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

Re: [PATCH v3 4/5] sha1_name: Parse less while finding common prefix

From
Junio C Hamano <gitster@pobox.com>
Date
Oct 4, 2017, 06:14 UTC
Message-ID
<xmqqlgkrcucj.fsf@gitster.mtv.corp.google.com>
In-Reply-To
<20171002145651.204984-5-dstolee@microsoft.com>
Derrick Stolee <dstolee@microsoft.com> writes:
Show 19 quoted lines
> diff --git a/sha1_name.c b/sha1_name.c
> index f2a1ebe49..5081aeb71 100644
> --- a/sha1_name.c
> +++ b/sha1_name.c
> @@ -480,13 +480,23 @@ struct min_abbrev_data {
>  	char *hex;
>  };
>  
> +static inline char get_hex_char_from_oid(const struct object_id *oid,
> +					 int pos)
> +{
> +	static const char hex[] = "0123456789abcdef";
> +
> +	if ((pos & 1) == 0)
> +		return hex[oid->hash[pos >> 1] >> 4];
> +	else
> +		return hex[oid->hash[pos >> 1] & 0xf];
> +}
> +
Show 9 quoted lines
>  static int extend_abbrev_len(const struct object_id *oid, void *cb_data)
>  {
>  	struct min_abbrev_data *mad = cb_data;
>  
> -	char *hex = oid_to_hex(oid);
>  	unsigned int i = mad->init_len;
> -	while (mad->hex[i] && mad->hex[i] == hex[i])
> +	while (mad->hex[i] && mad->hex[i] == get_hex_char_from_oid(oid, i))
>  		i++;

Assuming that [PATCH 3/5] makes sense, it is an obvious optimization to avoid writing the whole 20-byte out before comparing, and instead to grab hex digits as they become needed.

I assume that the "Base Time" in the log message was with whatever version of Git before [PATCH 3/5] and this one were applied (i.e. not comparing to "vanilla Git plus 3/5")?

Thanks.
Previous: Derrick StoleeNext: Derrick Stolee
Message 28 of 47 in “Improve abbreviation disambiguation”
  1. 0/5 Improve abbreviation disambiguationDerrick Stolee, Sep 25, 2017
  2. 2/5 p0008-abbrev.sh: Test find_unique_abbrev() perfDerrick Stolee, Sep 25, 2017
  3. Junio C HamanoSep 26, 2017
  4. Jeff KingOct 5, 2017
  5. Jeff KingOct 5, 2017
  6. 5/5 sha1_name: Minimize OID comparisons during disambiguationDerrick Stolee, Sep 25, 2017
  7. 1/5 test-list-objects: List a subset of object idsDerrick Stolee, Sep 25, 2017
  8. Junio C HamanoSep 26, 2017
  9. Jeff KingOct 5, 2017
  10. Junio C HamanoOct 5, 2017
  11. Jeff KingOct 5, 2017
  12. Junio C HamanoOct 5, 2017
  13. Derrick StoleeOct 5, 2017
  14. Jeff KingOct 6, 2017
  15. Derrick StoleeOct 7, 2017
  16. Jeff KingOct 7, 2017
  17. Junio C HamanoOct 8, 2017
  18. 4/5 sha1_name: Parse less while finding common prefixDerrick Stolee, Sep 25, 2017
  19. Stefan BellerSep 25, 2017
  20. Derrick StoleeOct 2, 2017
  21. 3/5 sha1_name: Unroll len loop in find_unique_abbrev_rDerrick Stolee, Sep 25, 2017
  22. 1/5 test-list-objects: List a subset of object idsDerrick Stolee, Oct 2, 2017
  23. Junio C HamanoOct 3, 2017
  24. 2/5 p0008-abbrev.sh: Test find_unique_abbrev() perfDerrick Stolee, Oct 2, 2017
  25. 0/5 Improve abbreviation disambituationDerrick Stolee, Oct 2, 2017
  26. Jeff KingOct 5, 2017
  27. 4/5 sha1_name: Parse less while finding common prefixDerrick Stolee, Oct 2, 2017
  28. Junio C HamanoOct 4, 2017
  29. 3/5 sha1_name: Unroll len loop in find_unique_abbrev_rDerrick Stolee, Oct 2, 2017
  30. Junio C HamanoOct 3, 2017
  31. Derrick StoleeOct 3, 2017
  32. Junio C HamanoOct 4, 2017
  33. Derrick StoleeOct 4, 2017
  34. Junio C HamanoOct 4, 2017
  35. Derrick StoleeOct 4, 2017
  36. Junio C HamanoOct 5, 2017
  37. Jeff KingOct 5, 2017
  38. Junio C HamanoOct 5, 2017
  39. 5/5 sha1_name: Minimize OID comparisons during disambiguationDerrick Stolee, Oct 2, 2017
  40. Stefan BellerOct 3, 2017
  41. Derrick StoleeOct 3, 2017
  42. Jeff KingOct 5, 2017
  43. cleanup: fix possible overflow errors in binary searchDerrick Stolee, Oct 6, 2017
  44. Jeff KingOct 6, 2017
  45. Derrick StoleeOct 6, 2017
  46. cleanup: fix possible overflow errors in binary searchDerrick Stolee, Oct 8, 2017
  47. Jeff KingOct 9, 2017

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.