[PATCH 00/15] Hierarchical reference cache (once again)

From: Michael Haggerty <mhagger@alum.mit.edu>

Here is another installment of the "hierarchical refs" saga. This patch series (obviously) builds on the changes in master that removed extra_refs. This group of patches does some cleanup then makes the transition to storing reference caches hierarchically. It also includes the first benefit of the hierarchical storage scheme, namely that searching for individual references becomes more efficient (though as yet all references are read into the cache if any references are needed, so the benefit is limited).

I will have limited Internet connectivity for the next week, so I apologize in advance if I don't react quickly to feedback. But I wanted to throw at least part of this reworked "ref-api" patch series back into the ring while the new release cycle is still fresh.

Michael Haggerty (15):
  refs.c: reorder definitions more logically
  refs: manage current_ref within do_one_ref()
  do_for_each_ref_in_array(): new function
  do_for_each_ref_in_arrays(): new function
  repack_without_ref(): reimplement using do_for_each_ref_in_array()
  names_conflict(): new function, extracted from is_refname_available()
  names_conflict(): simplify implementation
  is_refname_available(): reimplement using do_for_each_ref_in_array()
  free_ref_entry(): new function
  check_refname_component(): return 0 for zero-length components
  struct ref_entry: nest the value part in a union
  refs.c: rename ref_array -> ref_dir
  sort_ref_dir(): simplify logic
  refs: store references hierarchically
  do_for_each_ref(): only iterate over the subtree that was requested

 refs.c | 1036 +++++++++++++++++++++++++++++++++++++++++-----------------------
 refs.h |    7 +-
 2 files changed, 670 insertions(+), 373 deletions(-)