Tie::Discovery - Perl extension for `Discovery' hashes
A `discovery' hash is a hash that's designed to help you solve the data
dependency problem. It's based on the principle of least work; some
times, you may spend a lot of time in your program finding out paths,
filenames, operating system specifics, network information and so on
that you may not end up using. Discovery hashes allow you to get the
data when you need it, and only when you need it.
To use a discovery hash, first tie a hash as shown above. You will want
to keep hold of the object returned by `tie'. You can then add things to
discover by calling the `register' method as shown above. The above code
`$obj->register("os",\&discover_os);' means that when (and only when!)
the value `$config{os}' is fetched, the sub `&discover_os' will be
called to find it. The return value of that sub will then be cached to
save a look-up next time.
The real power comes from the fact that you may refer to the tied hash
inside of the discovery subroutines. This allows for fast, neat and
flexible top-down programming, and helps you avoid hard-coding values.
For instance, let us find the OS by calling the uname program.
$obj->register( os => sub { `$config{path_to_uname}` } );
Now we need code to find the program itself:
use File::Spec::Functions;
$obj->register( path_to_uname => sub {
foreach (path) {
return catfile($_,"uname") if -x catfile($_,"uname")
}
die "Couldn't even find uname"
};
Fetching `$config{os}' may now need a further call to fetch
`$config{path_to_uname}' unless the path is already cached. And, of
course, we needn't stop at two levels.
Fuller examples of this technique are seen in the forthcoming CTAN
and SysConf modules.
