More Dialyzer

[jkingsbery]

1. Added types for (most of) the greeting record.
2. In dj_al's commit, there was a check the encoding, but running dialyzer shows that the encoding can only ever be an atom, so the other checks in the pattern match were superfluous. I made the simplest change to get dialyzer passing again, but now the encoding_ok check seems superfluous (again, because the encoding field has to be an atom for dialyzer to pass). Should make dialyzer be added to the travis build?

[jkingsbery]

Hmm... I'll take a look at this tomorrow to see why it's failing.

[jlouis]

Ok. I think we should consider adding dialyzer to the travis build if possible. Since we are dialyzer-clean, we better keep it so. @dj_al, did you try running the dialyzer on this?

[jkingsbery]

The test that @dj_al added has a value for encoding that doesn't match the dialyzer types. We either need to change the test to match the types or fix the types in emysql.hrl to match the allowed encodings.

[jlouis]

Hmm, what do you propose we do here? I am all for making these things align up.

[jkingsbery]

I copied the type info over from the function spec. Dialyzer and tests both pass now.

[jlouis]

Awesome, pulling!