[0/10] removing nth_packed_object_sha1()

Message

Jeff King Feb. 24, 2020, 4:26 a.m. UTC

This series gets rid of nth_packed_object_sha1(), converting its
remaining callers to nth_packed_object_oid(). The latter provides better
type-safety in general, and it also allowed me to clean up some tricky
bits of the pack-verification code (patch 7 below).

This was inspired by brian's work in:

 https://lore.kernel.org/git/[e-mail]/

but I decide not to base it on that series. It's semantically
independent, and with the exception of one line, textually independent
(patch 4 here touches a different parameter in the same call as the
patch linked above). So I think it's easier to handle them independently
and just resolve the trivial conflict.

  [01/10]: nth_packed_object_oid(): use customary integer return
  [02/10]: pack-objects: read delta base oid into object_id struct
  [03/10]: pack-objects: convert oe_set_delta_ext() to use object_id
  [04/10]: pack-objects: use object_id struct in pack-reuse code
  [05/10]: pack-bitmap: use object_id when loading on-disk bitmaps
  [06/10]: pack-check: convert "internal error" die to a BUG()
  [07/10]: pack-check: push oid lookup into loop
  [08/10]: packed_object_info(): use object_id for returning delta base
  [09/10]: packed_object_info(): use object_id internally for delta base
  [10/10]: packfile: drop nth_packed_object_sha1()

 builtin/cat-file.c     |  2 +-
 builtin/pack-objects.c | 48 +++++++++++++++--------------
 midx.c                 |  2 +-
 object-store.h         |  2 +-
 pack-bitmap.c          | 16 +++++-----
 pack-check.c           | 22 ++++++--------
 pack-objects.c         |  4 +--
 pack-objects.h         |  2 +-
 packfile.c             | 69 ++++++++++++++++++------------------------
 packfile.h             | 15 +++------
 ref-filter.c           |  4 +--
 sha1-file.c            |  8 ++---
 sha1-name.c            | 13 ++++----
 13 files changed, 94 insertions(+), 113 deletions(-)

-Peff

Comments

brian m. carlson Feb. 25, 2020, 12:33 a.m. UTC | #1

On 2020-02-24 at 04:26:25, Jeff King wrote:
> This series gets rid of nth_packed_object_sha1(), converting its
> remaining callers to nth_packed_object_oid(). The latter provides better
> type-safety in general, and it also allowed me to clean up some tricky
> bits of the pack-verification code (patch 7 below).
> 
> This was inspired by brian's work in:
> 
>  https://lore.kernel.org/git/[e-mail]/
> 
> but I decide not to base it on that series. It's semantically
> independent, and with the exception of one line, textually independent
> (patch 4 here touches a different parameter in the same call as the
> patch linked above). So I think it's easier to handle them independently
> and just resolve the trivial conflict.

I think this is the right move.  The conflicts will be minor, as you
said.

The series looked good to me, and as mentioned, I appreciate you using
oidread everywhere.  It's nice to get rid of more code with "sha1" in
it.