and looking for the first non-contiguous ID in the file. It
doesn't need scan ID limits because EOF triggers that. Wouldn't it
be better to to factor the code out of the "new" script to find the
next id, and then have the new script call that?

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com