Using the Inflector comes with a big performance hit

[ahx]

I noticed a significant performance increase when not using Inflector#decode_type (no matter if I use dry-inflector (#36) or Inflecto)
In other words: Using the inflector seems to come with a pretty big cost.

🤔

I can provide a performance test if you are interested.

[doughsay]

I'm curious what you might be doing that causes the performance of inflecting to overshadow the latency of actually performing http requests to a server. It seems to me that any inflecting algorithm, even a slow one, wouldn't compare at all to the latency and performance of contacting external servers. So I'm basically wondering why it matters that it might be a bit slow?

On the other hand: if we really wanted to remove the inflector altogether, it would be harder to follow the resource naming conventions that jsonapi recommends. That may or may not be a deal-breaker for this library, I'd have to do some more research on how I'm using it to see how it would affect things.