GitLab

we should either/both consider revisiting the interface so we avoid the
issue as a result or require higher version of boost.  but for now this
should be okay.

the change introduced in this branch breaks the assumption of the
callback function used in the test code.

It caused the pycompile script to fail

There was a problem with placing the unix domain socket, it wanted to
put it into the source directory.

(they should have succeeded in the original version of implementation)

now the test passed.

separate RRsets.  As noted in the previous comment the previous behavior
was okay in practice within the expected restriction of the current
implementation, but as we are probably going to accept more generic forms
of input it would be prudent to fix it at this timing.

covered type and attaches the RRSIGs to the covered RRset.  It also performs
some additional validations on RRSIGs.  As a result an empty RRset is
now prohibited (which is correct in this context anyway), so some existing
tests were updated, too.

this is not absolutely necessary for the purpose of this branch, but
it will help make it less ugly.

And add few more to note where to catch them.

commiting this at my discretion.

There were two instances left the last time.

So the close happens every time.

It got outdated and inaccurate.

the tentative attempt of returning NSEC3 RRsets in negative cases were
canceled, and updated the doxygen doc accordingly.

themselves and special attributes.  for now, provided a quick-hack wrapper
for auth::Query so we don't have to all data source implementations at once.