You can't use exit() to exit in the same way as if you had
died with a signal; we tried to do some exit-code munging
but we did it one way and our tester script expected another
way.  we could resolve that, or use the approach here: try
to die with the same signal by sending it to ourselves.

The debug builds are taking a very long time to compile, but
when I reran one it completed very quickly.  Let's see if
this works+helps.

Add docs and tips for new contributors

try reverting these

This reverts commit 071ca0d8, reversing
changes made to aba975d1.

This reverts commit c78fdcb9.

Fixes

Instead, save the file data and parse that instead.
I think that should help with cases where the cached
file was getting trampled.

More travis-ci investigations

A fix to the ast interpreter gc visitor

more small perf work

Cache floats based on their bits, not their value

Fix some issues found during a search for another bug

now that the strings get interned anyway.  if we want to continue
down that road of interning BoxedStrings, we could probably do
away with InternedString's entirely.

Allocating things in-line, using malloc-friendly data structures, etc.

Fix a small test output difference on Ubuntu 15.04

- call PyType_Check instead of isSubclass
- unlikely()
- object_cls gets checked all the time but only has attributes that
  start with '_'

The help text has changed...

We were using a C 'double' as the hashmap key, which meant
that 0.0 and -0.0 would end up caching to the same thing.
Instead, extract the bits of the double, and use that as the
cache key instead.

Should fix #724

Add some more debugging for travis-ci builds

We had lost it since I guess `VAR=a cmd1 && cmd2` only applies
the VAR variable to the first command and not the second.

ie for the "repeatedly failing to parse file" error.

rewriter: release allocated scratch space

make attributes interned BoxedStrings

CMake already did this, and I think this is discrepancy is why
if you did `make quick_check`, then `make check_dbg` would fail.

Due to their non-standard bootstrapping, they ended up getting
initialized back to NORMAL hidden classes.  It's silly, but force
them back to SINGLETON.

One of the downsides of the BoxedString-as-attributes change
is that hidden classes become more complicated to gc-scan;
try to claw some of that back.

reenable PyList macros

speed calling of (some) capi code

The real benefit is that we intern the strings that
end up getting used as attribute names, so we can compare
them using pointer comparisons.  It should also reduce
the size overhead of hidden classes, since we no longer
have to copy the string data into the hidden class.