[Dbix-class] Using Class::Std under-the-hood

[Adrian Howard]

On 11 Aug 2005, at 08:05, Dan Kubb wrote:
>> What are the performance implications of using Class::Std?
>
> From Class::Std docs:
>
>   "Of the several popular methods of reliably enforcing  
> encapsulation in Perl,
>   inside-out objects are also by far the cheapest. The run-time  
> performance of
>   inside-out classes is effectively identical to that of regular  
> hash-based
>   classes. In particular, in both schemes, every attribute access  
> requires only
>   a single hash look-up. The only appreciable difference in speed  
> occurs when
>   an inside-out object is destroyed."
>
> And later Damian says:
>
>   "There is a simple, convenient, and utterly secure way to prevent  
> client code
>   from accessing the internals of the objects you provide. Happily,  
> that approach
>   also guards against misspelling attribute names (a common error  
> in hash-based
>   classes), as well as being just as fast as--and often more memory- 
> efficient
>   than--ordinary hash-based objects."
>
> As fast as -- and *more* memory efficient. ;)

</lurk>

Well - yes and no.

Since 5.004 perl is bright enough to only store hash keys once no  
matter how many hashes use them - so the overhead of having duplicate  
hash keys in each object disappears. Depending on what you use as the  
hash index inside-out objects can be a little bit smaller or a little  
bit larger than normal blessed hash objects - but nothing to write  
home about.

As for speed, depends on how you index the attribute hashes. If you  
just use the string $self then attribute access can be quite slow -  
about half the speed of normal hash based objects in some situations.  
If you do something moderately clever like caching refaddr $self in $ 
$self you can get access speeds that are of the same order as normal  
blessed hash objects.

As for Class::Std - remember this does a /lot/ more than just inside- 
out objects. I've just tacked on an example to a very basic benchmark 
[1] I used when looking at different ways to do inside-out objects  
and it comes in at about half the speed of blessed hash objects and  
hand rolled refaddr caching inside-out objects.

That said any large system will probably end up needing some of the  
extras that Class::Std provides - so will get the speed hit in any case.

Cheers,

Adrian

[1] <http://www.quietstars.com/perl/object-bench.pl> for those who  
care :-)

<lurk>