When I recently submitted a topic[1] to convert various things to
designated initializers Junio (presumably) ejected the refs changes
due to conflicts with ps/fetch-mirror-optim. The rest of the series
was down already in 20d34c07ea4 (Merge branch
'ab/c99-designated-initializers', 2022-03-06)
Now that that series has landed 1-3/5 are those previous patches
re-submitted (with a minor change for the new "read_symbolic_ref").
The 4/5 is then a deletion of packed-backend.c stub functions that I
think are not worth carrying anymore, and which has the fringe benefit
of squashing a noisy warning under SunCC.
5/5 is then a minor fix-up for ps/fetch-mirror-optim: It didn't define
a debug wrapper for refs_read_symbolic_ref(). Now that we do we don't
need an implementation that'll ever fall back on
refs_read_raw_ref(). It's now a stand-alone API function without a
fallback (which in effect it was before, sans debug backend).
Aside: It seems that the GIT_TRACE_REFS facility has been broken since
it was added, i.e. running the test site with e.g.:
GIT_TRACE_REFS=/tmp/log.txt make test
Will fail, and not due to a test being confused about the logs. I
think something in how the iterators are wrapped is confusing it, but
that's a long-standing issue in any case...
1. https://lore.kernel.org/git/cover-00.12-00000000000-20220224T092805Z-avarab@xxxxxxxxx
Ævar Arnfjörð Bjarmason (5):
refs: use designated initializers for "struct ref_storage_be"
refs: use designated initializers for "struct ref_iterator_vtable"
misc *.c: use designated initializers for struct assignments
packed-backend: remove stub BUG(...) functions
refs debug: add a wrapper for "read_symbolic_ref"
attr.c | 2 +-
notes-merge.c | 1 +
object-file.c | 9 +--
refs.c | 13 +----
refs/debug.c | 82 ++++++++++++++++++---------
refs/files-backend.c | 64 ++++++++++-----------
refs/iterator.c | 18 +++---
refs/packed-backend.c | 128 ++++++++++--------------------------------
refs/ref-cache.c | 6 +-
9 files changed, 135 insertions(+), 188 deletions(-)
--
2.35.1.1384.g7d2906948a1
