I\'m relatively new to RoR and I\'m curious about why Rails compiles assets both with and without md5 hash for production?
I run bundle exec rake assets:clean<
The reason it does it is so that you can access the files without knowing the MD5 fingerprint (for example in a non-rails application, or a file within the rails app which isn't compiled or run by the rails stack (e.g. a 500/502 status error page). In this case you would have to compile the assets then change the css/js links in the static HTML files each time you updated the code (thus causing a change in the MD5 hash).
So instead rails produces 2 copies of each asset file, one with the fingerprint in the filename, the other without (e.g. application-731bc240b0e8dbe7f2e6783811d2151a.css, and application.css). The fingerprinted version is obviously preferred (see 'what is fingerprinting and why should I care' in the rails asset pipeline guide). But the non-digested version is there as a fallback.
As a final thought on the matter I'd take a read of the following pull request to the rails git repo: https://github.com/rails/rails/pull/5379 where they are discussing the pros and cons of the non-digested filenames, and the possibility of being able to turn off compilation of them.
HTH