Marek Vavruša authored 7 years ago and Grigorii Demidov committed 6 years ago

The AD indicates validation request (but not request for DNSSEC records).
If the response can't be validated, resolver flips the AD to 0.

distro: packaging updates

See merge request !567

daemon: allow per-request variables in Lua

See merge request !533

The handlers in Lua can now store per-request variables that are automatically
GC'd when the request is finished. This is useful for stateful modules,
such as DNS64 that uses internal option flags for state tracking.

The layers can now get a variable table like so:

```
local vars = kres.request_t(r):vars()
vars.hello = true
```

The variables are persisted between different layers for each request.

cache/entry_rr: fixed undefined behavior

See merge request !595

Pointer arithmetic with 'void *' is undefined, it only works as GNU extension.

synchronize distro/deb with official debian packaging

See merge request !593

Daniel Kahn Gillmor authored 6 years ago and Tomas Krizek committed 6 years ago

Note that this is not an exact synchronization, just the salient parts
where i believe upstream wants to sync up with debian.

I've left alone indications of upstream package maintainership, and
I've left debhelper back at version 9, for example.  and i've left the
differences between the shipped debian/kresd.conf (debian package
ships a minimal kresd.conf that doesn't load those four modules, in
particular).

documentation nitpicks

See merge request !592

I don't think it's good to write that we "provide a library",
as it currently doesn't seem suitable for usage outside kresd.

It seems like implementation and docs got de-synchronized,
probably at some point very long ago (years).

doc: systemd -- clarify how to manually start all services.

See merge request !591

Daniel Kahn Gillmor authored 6 years ago and Tomas Krizek committed 6 years ago

See https://github.com/systemd/systemd/issues/9080 for
details/discussion.

ci: add pytest-xdist dependency for Deckard

See merge request !590

handle htobe32 et al. on glibc systems with a non-Linux kernel

See merge request !588

This fix copies over an updated #if clause from libknot's
src/contrib/endian.h.

This should resolve:

    https://gitlab.labs.nic.cz/knot/knot-resolver/issues/348

See also discussion about this same problem in libknot from a couple
years ago:

    https://bugs.debian.org/840460

I note that contrib/wire.h in knot-resolver is out of sync with
src/contrib/wire.h (and src/contrib/endian.h) from libknot.  I don't
know whether there's any upstream preference for keeping these in sync
in some more reliable way than manual comparisons.  For now i'm just
providing a narrow fix for the specific problem.

iterate: avoid turning off qname minimization in a case

Closes #339

See merge request !584

Thanks to @ spakka for discovering this and authoring an earlier version
of this commit.

In case of referrals the authoritative server might add also another NS
record(s), and this might lead to downgrading the corresponding zones.
Regressed probably in f0da0a35 !505.

We don't need the end, but I prefer to get rid of the exception,
as we now require libknot > 2.6 anyway and it only implies adding
a few more binding lines.

Our Makefile required >= 2.6.4 for some time already.
It's pretty straightforward, except for zonefile.lua.in -> zonefile.lua

ci: respdiff - use relative path to restart script

See merge request !580