Hello all!

We have an application that remotely builds docker images and recently it started to fail to list all images on the host because this line fails to convert the huge buffer (over 500MB of text) into a string.

We don't actually require all the conent from the result (only the id and a few other) but docker api does not offer the possibility to "select" only the relevant fields. One other possibility would be to use https://www.npmjs.com/package/stream-json in order to parse the buffer instead of creating the string and then JSON parsing it. Would you be open to a PR adding this dependency and treating it?