Question

Context

• I have an app that does time-consuming processing of texts in Ruby.
• When a users submits a text, a background process is forked to take care of that text.
• This process continuously outputs new parts of the text as they have been processed.
• Meanwhile, the user is redirected to the page where the processed text will be output.

My question is: how can the client’s Javascript communicate with a process running on the server, given its PID? Is that possible? Do I need to wrap each background process inside its own servlet running on its own port?

Answer 1

Here is the strategy I ended up using:

• Create a small Sinatra servlet to manage the different jobs.
• Each job request received is assigned a job ID and started in a detached forked process.
• The forked process sets up a DRb service with an array that is used as a stack.
• It then performs a “GET” request on the servlet that associates the job ID with the DRb URI.
• As the process continues execution in the background, it adds elements to the stack.
• The main thread can be polled to return the next element in the stack.
  • Connection to the remote array is established within the servlet, and an element is shifted from the stack.
  • Polling continues until there are no more elements to add on the stack and no more elements to return to the client.
• When the stack is empty, the process terminates.

Any thoughts?