Release 2016.10 First, this release reverts a previous change to use a mtime of `1` back to `0`. This will affect flatpak and users who have set up build systems to set that mtime, but it restores compatibility with systems which were using the previous zero value. For more information, see https://github.com/ostreedev/ostree/pull/495 Another notable bit is we fixed a race condition in pull when the upstream ref hadn't changed. OSTree also learned about split metadata/content fetches and `mirrorlist` URIs. Generally, the idea with this is that you set up a centralized set of metadata servers that are secured via pinned TLS etc., and content fetches can occur via more local mirrors. Besides that, there is continued work on initramfs-less systems, and some fixes for resource use during delta generation. Colin Walters (8): repo: Revert default timestamp from 1 back to 0 delta: Add missing `goto out` for failure to mmap() repo: Only use mmap() for metadata > 16k delta: Unreference files we've processed fetcher: Fix another finalization deadlock sysroot: Avoid double cleanup, and ensure no cleanup if specified core: Make OSTREE_TIMESTAMP public API Release 2016.10 Dan Nicholson (3): build: Set --enable-man during distcheck build: Distribute man page XML source build: Actually distribute man page XML source Giuseppe Scrivano (1): gpg: do not segfault when the algorithm name is not known Jonathan Lebon (7): pull code: support contenturl setting trivial-httpd: prepend timestamp in log file pull: drop fetching_sync_uri pull: add mirrorlist support libtest: add has_gpgme() helper function tests: add tests for contenturl and mirrorlist pull code: clean up mirrorlist hack William Manley (4): switchroot: Fix build on Ubuntu switchroot: Fix test-switchroot now autotools can build static ostree-prepare-root: Error if realpath fails ostree-prepare-root: Fix running with musl Git-EVTag-v0-SHA512: affd25071654468e3fcc1d321afb6ade531550dc4f31fe3cd425adeeee9ecd4f56108a12aadd4c5404256409cf00c7821ff10d33afaa50dc38604c2323dc7901