Question

trying to wrap my head around some basic async programming concepts.

Right now I essentially have two functions, for now lets call them getOne and getAll. getAll essentially loops of a series and makes a call to getOne, then maps it to a hash and sends it back down the chain. Here is basically what I have, in coffeescript:

getOne = (key, callback) ->
    try
        # find object referenced by key
        object = "blah blah"
        callback(null, object)
    catch(error)
        callback(error, null)

getAll = (keys, callback) ->
    results = {}
    try
        count = keys.length
        for i, key of keys
            count--
            getOne key, (err, result) ->
                if(err)
                   # do something
                else
                   results[key] = result
                if count is 0
                   callback "", results
    catch
        callback error, results

The above code essentially outputs a hash where the keys don’t necessarily lineup with the expected data, which is expected the way a for loop is async processed. However, I can’t seem to come up with a solution that doesn’t block the loop and cause everything to be processed in a series instead of maintaining parallel processing.

What am I missing?

Answer 1

There isn’t anything inherently asynchronous about the code you’ve written. If you’re worried about the callbacks in getOne taking a long time to execute, you have a couple of options.

• use the async library
• use process.nextTick to defer execution of each iteration of your for loop

The first option is definitely easier.

The second option involves rewriting your getAll function as a recursive function (instead of using for) and wrapping recursive subcalls in process.nextTick.

This article is reasonably useful for understanding how to use process.nextTick to accomplish this: Understanding process.nextTick()