Cache processing results

[asiniy]

Now all processed files (especially images) are generated on each query.

It's not difficult if you have only couple of images to be rendered on the page. But this becomes a problem when you have a webpage which requires 20-30 images to be generated simultaneously. They are returned one after another, even on powerful VPN.

In my opinion, exfile should have a cache ability, like:

config :exfile, Exfile,
  cache: %{
    size: '10GB',
    path: '/tmp/exfile-other' # default is `/tmp/exfile-cache-#{env}`
  }

All processed images will be stored to path and will have the mark of their earlier usage assigned, and there also will be a supervisor which monitors size of the path folder. If it oversizes limit, it will destroy the oldest unused images in it.

@keichan34 WDYT?

[keichan34]

Sorry about the late reply on this.

Since files are fingerprinted and processing inputs are in the URL, the requests can be cached by a normal HTTP cache (nginx and/or any pull-based CDN like CloudFront, CloudFlare, Fastly, etc). I don't think it's in the scope of Exfile to cache outputs when a HTTP cache can do it equally as well.

[asiniy]

@keichan34 interesting idea! Let me test it, I'll response in ~ 1 week

[scarfacedeb]

@asiniy you could also put nginx in front of elixir app and cache exfile responses with it.