It creates new callback functions for every request which
uses "callback chaining" but these should be rare.

It seems there is no reason to keep this function private in policy
module.

Attempt to avoid duplicating ten lines in debug_logfinish_cb lead me
to splitting kr_log_qverbose_impl into two functions kr_log_q and kr_log_req.
This is another minor change to API exposed to modules.

Formerly both logs used slightly different formats and duplicated code.
From now on verbose log and request tracing are generated using the same
code.

This required a small change to request trace_log_f definition so it
might affect external modules.

Normally the types are equal, but that's not true e.g.
on 32-bit systems and macOS.

In FreeBSD it's disabled by default, causing this error.
I tried disabling it on Linux via sysctl, but that didn't seem
to cause any errors.

Petr Špaček authored 5 years ago and Vladimír Čunát committed 5 years ago

These files did not have GNU GPL v3 boilderplate in them so
I've added machine readable tag with appropriate license.

Petr Špaček authored 5 years ago and Vladimír Čunát committed 5 years ago

Long GNU GPLv3 boilderplate was automatically replaced
with machine readable tag.

Vladimír Čunát authored 5 years ago and Petr Špaček committed 5 years ago

It just reports a failure to show the key-pin hash in the log.

GnuTLS 3.3 contained these as public symbols.  This doesn't seem
important, but why not use more collision-resistant names anyway?
Details/discussion: https://github.com/CZ-NIC/knot/pull/14
Basically noone should notice, so I didn't include any NEWS entry.

This also fixes a build error with GCC 10 on fedora.

Journald was missing some error messages when kresd failed to start.

Well, the part about giving semantics to nil actually
decreases strictness.

The preconfig is used to set distro-specific values to avoid messing
with user config, in partciular:

- binding to control sockets under systemd
- setting default cache location

- reduce visibility in docs
- add deprecation warnings (docs and stderr)

- consistent order of the options on the three places in parse_args()
- drop remains of -m -k -K
  It didn't seem to cause any problems except confusion in
  $ kresd -k
  kresd: option requires an argument -- 'k'

Unlink the unix address in case the file exists, e.g. from previous
unclean run.
Also attempt to create the directory for unix socket in case it doesn't
exist.