GitLab

for python logging message scripts.

briefly discussed on jabber, didn't see any objection.

with fixing conflicts in src/lib/datasrc/tests/database_unittest.cc

with fixing conflicts in database_unittest.cc.

RRSIG support for database DataSources

with fixing conflicts.  The conflicts were only in database_unittest.cc,
and the necessary fix is straighforward updates to make newly added tests
work as TYPED_TESTs.

with the mock accessor and sqlite3 accessor.

operations with multiple adds and deletes.

of the in-memory data source client.

of class member variables (which are now removed).  also added a test to
detect a bug in the previous code where the sigtype field for addRRset
was incorrectly carried over after adding an RRSIG.

Also updated doc and tests accordingly.

* Avoid infinite recursion - the operator | kind of happened to call
  itself by accident.
* Re-added lost | dnssec_opt_.

the MockAccessor class definition.

catch exception in rollback explicitly.  added a test case for that.

implementation completely within .cc as even the class name doesn't have to
be visible.

startUpdateZone() to be more consistent with getIterator()).

result value of system() was unused. And if we're catching it anyway, might as well error on it if the command failed.
reviewed on jabber

While backends are allowed to pass DNSSEC data when not asked to, they
don't have to, so we need to ask for it. Test backend therefore doesn't
return the DNSSEC data when not asked to, to check it really asks.

Only positive matches for now.

This looks like someone already prepared for it, it only needs to
passing the intent to few functions and everything handles itself well.

Also updated tests, since the utility function gets confused by the
RRSIGs. Not fixing the function for now, it looks complicated, because
the problem is in it's design - if it happens to be just this one case,
it will be easier.

by update tests.  the semantics should be preserved at the moment.
making a commit at this point mainly as a checkpoint.

for database_unittests (due to too many conflicts).

also use is_open() instead of != NULL (it's strange that the original code
did even compile, but in any case this one should be better in clarity and
possibly in portability).
okayed on jabber, should be minor enough, directly pushing.