I agree. When it comes to ruby, System.fork is the way to go. Ruby has taught me (forced me?) to think more about parallel processes more than multi-threading. Separating different jobs into their own process has yielded great results.

On the other hand, forking works well & it’s at least as easy to implement as threads–for me, it’s probably a better solution anyway, as I’m building a distributed app using RabbitMQ. I have long-running workers that will have happier lives if the actual work is done in a subprocess, & the aws-s3 gem plays nicely with my low-memory-footprint requirements by allowing all uploads & downloads to stream.

Sure, I could rewrite Amazon’s code to handle streaming, and it’d definitely be easier than trying to make the aws-s3 gem thread-safe, but sometimes, when I find a fork in the road, I take it, y’know?

FWIW.

From what I understand you store an object with the name “foo/bar/baz” and then your can list all objects in that “dir” by giving the parameter “foo/bar/” which lists all objects that start with that string. The slash isn’t special in any way. My impression is that this is done on the server side (Amazon side), and you therefore *have* directories.

My problem is with the number of characters or the max length of file name that S3 bucket can have.

Any ideas?