This distribution includes a stripped down version of Template::Stash::XS
for the purpose of tracking down a bug when using DateTime objects.
See http://rt.cpan.org/Public/Bug/Display.html?id=48020

UPDATE: TJC's report seems to suggest that the bug has nothing to do with 
DateTime, but is caused by a string eval in a subroutine.  See the new
t/string_eval.t test.  https://rt.cpan.org/Ticket/Display.html?id=47929

To build and run the tests:

  $ perl Makefile.PL
  $ make 
  $ make test

The bug only appears to manifest itself on 64 bit systems.

If you don't have the bug on your system then you should see output like this:

  [abw@shiny ~/tt/TT-XS-Stash-DateTime] make test
  t/datetime......- dotop(date_time)
  - fetch item: date_time
  - fetching hash item
  - got value, triggering any tied magic
  - TT_RET_OK
  t/datetime......1/2 - dotop(date_time_sub)
  - fetch item: date_time_sub
  - fetching hash item
  - got value, triggering any tied magic
  - calling coderef
  - in call_coderef()
  - about to push args
  - pushed args
  - calling call_sv()
  # Creating DateTime object
  # Created DateTime object, returning
  - called call_sv()
  - called coderef, returning result
  - TT_RET_CODEREF
  t/datetime......ok   
  All tests successful.
  Files=1, Tests=2,  0 wallclock secs ( 0.01 usr  0.01 sys +  0.20 cusr  0.02 csys =  0.24 CPU)
  Result: PASS

Or you can run the t/datetime.t script by itself (make sure you run 'make' first)

  [abw@shiny ~/tt/TT-XS-Stash-DateTime] perl t/datetime.t 
  1..2
  - dotop(date_time)
  - fetch item: date_time
  - fetching hash item
  - got value, triggering any tied magic
  - TT_RET_OK
  ok 1 - The year is 2009 (DateTime object)
  - dotop(date_time_sub)
  - fetch item: date_time_sub
  - fetching hash item
  - got value, triggering any tied magic
  - calling coderef
  - in call_coderef()
  - about to push args
  - pushed args
  - calling call_sv()
  # Creating DateTime object
  # Created DateTime object, returning
  - called call_sv()
  - called coderef, returning result
  - TT_RET_CODEREF
  ok 2 - The year is 2009 (subroutine returning DateTime object)

If the bug is present on your system then it won't proceed past this point:

  # Created DateTime object, returning

The final 4 lines will be missing

  - called call_sv()
  - called coderef, returning result
  - TT_RET_CODEREF
  ok 2 - The year is 2009 (subroutine returning DateTime object)

UPDATE: the new t/string_eval.t test.  This fails on my OSX laptop even
though the t/datetime.t test passes.  I can only assume this is a Good 
Thing as it makes the bug more reproducible.

  [abw@shiny ~/tt/TT-XS-Stash-DateTime] perl t/string_eval.t 
  1..3
  - dotop(eval_fail)
  - fetch item: eval_fail
  - fetching hash item
  - got value, triggering any tied magic
  - calling coderef
  - in call_coderef()
  - about to push args
  - pushed args
  - calling call_sv()
  # About to use NonExistantClass
  # Caught expected failure to use Nonexistent::Class at t/string_eval.t line 27.
  # No tests run!

If you haven't got the bug then you'll see the last 4 lines.

  - called call_sv()
  - called coderef, returning result
  - TT_RET_CODEREF
  ok 1 - eval failed as expected

If you don't see those lines then congratulations, you've reproduced the 
bug.  Now, can you fix it?  Please?  :-)