At 07:29 AM 9/19/00 -0700, Dave Storrs wrote:
> I guess, if I had to write an explanation of pack/unpack based on
>my limited understanding, it would be something like:
>
> "Unpack takes binary data in some particular format and
>disassembles it, assigning various pieces of it to variables according to
>formatting that you supply. Pack does the opposite, using your supplied
>formatting to crunch Perl scalar variables into binary data that is
>represented in some specific way. The binary data used by (un)pack will
>belong to exactly one type of C numeric variable, meaning that it will be
>limited in what kinds of numbers it can store and how it will represent
>them."
>
> Is this definition completely off-base?
I think you are on track, but I would probably have defined it the other
way around:
"The perl function pack() takes Perl values and converts them to a compact
binary representation based on a user-specified template. This is useful
for communicating with external data formats that expect data to be in a
particular binary format that is not native to perl. The perl function
unpack() takes a compact binary representation of data and converts it to
Perl values based on a user-specified template. Pack() and unpack() are
inverse-functions of each other: if used with the same template, unpack can
be used to recover data that has been packed.
Example:
$Tudpheader = "nnnn"; # source, destination, length, checksum
$Tudppseudoheader = "NNxcn"; # source IP, dest IP, protocol (must be 17),
length
$udplength = length($data)+8;
$udpcheck = (pack $Tpseudoheader, $sip,$dip,17, $udplength)
. $(pack $Tudbheader,$sport,$dport,$udplength, 0)
. $data;
$checksum = udpchecksum($udpcheck);
$udppacket = (pack $Tudbheader,$sport,$dport,$udplength,$checksum) . $data
"
> Dave
