Cache API data

[marnen]

We'll probably cache API data for 24 hours.

• Figure out a cache store (Redis Cloud addon seems like a good bet)
• Set up caching (probably with api_cache and moneta gems)

[marnen]

See also #10.

[marnen]

I'll go with Redis Cloud for right now as planned, though I suspect Google will be a better idea (if nothing else, their free tier gives more space). But the Moneta gem should mostly let us abstract that difference (famous last words, I know).

[marnen]

Since we're doing Redis Cloud for the moment, I guess I'll set up a local Redis instance.

[marnen]

Even better if it works: https://github.com/iainbeeston/faraday_api_cache. If not, I'll fall back to using api_cache manually.

[marnen]

Well, that gem hasn't been touched since 2014, but it looks like https://github.com/sourcelevel/faraday-http-cache is maintained. Trying it.

[marnen]

Nope, that doesn't support current Faraday. Something probably does, but I'll use api_cache manually if I have to.

[marnen]

Looks like the version at sourcelevel/faraday-http-cache#116 should support current Faraday. Let's see if it works...

[marnen]

I think I should first create a Faraday::Connection instance that underlies all API requests, then I can add caching to that.

[marnen]

Unbelievably, it's hard to get the URL from a Faraday request before I make it. Considering switching to a different library (HTTP-rb?) so I can have more control of what requests I'm making.

[marnen]

Incredible. HTTP allows you to build a request and see its URL, but then not fire it as far as I can tell, or to fire it but not see its URL. WTF? Anyway, rest-client may do the trick.

[marnen]

It didn't, but after all that, Typhoeus will. Plus it may be nice to run 30 requests through the hydra at some point.