Gibbler: Git-like Hashes and History for Ruby Objects

http://www.rubyinside.com/gibbler-git-like-hashes-and-history-for-ruby-objects-1980.html

by luccastera • 17 years ago

11 5 17 years ago

rubyinside.com

jrockway3• 17 years ago

We use this pattern for immutable objects in our Perl applications, although I think our implementation is cleaner:

http://github.com/nothingmuch/kiokudb/blob/96f9a380c715b7aa1...

delano2• 17 years ago

Interesting! I'm not familiar with KiokuDB or Moose so I hope you don't mind a couple questions:

- what is the return value of $self->data in the example you link to?

- do you store the digest or calculate it each time?

- where can I find an example of the digests being used? I looked around the repo a bit but couldn't find one.

jrockway3• 17 years ago

$self->data returns the value of the data attribute. The class is responsible for generating the identity of instances.

The digest is only calculated when the object is stored. When it is retrieved and stored again (a no-op, since it's immutable), the ID is remembered and not recalculated. (Creating an identical new object will result in the digest being calculated again, of course.)

The basic pattern for using this, once you've written the class:

    my $foo = Class->new( ... );
    my $id = $db->store($foo); # 29837a387bffe...
    my $foo2 = $db->lookup('29837a387bffe...');

Then this is true (the linker keeps note of the in-memory object IDs):

    refaddr $foo == refaddr $foo2

Gibbler: Git-like Hashes and History for Ruby Objects

Related Stories