in reply to Re^4: Data Structures
in thread Data Structures
What about the argument for using object oriented code...
Don't. Three reasons.
But you can already do that with a hash, without the extra effort.
The reasoning is that it will allow you to change your structure later, if necessary, and so save time.
But expending effort now to save time for an eventuality that may never happen, that will take more memory and run more slowly, is total folly.
Expend the effort when you know you need to and when you know in what way things have to chage.
while( <FILE> } { my( @attributes = unpack '...', $_; my $object = Siesmic::Line->new( @attributes ); ## Now what? ## ???
Where are you going to put you objects so that you can retrieve them when you need them??
### IN A HASH OF COURSE!!!! $lines{ $object->getLineNo() }{ $object->getStnNo() } } = $object; }
So now, you've still got the HoHs in order to find the one you want, but instead of each leaf being an small anonymous array, it's an object (or nest of objects) that requires more memory, runs more slowly, uses clumsy unfamiliar syntax, and requires at least 5 times more effort to develop.
For what? Just in case?
My advice is to stick with the HoHoAs. If the needs change, you can make objects later, but you'll still need the container to hold them and find the one you want.
|
---|
Replies are listed 'Best First'. | |
---|---|
Re^6: Data Structures
by YYCseismic (Beadle) on May 02, 2008 at 21:26 UTC |