Reverse log level order for all levels so that syslog can be used again

[mheap]

Fixes #249. As syslog level numbers now conform to the RFC, we need to reverse the order of the NPM and CLI loggers too, as well as the comparison which checks levels before logging

[mheap]

Any idea if this can be merged in?

[3rd-Eden]

@mheap Could you add a test to prevent regression as this was already once broken before?

[mheap]

@3rd-Eden Had to do a little refactoring to get the data I needed out, but I've added three tests, one for each supplied set of config levels

[3rd-Eden]

@mheap It seems that some of your changes causes the test suite to fail on Node 0.6

[mheap]

Runs fine for me on 0.6.21 locally. According to Travis, the build stalled rather than failing.
Anyone with access can restart the build at https://travis-ci.org/flatiron/winston/builds/10067336

Edit: Just spotted that it was restarted 2 hours ago. Not sure what's going on there.

[mheap]

@3rd-Eden Any ideas what might be causing the build to stall on Travis?

[mheap]

@3rd-Eden Ping. Can we merge this anyway? It's a blocker for a lot of people that are using syslog levels

[3rd-Eden]

@mheap I'm not merging this in until the tests are passing. I simply don't want new code to break the library as those tests we're added for a reason.

[mheap]

@3rd-Eden Any ideas why they're stalling on Travis? I can run them fine locally with the exact version that Travis uses

[mheap]

@3rd-Eden Interestingly, it builds fine for me - https://travis-ci.org/mheap/winston/jobs/10511961

The Node 0.8 tests stall instead, though

Edit: Interestingly, the last test run is the same both times: " An instance of the Daily Rotate File Transport after the logs have flushed the query() method using a bad from and until option
✓ should return no results"

[mheap]

@3rd-Eden Tests seem to be passing now on Travis. No change to the code, but as the error was that it was stalling and I couldn't reproduce locally, I'm not convinced there was ever an issue there

My test run is located at : https://travis-ci.org/mheap/winston
If you can trigger the build again at https://travis-ci.org/flatiron/winston/builds/14151171, we can see if it completes and if it does, get it merged in

[julien51]

Any progress on this?

[mheap]

@julien51 I can't reproduce the travis stalling issue consistantly. It looks like the tests just hang (I think it's to do with an early return if some setup fails, that doesn't call the callback) but I haven't had time to really dig into it

[julien51]

Dammit. Is stalling just for 0.6 ? How far back does winston has to support?

[julien51]

Hum. At this point, even master fails to build for me :(

[mheap]

@indexzero It's been a while, but I just triggered another build of this and the tests all passed. I think it was to do with some test setup previously (the failing tests were nothing to do with my changes).

As it's passing now, any chance of getting it merged?

[mheap]

cc @3rd-Eden on this PR, now with tests passing

[indexzero]

This has been resolved in the 2.x branch. We will be releasing winston@2.0.0 shortly with this fix.

[indexzero]

And thanks again for your contribution!