Extracting caching layer

[cameron-martin]

Would you accept a PR for extracting the redis caching into it's own gem, as a pluggable layer?

It doesn't seem right having redis as a hard-coded dependency of this gem, and doing the above would open up the possibility of using other caching strategies.

I also wanted to create something to automatically combine the requests which take multiple ids occuring in a certain timeframe, and split them back up again, transparently to the api consumer. Adding the above infrastructure would make this easy.

I wanted to get your approval before doing it, since this is a non-trivial change. Let me know what you think.

[intinig]

For now I don't mind having redis as a dependency, since it will install even if you have no redis on your box.

An extraction could be a good thing in the future but for now it seems ovecomplication, also because it's written in a way that already supports eventual new caching layers.

Regarding the request splitting and merging I don't think it's useful. Riot will increase rate limits without problems if you need them, and this would be complicating the library a little bit too much.

[cameron-martin]

Yeah fair enough, I did some memory profiling and loading the redis gem only uses 5MB of extra memory. For this amount of memory savings, its not worth it.

Regarding the request splitting, I wasn't proposing adding it to this library, but implementing it in another gem in the same way that the caching layer would. However, I agree that this would probably be pointless; ideas always seem better when you first come up with them.