Add Travis-CI job

[hvr]

As currently dlist doesn't build with GHCs prior to GHC 7.6.1 (but claims to, according to the .cabal contents), it would be a good idea IMO to use Travis-CI to make sure it is detected early if .cabal and reality start to diverge :-)

[hvr]

btw, I had to comment out cabal test as it fails due to an out-of-mem error even on a 32GiB machine, e.g.:

Preprocessing test suite 'test' for dlist-0.5...
[1 of 1] Compiling Main             ( tests/Main.hs, dist/build/test/test-tmp/Main.o )
Linking dist/build/test/test ...
Running 1 test suites...
Test suite test: RUNNING...
8: empty                    : +++ OK, passed 100 tests.
4: singleton                : +++ OK, passed 100 tests.
1: head                     : +++ OK, passed 100 tests.
7: snoc                     : +++ OK, passed 100 tests.
3: model                    : +++ OK, passed 100 tests.
6: cons                     : +++ OK, passed 100 tests.
4: tail                     : +++ OK, passed 100 tests.
2: append                   : +++ OK, passed 100 tests.
4: map fusion               : +++ OK, passed 100 tests.
1: unfoldr                  : +++ OK, passed 100 tests.
2: read . show              : +++ OK, passed 100 tests.
4: show . read              : +++ OK, passed 100 tests.
6: map                      : +++ OK, passed 100 tests.
5: concat                   : +++ OK, passed 100 tests.
7: foldr                    : +++ OK, passed 100 tests.
test: out of memory (requested 2097152 bytes)
Test suite test: FAIL
Test suite logged to: dist/test/dlist-0.5-test.log
0 of 1 test suites (0 of 1 test cases) passed.

[spl]

@hvr Thanks so much! I didn't know you could run so many versions of GHC on Travis. And thanks for the cabal corrections. Forgot about them and hadn't gotten around to testing sdist yet.

[spl]

@hvr Also, I hadn't yet been able to run the tests to completion. The one time I tried, my machine froze. So, thanks for that info, too.

[spl]

@hvr I tried extending the Travis-CI configuration to test dlist with GHC 6.12.3, but I'm unsuccessful at it. Do you have any idea what the problem is here?

[hvr]

Probably, this is another case of cabal sdist needing some features that were not available with ghc-6.12's bundled Cabal library version. Personally, I've given up a bit on supporting GHC 6.12.x as it requires a few contortions (mostly, that you can't assume Cabal-1.10 & you can't use Haskell2010)