Inline command write methods

[robd]

I had another look through #38 and spotted a possibility for a bit of simplification in the ConsoleFormatter. I think this is an improvement, and all the tests pass.

One thing I wondered about is deferring the command = decorate(command) lines until after the return if debug?(command) lines, but the decorate method has some side effects in Context.decorate_command. I was wondering if we could make those side effects more obvious in the ConsoleFormatter.

The other thing is, I think these side effects only ever occur when decorate is called from the log_command_start method. Therefore, I was wondering if we should just have a new explicit method on context and call that from log_command_start. This would make context.decorate_command side effect free and I think it would make the code easier to reason about.

If you think this is worth investigating, I will work on it a bit and see what I can come up with.

[mattbrictson]

I haven't had time to fully review this PR yet, but your comments about the side effects of decorate before debug? make me think that it is related to bug #43. Would you be willing to investigate?

[robd]

Hey - yes I think you are right. I believe that #43 is definitely related and may well be fixed by moving the calls to decorate after the debug?(command). I can look into this later tonight. I think it should be possible to expose the behaviour via a test. I think what we are missing for #43 is a test case which runs info and debug in an interleaved way.

[robd]

OK, I've had a look at this and you are right - #43 is due to calling decorate before debug. I have written a failing test for interleaved debug / info scenario. I think the refactor in this PR, or something very similar is a pre requisite to move the decorate after the debug call. Therefore, I will fix #43 on this branch. Hope that's OK.

[mattbrictson]

Sounds good! I'll wait for your patch and then merge this PR.

[robd]

Closed in favour of #45