GitLab

It was obviously forgotten (and cppcheck complained).

The option would be used often. While the produced code would probably
be the same, because the compiler can see it's equivalent, this is more
convenient to use.

Just to make it easier for some editors to parse.

When we support $INCLUDE, it should report reasonable place of error,
not :0.

In this case, the user should be aware of the error as much as possible.
In case of MANY_ERRORS, we can't do this, exception is able to handle
just one error.

Instead of file name. And use it for some of the tests, to avoid
juggling with files.

It claimed it throws every time and returns NULL at the same time. It is
impossible and the first one is not true.

Reject invalid parameters and invalid attempts to load twice.

By catching all exceptions. It is questionable, if we want to do it that
way, but it seems to work at least for now.

Generate several broken input and see all of them fail. Also check they
continue to parse or not according to parameters.

Tested at least on the domain names. There is no known Rdata type
implementation handling relative names right now.

As we pass exactly one RR each time, RRset is overkill. We just pass
bunch of parameters to the callback.

It seemed the wrong pushSource was used due to some automatic
conversion. Also, it complained it couldn't find the file, because a
slash was missing (and nobody noticed before).

And fill it into the rrset.

This was from the template in trac and it is currently not needed.

The tentative interface of the MasterLoader was missing a callback to
add RRsets.

Also, include some small changes to the interface - pass RRClass as
reference and have DEFAULT options value for no special options.