so that we can test with an empty diffs table.

so that we can test them with SQLite3 as much as possible.

almost empty with bootstrap tests.

a specific order of the underlying datasrc SQLite3 query result, but it
doesn't appear to be the case for all systems.

(I made a couple of piggy-back small fix/cleanup: avoid unexpected exception
in an assert failure message for an empty (test) RRset; suppress log output
(by default))

Resolved confusion over "operation" values for adding and deleting
records.  Added ORDER BY clause to SQL statement and removed
redundant variable declaration.

Conflicts:
	src/lib/datasrc/database.cc
	src/lib/datasrc/tests/database_unittest.cc

there's even any compiler that allows it, but at least it shouldn't
be portable if not illegal.)

Conflicts:
	ChangeLog

Conflicts:
	src/bin/xfrin/tests/xfrin_test.py
	src/bin/xfrin/xfrin.py.in

Inadvertantly removed a check for NULL.

In particular, use checkRR to compare RRs returned from
iterators.

to an existing entry)

A different exception adds nothing to the error and has the
complication of requiring an additional query.

As the application can't call the constructor directly, it will be
passed from here.

Defaults to false as a backward compatibility (might be temporary)

hid the definition of uint32_t.  also, replaced an unused boost header file
(foreach) to a more relevant one (shared_ptr).
may be beyond a super-trivial level, but as an urgent care fix I'll push it
now.  it was confirmed on our solaris box.

complains about it.