Dedicated handling of AWS Lambda environment

[medikoo]

Recently I'm working with projects that stand on Serverless architecture (AWS Lambda), and found that debug doesn't play well with that.

Thing is that in AWS Lambda ( http://docs.aws.amazon.com/lambda/latest/dg/nodejs-prog-model-logging.html ) all console.* methods are internally decorated, so then all logs are combined with meaningful info on time and invocation id), so outcome of console.log("this is console log"); etc. ends as:

Now debug (as run at v2.6.8) by defaults logs as:

This patch ensures that eventual debug in case of being run in AWS Lambda outputs logs via console.error therefore comes out as following:

[coveralls]

Coverage decreased (-0.7%) to 63.095% when pulling b80437c on medikoo:aws-lambda-friendly into a45d4a0 on visionmedia:master.

[medikoo]

Coverage decreased (-0.7%)

I can configure some tests that try to emulate AWS Lambda env, and run basic tests, but it may need some sophisticated handling in a module (e.g. module cache invalidation). Let me know

[coveralls]

Coverage increased (+0.9%) to 73.585% when pulling 196d518 on medikoo:aws-lambda-friendly into 13e1d06 on visionmedia:master.

[medikoo]

Updated for v3

[medikoo]

Closing this, for following reasons:

• After Use Date#toISOString() instead to Date#toUTCString() when output is n… #418 got in, the default output differs only in not having invocation id. which is not that relevant, as there's no concurrency within single lambda in CloudWatch we can clearly deduct which invocation id applies to which debug line).
• It's too custom, which seems bad to have hard-coded in. it's best if addressed by external tweaking, which probably is doable now and if not, should be after major overhaul is done.