hard drive sounded bad. TODO includes free and copy updated and a bunch
of refactoring / redesign of the solution

"glr(1)" and call compile for the appropriate backend. Needs clean up
and refactor, and tests, but it compiles and my hard-drive is
periodically making threatening noises...

better_backend_handling

Conflicts resolved:
	src/SConscript
	src/test_suite.h

Fix uninitialized symbol warning in cfgrammar.c

See merge request hammer/hammer!42

Fix feature macros

See merge request hammer/hammer!41

Restore -D_POSIX_C_SOURCE=200809L in build flags on Linux; we'll figure out report of Mac OS breakage later

Using _POSIX_C_SOURCE in the middle of a header is incorrect and may work unreliably because of multiple-include guards

add regression test for issue 87 - same as issue 91

Closes #87

See merge request hammer/hammer!37

Resolve "empty follow set assertion triggered with LALR"

Closes #92

See merge request hammer/hammer!35

add TT_DOUBLE and TT_FLOAT as user-accessible token types

See merge request hammer/hammer!32

Resolve "_POSIX_C_SOURCE breaks build on MacOS X"

Closes #93

See merge request hammer/hammer!38

Resolve assertion failure with rules that can never match

Closes #83

See merge request hammer/hammer!39

fixes issue #83

hammer/hammer#83

doh! fixes issue 92 again.

tentative commit. this should solve issue 92, but isn't working, yet.

the idea is to taint any memoized follow set that hit a recursive loop,
directly or indirectly. the recursive calls (h_follow_rec) are allowed to use
these intermediates, but propagate the taint. the top-level routine (h_follow)
places the original taint on its result and may remove it at the end. it will
not return memoized results that are tainted (but compute them afresh).

without it, our test for issue 92 is no good (will always pass).
we could probably officially put it back if we simplify "dead"
symbols (that have no productions) out of the grammar beforehand.

there is a case where fs can be empty: when reducing by lhs would lead to
certain parse failure, by means of h_nothing_p() for instance.

fixes issue 83.

cf.
#83

this makes the computation more wasteful but avoids issue 92. intermediate
results could end up in the memoization table and be treated as final by
later calls to h_follow(). this problem could appear or not depending on the
order of nonterminals (i.e. pointers) in a hashtable.