buildx.sh: download tarballs using keepalive

Executing wget like

  wget http://example.com/file1 http://example.com/file2 ...

can do HTTP keepalive.

Quit calling wget with single file. Downloading multiple files in one
http connection is a little bit faster than executing wget more than
100 times. And parallelly execute two instances of wget to increase
download speed.

xorg/X11R7.6/buildx.sh

@@ -23,23 +23,19 @@
 # debian packages needed
 # flex bison libxml2-dev intltool xsltproc xutils-dev python-libxml2 g++ xutils
 
-download_file()
+download_all_files()
 {
-    local file url status
-    file=$1
+    # download files parallelly using keepalive
+    # a little bit faster than calling wget with single file more than 100 times
+    < x11_file_list.txt cut -f1 -d: | sed -e "s|^|${download_url}/|" | \
+        xargs -P2 -n $(expr $num_modules / 2 + 1) \
+        wget \
+          --directory-prefix=downloads \
+          --no-verbose \
+          --timestamping \
+          --continue
 
-    # if we already have the file, don't download it
-    if [ -r downloads/$file ]; then
-        return 0
-    fi
-
-    echo "downloading file $download_url/$file"
-
-    cd downloads
-
-    wget -cq $download_url/$file
     status=$?
-    cd ..
     return $status
 }
 
@@ -74,15 +70,6 @@ extract_it()
         return 0
     fi
 
-    # download file
-    if ! download_file $mod_file
-    then
-        echo ""
-        echo "failed to download $mod_file - aborting build"
-        echo ""
-        exit 1
-    fi
-
     cd build_dir
 
     # if pkg has not yet been extracted, do so now
@@ -263,6 +250,14 @@ if ! NPROC=`nproc`; then
     NPROC=1
 fi
 
+if ! download_all_files; then
+    echo ""
+    echo "download failed - aborting build"
+    echo "rerun this script to resume download/build"
+    echo ""
+    exit 1
+fi
+
 while IFS=: read mod_file mod_dir mod_args
 do
     mod_args=`eval echo $mod_args`