--- trunk/lib/WebPAC/Input.pm	2005/06/25 20:23:23	1
+++ trunk/lib/WebPAC/Input.pm	2005/07/16 11:07:38	3
@@ -5,7 +5,7 @@
 
 =head1 NAME
 
-WebPAC::Input - The great new WebPAC::Input!
+WebPAC::Input - core module for input file format
 
 =head1 VERSION
 
@@ -17,27 +17,49 @@
 
 =head1 SYNOPSIS
 
-Quick summary of what the module does.
+This module will load particular loader module and execute it's functions.
 
 Perhaps a little code snippet.
 
     use WebPAC::Input;
 
-    my $foo = WebPAC::Input->new();
-    ...
+    my $db = WebPAC::Input->new(
+    	format => 'NULL',
+	config => $config,
+    );
+
+    $db->open('/path/to/database');
+    print "database size: ",$db->size,"\n";
+    while (my $row = $db->fetch) {
+	...
+    }
+    $db->close;
 
-=head1 EXPORT
+=head1 FUNCTIONS
 
-A list of functions that can be exported.  You can delete this section
-if you don't export anything, such as for a purely object-oriented module.
+=head2 new
 
-=head1 FUNCTIONS
+Create new input database object.
 
-=head2 function1
+  my $db = new WebPAC::Input( format => 'NULL' );
+
+This function will load needed wrapper module and
 
 =cut
 
-sub function1 {
+sub new {
+        my $class = shift;
+        my $self = {@_};
+	bless($self, $class);
+
+	$self ? return $self : return undef;
+}
+
+=head2 open
+
+=cut
+
+sub open {
 }
 
 =head2 function2
@@ -47,19 +69,42 @@
 sub function2 {
 }
 
-=head1 AUTHOR
 
-Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
 
-=head1 BUGS
+=head1 MEMORY USAGE
+
+C<low_mem> options is double-edged sword. If enabled, WebPAC
+will run on memory constraint machines (which doesn't have enough
+physical RAM to create memory structure for whole source database).
+
+If your machine has 512Mb or more of RAM and database is around 10000 records,
+memory shouldn't be an issue. If you don't have enough physical RAM, you
+might consider using virtual memory (if your operating system is handling it
+well, like on FreeBSD or Linux) instead of dropping to L<DBM::Deep> to handle
+parsed structure of ISIS database (this is what C<low_mem> option does).
+
+Hitting swap at end of reading source database is probably o.k. However,
+hitting swap before 90% will dramatically decrease performance and you will
+be better off with C<low_mem> and using rest of availble memory for
+operating system disk cache (Linux is particuallary good about this).
+However, every access to database record will require disk access, so
+generation phase will be slower 10-100 times.
+
+Parsed structures are essential - you just have option to trade RAM memory
+(which is fast) for disk space (which is slow). Be sure to have planty of
+disk space if you are using C<low_mem> and thus L<DBM::Deep>.
+
+However, when WebPAC is running on desktop machines (or laptops :-), it's
+highly undesireable for system to start swapping. Using C<low_mem> option can
+reduce WecPAC memory usage to around 64Mb for same database with lookup
+fields and sorted indexes which stay in RAM. Performance will suffer, but
+memory usage will really be minimal. It might be also more confortable to
+run WebPAC reniced on those machines.
 
-Please report any bugs or feature requests to
-C<bug-webpac-input@rt.cpan.org>, or through the web interface at
-L<http://rt.cpan.org/NoAuth/ReportBug.html?Queue=WebPAC>.
-I will be notified, and then you'll automatically be notified of progress on
-your bug as I make changes.
 
-=head1 ACKNOWLEDGEMENTS
+=head1 AUTHOR
+
+Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
 
 =head1 COPYRIGHT & LICENSE