Closes #125 and #208.

- sort the list instead of just picking the best one
- prefer unknown RTTs to probe them
- verbose output of the choice

Fixes https://gitlab.labs.nic.cz/knot/resolver/issues/125
Fixes https://gitlab.labs.nic.cz/knot/resolver/issues/208

They're just useless when we have kr_inaddr*.

There would be no real resource leak, as the memory is on a mempool,
so the lifetime is limited by kr_request finishing.  Also, it's only
on an unlikely error branch, but still...

Some people might get surprised, as it's the default location for quite
some data nowadays ($XDG_DATA_HOME).

I just found it by accident when studying RFCs again.
Hmm, I'd like an operator for logical implication; I'm spoiled by Maths.

daemon: check existence of config file

See merge request !317

Without it we were failing to download fstrm tarball,
and the shown errors were very confusing.
Tested building an image locally.

... even if rundir isn't specified.  No other changes in semantics.
Before this a typo in config path would pass silently.

Grigorii Demidov authored 7 years ago and Vladimír Čunát committed 7 years ago

unecessary queries in some circumstances; some minor bugfixes

Grigorii Demidov authored 7 years ago and Vladimír Čunát committed 7 years ago

it can contain records which have not been validated by validator

Grigorii Demidov authored 7 years ago and Vladimír Čunát committed 7 years ago

Don't check proof of nonexistance in NODATA or NAME ERROR answers
which contains a CNAME.

In particular, don't mention things that probably don't work (well):
Windows, amalgamation, very old OpenSUSE package.

I also verified there's no other usage of the `moduledir` symbol from
lua.  Bug introduced in 2f81b111 (within !298).

Leo Vandewoestijne authored 7 years ago and Vladimír Čunát committed 7 years ago

(the port will soon be 1.3.0 also)
Close https://github.com/CZ-NIC/knot-resolver/pull/43.

Leo Vandewoestijne authored 7 years ago and Vladimír Čunát committed 7 years ago

This tiny addition was needed to be able to install on FreeBSD.
Close https://github.com/CZ-NIC/knot-resolver/pull/44

lua kres-gen: add a guide to re-generating

See merge request !310

http://www.root-servers.org/news/b-root-begins-anycast-in-may.txt

Update the final date for 1.3.0 release

See merge request !308

iterate: be more precise when detecting CNAME loops

See merge request !307

Iterating over sibling sub-queries isn't precise enough,
and in particular in forwarding mode it reported non-existing loops.

Example ATM: www.dpo.cz - there forwarding spawns multiple sub-queries,
and one of those is a CNAME to another of them.  Due to them being siblings,
the old code misdetected that as a CNAME loop.  Now that we have these
cname_parent pointers, we can do a precise detection.