Fix more 0.7 deprecations

[stevengj]

No description provided.

[stevengj]

@tkelman, the failure on Travis/Linux with Julia 0.6 is somewhat mysterious to me. (The tests pass for me on my local machine.) Is there any way to get more test output? e.g. is it failing when precompiling?

[tkelman]

signal 9? maybe oom killed? try a debug build and ssh'ing in. and/or adding sudo dmesg output on failure

[codecov-io]

Codecov Report

Merging #505 into master will increase coverage by 0.97%.
The diff coverage is 89.13%.

@@            Coverage Diff             @@
##           master     #505      +/-   ##
==========================================
+ Coverage   64.61%   65.58%   +0.97%     
==========================================
  Files          18       18              
  Lines        1478     1485       +7     
==========================================
+ Hits          955      974      +19     
+ Misses        523      511      -12

Impacted Files	Coverage Δ
src/startup.jl	0% <ø> (ø)	⬆️
src/callback.jl	89.28% <ø> (-1.04%)	⬇️
src/numpy.jl	58.92% <100%> (+0.24%)	⬆️
src/pytype.jl	80.88% <100%> (-0.23%)	⬇️
src/gc.jl	100% <100%> (+18.18%)	⬆️
src/PyCall.jl	73.81% <100%> (+0.09%)	⬆️
src/serialize.jl	100% <100%> (ø)	⬆️
src/exception.jl	64.81% <100%> (+0.66%)	⬆️
src/pyeval.jl	69.47% <100%> (ø)	⬆️
src/pyiterator.jl	85.71% <100%> (+7.14%)	⬆️
... and 9 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 53d870d...1ada506. Read the comment docs.

[stevengj]

Thanks for the suggestion @tkelman; doing a debug build means messing around with Travis authorization tokens, unfortunately, but I may have to resort to that if I can't figure out how to reproduce the problem offline.

[stevengj]

It looks like defining all the constructor methods explicitly was causing compilation resources to explode in 0.6... hopefully it is now fixed.