The following is no duplicate of How to determine the size of a package while using apt prior to downloading?, it is more an additional question.

Consider the following two implementations to get the required disk space:

#!/bin/bash

echo n | sudo apt-get install $LIST | grep "disk space" | awk '{ print $4, $5 }'

and

#!/bin/bash



PACKAGE_LIST=$(dpkg --get-selections | awk '{ print $1 }' | grep -v -e "-dbg" | cut -f1 -d":")



getsize () {

   size=$(apt-cache --no-all-versions show $1 | grep Installed-Size | awk '{ print $2 }')

   ((NEEDED+=$size))

}



for package in $LIST; do

   getsize $package

   DEPENDENCIES=$(apt-cache depends $package | grep Depends | awk '{ print $2 }')

   for dependency in $DEPENDENCIES; do

       if [[ ! $PACKAGE_LIST =~ [^.[:alnum:]-]"$dependency"[^.[:alnum:]-] ]]; then

          getsize $dependency

       fi

   done

done



echo "$NEEDED kb"

In theory they should return similar results, but for big packages in $LIST they return very different results, the second implementation returns much bigger results (for example 13.5 GB vs 16.2 GB).
I know the second implementation is slower, because it tries to do the job of apt-get in bash, which is a lot slower, but my intention of writing the second version was to get the result in kb, because I was not able to find out how to get the result in kb from the first implementation, mainly because apt-get scales the unit, from bytes up to gigabytes and I do not know, whats the biggest unit it considers.

Could someone help me find out why there is such a big difference?