[ccan] CCAN News for December

[Rusty Russell]

Hi all!

   Happy new hacking year, everyone.  Things have slowed a bit with the
traditional end of decade festivities, but lots of minor improvements to
ccanlint.

   Highlight this month is tlist: typesafe variant of ccan/list.  It's not
perfect, but it should be a minimally disruptive change for those used to
Linux-style list.h.

   I'm giving a talk at linux.conf.au on CCAN, which I'm preparing now.
Expect most of my hacking this month to be related to that!

Infrastructure:
	- configurator: tweaks and some new config.h defines.
	- ccanlint:
		- We don't use -Werror any more, so we distinguish "failed
		  compile" from "compiled with warnings".
		- Report definite memory leaks in tests from valgrind.
		- Tests run in alphabetical order, so you can order them from
		  simple to complex.
		- ccanlint: section in _info replaces old "Fails" section.
		- Build depends privately if they're not already built.
		- Lots of minor fixes and cleanups.

New modules:
	- tlist: typesafe list, inspired by Vegard Nossum.

Changed modules:
	- daemonize: set stderr to /dev/null, misc infrastructure.
	- idtree: fixes imported from SAMBA upstream.
	- list: list_del_from and LIST_HEAD_INIT, misc cleanups.
	- nfs: now a first-class CCAN module.
	- rbtree: trbt_delete32() fix from Ronnie.
	- tdb: simplify tdb_summary for upstream submission.
	- tdb2: functionally complete, performance improved massively.
	- typesafe_cb: fix compilation warnings under gcc-4.5.

Cheers,
Rusty.