Using NULL for undefined values is confusing since maybe
something actually returned that.  So try using -1.

Somewhat ad-hoc at the moment; only added for a few types
of operations.

Add a couple tests to the travis-ci configuration

correct output filenames for the new .so's

- Build and run in Debug mode (but not the extra tests)
- make sure that building + running doesn't dirty the repository.

fix null check

add support for the _curses, bz2, and grp modules (mercurial needs them)

don't do isNonheapRoot first, since .so's loading at arbitrary accesses can put max/min on either side of our gc heap.

add long.__float__ attribute

Revert "Merge pull request #685 from undingen/interp_gc2"

This reverts commit 78493220, reversing
changes made to 194d51eb.

This was a merge conflict between #656 and #693

Fix lea

baseline jit: directly emit patchpoints instead of using runtime ICs

Don't guard types sometimes

I encountered this issue sometimes with the protobuf integration test.

this also adds support for calling 8 arg functions (=2 stack args)

rewrite some tp_call cases, and reduce calls to getTopPythonFrame

use inline functions instead of casting function pointers

Fewer mallocs

interpreter: Don't GC allocate ASTInterpreter instances.

Print out the last N lines of a failing test instead of first N

after adding these modules, virtualenv_test now tries to access file.mode, which we hadn't exposed before.

And also bump N from 200->500

Fix analysis_unittest after tiering changes

It was failing an assert that doesn't get checked on travis-ci,
which only tests release mode :/

GC allocating the huge number of large ASTInterpreter instances is slow,
instead add a small wrapper object which gets GC allocated and notifies the ASTInterpreter.

tiering refactoring