Re: [PATCH v2 00/10] object_id part 2

[Michael Haggerty <mhagger@alum.mit.edu>]

On 06/14/2015 12:16 AM, brian m. carlson wrote:
> This is another series of conversions to struct object_id.
> 
> This series converts more of the refs code and struct object to use
> struct object_id.  It introduces an additional helper function,
> has_object_file, which is the equivalent of has_sha1_file.  The name was
> chosen to be slightly more logical than has_oid_file, although it can be
> changed if desired.
> 
> The next-to-last patch in this series is the conversion of struct object
> to use struct object_id.  This is a necessarily large patch because of
> the large number of places this code is used.  That patch will most
> likely be too large to make it through to the list, so this series is
> available in the object-id-part2 branch from:

FWIW I've skimmed patches 01-08 inclusive and nothing jumped out at me.

Brian, what was your experience when writing these patches? Did they
tend to work as soon as they compiled without errors (i.e., not super
risky) or did you often have test suite failures that you had to go back
and fix (i.e., risky)? If the latter, what kinds of code patterns tended
to be problematic? Your answers might help reviewers decide how much
diligence is needed when reviewing these patches and what kind of
changes to inspect extra carefully. Because doing a thorough review of
all of the patches would be quite a bit of work.

Michael

-- 
Michael Haggerty
mhagger@alum.mit.edu