distclean and clean are different. "clean" is part of build and conditional
casing for "clean" therefore makes sense.

It will mean more to the user than the internal name

The latter caused build breakage on some systems.

libntp/wscript:40:1: E101 indentation contains mixed spaces and tabs
libntp/wscript:40:1: W191 indentation contains tabs
libntp/wscript:41:1: E101 indentation contains mixed spaces and tabs
pylib/util.py:473:1: E302 expected 2 blank lines, found 1
pylib/util.py:489:39: W291 trailing whitespace
pylib/util.py:1059:5: E303 too many blank lines (2)
tests/pylib/test_agentx.py:94:1: W293 blank line contains whitespace
tests/pylib/test_agentx.py:1851:1: W293 blank line contains whitespace
wafhelpers/pythonize-header:30:32: E261 at least two spaces before inline comment
wscript:236:5: E265 block comment should start with '# '

This change propagates compiler dialect switches to both the host and main
parts of the build.  Doing this with -std=c99 exposed a couple of minor
port bugs, which are also fixed.

This should fix at least one build issue under Solaris.

NOTE: While this change is not *known* to create any port problems with
unusual platforms, it's the kind of thing that very well could. Consider
bisecting here in case of port breakage.

One probleem in this issue - distclean not deleting ntpd/version.h or
ntpclients/ntp - can't be fixed.  clean does this through a hook in
wscript, but distclean has a separate, weird execution path that there
is no apparent way to set hooks in.

struct msghdr has all of the members specified by POSIX, if you set the defines
correctly.

Fixes Gitlab issue #342

related to issue #320 and mailing list comment https://lists.ntpsec.org/pipermail/devel/2017-May/004519.html
this patch makes the Trimble refclock depend on libparse

“This parrot is no more. It has ceased to be. It's expired and gone
to meet its maker. This is a late parrot. It's a stiff. Bereft of life,
it rests in peace. If you hadn't nailed it to the perch, it would be
pushing up the daisies. It's rung down the curtain and joined the choir
invisible. This is an ex-parrot.” -- John Cleese

It was always defined as __attribute__((noreturn)).  Code uses
that directly in other places, so it is just obfuscation.

FreeBSD has to be different...

Might as well hook them together as they both need the symbol table.
We'll if it needs changing when people use it.

Change anem from BACKTRACE_GCC to BACKTRACE_UNWIND since this is available
also on OS X with llvm.

Instead of manually set HAVE_LIBCTRACE use automatic HAVE_BACKTRACE_SYMBOLS_FD

Introduced in "Setup to help testing leap seconds"

Needed on openSUSE, at least, where the libseccomp headers are not installed in
/usr/include

Fixes GitLab issue #308

We need to avoid removing the cache of DEFINES when we write the config.h

Also, consolidate on one symbol for whether seccomp is enabled in config.h
(HAVE_SECCOMP_H)

Per issue #261

This shows serious problems on RasPi that need fixing, but put it
back in the closet for now.

Instead of sweeping inconveniennt truths under the rug, put them
in the closet.  On my list to get fixed.

-Winline now only with --enable-debug-warnings

The -Winline option produce two warnings that are (probably correctly)
indicating that inlining normalize_tspec() and dtolfp() is inefficient.
I judge that it's not worth the hassle of reorganizing the code to fix
these, so we'll just tell the compiler to stifle.

so that we can popen with a list later