[Catalyst] Non-real world irrelevant benchmarks

[Matt S Trout]

catalyst.20.chsg at spamgourmet.com wrote:
> In victori's remarks, he calls for a change in Catalyst and points to 
> the other advantages to to this framework, mostly related to ease of 
> coding.  While the whole reason I came to Catalyst is because I'm 
> comfortable with Perl and don't want to learn Ruby, I'm worried that my 
> Catalyst application won't perform as well when/if my app usage becomes 
> very significant.  Should I be concerned?

No. Six apart certainly aren't, given http://vox.com/ is a Catalyst app.

> Again, I'm not interesting in hearing about how Rails/Ruby/Django/Python 
> sucks, but in facts about real performance of Catalyst.

The reality is that victori's benchmark successfully measures only the base 
overhead of a request in an app with very few URL endpoints; Catalyst's 
dispatch mechanism is optimised for speed of dispatch for large applications 
and for allowing complex dispatch logic elegantly.

Besides which, I've never yet seen a production application (and between 
Shadowcat's client portfolio I've seen not a small number thereof) where the 
dispatch overhead was even statistically significant to the overall 
performance - the bottleneck has always been either data retrieval or template 
rendering.

-- 
      Matt S Trout       Offering custom development, consultancy and support
   Technical Director    contracts for Catalyst, DBIx::Class and BAST. Contact
Shadowcat Systems Ltd.  mst (at) shadowcatsystems.co.uk for more information

+ Help us build a better perl ORM: http://dbix-class.shadowcatsystems.co.uk/ +