[pacman-dev] [PATCH 2/2] pkgdelta: implement requirments for delta generation

[Dave Reisner]

On Fri, Apr 06, 2012 at 08:56:05PM +0200, Florian Pritz wrote:
> Big deltas or deltas for very small packages are not needed so we should
> check that and not generate any.
> 
> Signed-off-by: Florian Pritz <bluewind at xinu.at>
> ---
>  scripts/pkgdelta.sh.in |   33 ++++++++++++++++++++++++++++++---
>  1 file changed, 30 insertions(+), 3 deletions(-)
> 
> diff --git a/scripts/pkgdelta.sh.in b/scripts/pkgdelta.sh.in
> index 806ae23..ff34d88 100644
> --- a/scripts/pkgdelta.sh.in
> +++ b/scripts/pkgdelta.sh.in
> @@ -30,6 +30,13 @@ declare -r myver='@PACKAGE_VERSION@'
>  
>  QUIET=0
>  
> +# minimal of package before deltas are generated (bytes)
> +min_pkg_size=$((1024*1024))
> +
> +# percent of new package above which the delta will be discarded
> +max_delta_size=80
> +
> +
>  # ensure we have a sane umask set
>  umask 0022
>  
> @@ -46,6 +53,8 @@ This delta file can then be added to a database using repo-add.\n\n")"
>  	echo
>  	printf -- "$(gettext "Options:\n")"
>  	printf -- "  -q                ""$(gettext "quiet\n")"
> +	printf -- "  --min-pkg-size    ""$(gettext "minimal of package before deltas are generated (bytes)\n")"
> +	printf -- "  --max-delta-size  ""$(gettext "percent of new package above which the delta will be discarded\n")"
>  }
>  
>  version() {
> @@ -96,6 +105,13 @@ create_xdelta()
>  	newver="$pkgver"
>  	newarch="$arch"
>  
> +	pkgsize="$(@SIZECMD@ -L "$newfile")"
> +
> +	if ((pkgsize < min_pkg_size)); then
> +		msg "$(gettext "Ignoring small package: %s")" "$pkgsize"
> +		return 0
> +	fi
> +
>  	if [[ $oldname != "$newname" ]]; then
>  		error "$(gettext "The package names don't match : '%s' and '%s'")" "$oldname" "$newname"
>  		return 1
> @@ -119,10 +135,19 @@ create_xdelta()
>  	if (( ret )); then
>  		error "$(gettext "Delta could not be created.")"
>  		return 1
> -	else
> -		msg "$(gettext "Generated delta : '%s'")" "$deltafile"
> -		(( QUIET )) && echo "$deltafile"
>  	fi
> +
> +	deltasize="$(@SIZECMD@ -L "$deltafile")"
> +
> +	if ((max_delta_size * pkgsize / 100 < deltasize)); then
> +		msg "$(gettext "Ignoring big delta: %s vs %s")" "$deltasize" "$pkgsize"
> +		rm -f "$deltafile"
> +		return 0
> +	fi
> +
> +	msg "$(gettext "Generated delta : '%s'")" "$deltafile"
> +	(( QUIET )) && echo "$deltafile"
> +
>  	return 0
>  }
>  
> @@ -134,6 +159,8 @@ while (( $# )); do
>  		-h|--help)    usage; exit 0 ;;
>  		-V|--version) version; exit 0 ;;
>  		-q|--quiet) QUIET=1;;
> +		--min-pkg-size) min_pkg_size=$2; shift;;
> +		--max-delta-size) max_delta_size=$2; shift;;

You never do any checking for these inputs to ensure they're strictly
numeric. You can do this simply with a simple glob test:

  isnumeric() {
    [[ $1 != *[!0-9]* ]]
  }

  ...
  --min-pkg-size)
    if ! isnuermic "$2"; then
      echo "invalid blah blah blah"
      exit 1
    fi
    min_pkg_size=$2
    shift
    ;;
  ...

>  		*) args+=("$1");;
>  	esac
>  	shift
> -- 
> 1.7.9.6
>