Stefan Beller <sbeller@xxxxxxxxxx> writes: > grep "fatal: test-blob-1 is neither a commit nor blob" actual OK, that might be somewhat unsatisfying from end-user's point of view (logically "test-blob-1" is already a name based on the 'graph relations' that is satisfactory). [side note] should that grep become test_i18ngrep? If this machinery is to find an object that is *bad* (in the sense in one of your earlier messages), I suspect it may be quite easy to use the new mechanism added by the series to also locate a tree object and report "that tree appears in this commit at this path" and it would be equally useful as the feature to find a blob. Thanks.
