git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v2] abbrev: allow extending beyond 20 chars to disambiguate

From
Derrick Stolee <stolee@gmail.com>
Date
Aug 12, 2025, 13:28 UTC
Message-ID
<55cbc80e-2585-49dd-80e9-33f200a9fac1@gmail.com>
In-Reply-To
<xmqqzfc51xvk.fsf@gitster.g>
On 8/11/2025 3:06 PM, Junio C Hamano wrote:
Show 12 quoted lines
> When you have two or more objects with object names that share more
> than half the length of the hash algorithm in use (e.g. 10 bytes for
> SHA-1 that produces 20-byte/160-bit hash), find_unique_abbrev()
> fails to show disambiguation.
> 
> To see how many leading letters of a given full object name is
> sufficiently unambiguous, the algorithm starts from a initial
> length, guessed based on the estimated number of objects in the
> repository, and see if another object that shares the prefix, and
> keeps extending the abbreviation.  The loop stops at GIT_MAX_RAWSZ,
> which is counted as the number of bytes, since 5b20ace6 (sha1_name:
> unroll len loop in find_unique_abbrev_r(), 2017-10-08);
Wow. My first Git contribution.

Some style issues that you're opportunistically cleaning up are due to my newness, for sure.

Show 39 quoted lines
> before that
> change, it extended up to GIT_SHA1_HEXSZ, which was the correct
> limit because the loop is adding one output letter per iteration
> and back then SHA256 was not in the picture.
> 
> Pass the max length of the hash being in use in the current
> repository down the code path, and use it to compute the code to
> update the abbreviation length required to make it unique.
> 
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  object-name.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/object-name.c b/object-name.c
> index 11aa0e6afc..8f9af57c0a 100644
> --- a/object-name.c
> +++ b/object-name.c
> @@ -680,6 +680,7 @@ static unsigned msb(unsigned long val)
>  struct min_abbrev_data {
>  	unsigned int init_len;
>  	unsigned int cur_len;
> +	unsigned int max_len;
>  	char *hex;
>  	struct repository *repo;
>  	const struct object_id *oid;
> @@ -699,12 +700,12 @@ static inline char get_hex_char_from_oid(const struct object_id *oid,
>  static int extend_abbrev_len(const struct object_id *oid, void *cb_data)
>  {
>  	struct min_abbrev_data *mad = cb_data;
> -
>  	unsigned int i = mad->init_len;
> +
>  	while (mad->hex[i] && mad->hex[i] == get_hex_char_from_oid(oid, i))
>  		i++;
>  
> -	if (i < GIT_MAX_RAWSZ && i >= mad->cur_len)
> +	if (mad->cur_len <= i && i < mad->max_len)
>  		mad->cur_len = i + 1;

This logic is all about not extending the abbreviation length to beyond the length of the hex array, so your limits make sense. Moving the comparisons here helps with readability.

Show 6 quoted lines
>  	return 0;
> @@ -864,6 +865,7 @@ int repo_find_unique_abbrev_r(struct repository *r, char *hex,
>  	mad.repo = r;
>  	mad.init_len = len;
>  	mad.cur_len = len;
> +	mad.max_len = hexsz;
This new parameter required for allowing SHA256 is valuable.

I agree that we shouldn't need a test case to guarantee this in the future. Good to clean up unnecessary uses of the macro limits.

Thanks, -Stolee

Previous: brian m. carlsonNext: René Scharfe
Message 5 of 14 in “abbrev: allow extending beyond 20 chars to disambiguate”
  1. abbrev: allow extending beyond 20 chars to disambiguateJunio C Hamano, Aug 11, 2025
  2. Junio C HamanoAug 11, 2025
  3. abbrev: allow extending beyond 20 chars to disambiguateJunio C Hamano, Aug 11, 2025
  4. brian m. carlsonAug 11, 2025
  5. Derrick StoleeAug 12, 2025
  6. René ScharfeAug 12, 2025
  7. Junio C HamanoAug 12, 2025
  8. René ScharfeAug 12, 2025
  9. abbrev: allow extending beyond 32 chars to disambiguateJunio C Hamano, Aug 14, 2025
  10. brian m. carlsonAug 11, 2025
  11. Junio C HamanoAug 11, 2025
  12. Junio C HamanoAug 11, 2025
  13. Jon ForrestAug 12, 2025
  14. René ScharfeAug 12, 2025

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.