[DBIx-Class-Devel] Re: [offtopic] Shortsightedness among developers

[Peter Rabbitson]

On Fri, Dec 14, 2012 at 08:57:50AM -0800, John Napiorkowski wrote:
> I guess I was just just thinking at random, regarding your thoughts on the weirdness of some of the behaviors of important classes, and that partly we have some many oddball ways of doing stuff, maybe that hurts us a bit for important common infrastructure.  and again like with 'base' no one wanted to fix it presumable because it wasn't fun to do so, or hard to do so, or fear of busting the world if it was changed.
> 
> Maybe is not a well formed thought but I wonder if it might be a good idea if we had a more restricted sense of allowed code and API designs for the most important bits of infrastructure.  And communication.  Like until just now I didn't know base was considered ok again.

Actually it isn't considered ok by the masses. What makes me stand out is
that I did multiple NYProfile-ing and benchmarking of large schema loads.
The loading of parent.pm in addition to base.pm (which is always loaded no
matter what), made the code slower even in the case of a Schema with 300
Resultclasses using parent.pm and the underlying deps using base.pm.

Basically the benefit of parent.pm can be realized only if base.pm is never
loaded at all, which in reality will never happen.

Also note that we are talking about differences under 100msecs, so in the
grand scheme of things it is a wash either way.

Cheers