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

Re: [PATCH v3 6/6] diff: shortcut for diff'ing two binary SHA-1 objects

From
Junio C Hamano <gitster@pobox.com>
Date
Aug 14, 2014, 17:17 UTC
Message-ID
<xmqqy4ur0z46.fsf@gitster.dls.corp.google.com>
In-Reply-To
<1407927454-9268-7-git-send-email-pclouds@gmail.com>
Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
Show 34 quoted lines
> If we are given two SHA-1 and asked to determine if they are different
> (but not _what_ differences), we know right away by comparing SHA-1.
>
> A side effect of this patch is, because large files are marked binary,
> diff-tree will not need to unpack them. 'diff-index --cached' will not
> either. But 'diff-files' still does.
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  diff.c           | 13 +++++++++++++
>  t/t1050-large.sh |  8 ++++++++
>  2 files changed, 21 insertions(+)
>
> diff --git a/diff.c b/diff.c
> index d381a6f..b85bcfb 100644
> --- a/diff.c
> +++ b/diff.c
> @@ -2324,6 +2324,19 @@ static void builtin_diff(const char *name_a,
>  	} else if (!DIFF_OPT_TST(o, TEXT) &&
>  	    ( (!textconv_one && diff_filespec_is_binary(one)) ||
>  	      (!textconv_two && diff_filespec_is_binary(two)) )) {
> +		if (!one->data && !two->data &&
> +		    S_ISREG(one->mode) && S_ISREG(two->mode) &&
> +		    !DIFF_OPT_TST(o, BINARY)) {
> +			if (!hashcmp(one->sha1, two->sha1)) {
> +				if (must_show_header)
> +					fprintf(o->file, "%s", header.buf);
> +				goto free_ab_and_return;
> +			}
> +			fprintf(o->file, "%s", header.buf);
> +			fprintf(o->file, "%sBinary files %s and %s differ\n",
> +				line_prefix, lbl[0], lbl[1]);
> +			goto free_ab_and_return;
> +		}
A tangent.

I think one and two can point at the same object only when this filepair is involved in rename/copy. In other words, one and two with the same <mode,sha1,name> would not be given to this code. And must-show-header would be set to true long before we get here in fill-metainfo in such a case.

I think this new code and the original below which you copied this one from can probably be simplified. It already felt wrong to see two copies of "fprintf(o->file "%s", header.buf)" and now we have four of them. Because this is a copy-and-paste of the identical logic from below, I do not want you to attempt fixing this tangent in this patch, though.

Thanks.
Show 22 quoted lines
>  		if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
>  			die("unable to read files to diff");
>  		/* Quite common confusing case */
> diff --git a/t/t1050-large.sh b/t/t1050-large.sh
> index 711f22c..b294963 100755
> --- a/t/t1050-large.sh
> +++ b/t/t1050-large.sh
> @@ -116,6 +116,14 @@ test_expect_success 'diff --stat' '
>  	git diff --stat HEAD^ HEAD
>  '
>  
> +test_expect_success 'diff' '
> +	git diff HEAD^ HEAD
> +'
> +
> +test_expect_success 'diff --cached' '
> +	git diff --cached HEAD^
> +'
> +
>  test_expect_success 'hash-object' '
>  	git hash-object large1
>  '
Previous: Duy NguyenNext: Nguyễn Thái Ngọc Duy
Message 42 of 49 in “Git chokes on large file”
  1. Dale R. WorleyMay 27, 2014
  2. Duy NguyenMay 28, 2014
  3. Junio C HamanoMay 28, 2014
  4. Dale R. WorleyMay 28, 2014
  5. Dale R. WorleyMay 28, 2014
  6. David LangMay 28, 2014
  7. Dale R. WorleyMay 28, 2014
  8. David LangMay 28, 2014
  9. Dale R. WorleyMay 29, 2014
  10. Junio C HamanoMay 28, 2014
  11. David LangMay 28, 2014
  12. 1/4 wrapper.c: introduce gentle xmallocz that does not die()Nguyễn Thái Ngọc Duy, May 29, 2014
  13. 2/4 fsck: do not die when not enough memory to examine a pack entryNguyễn Thái Ngọc Duy, May 29, 2014
  14. 3/4 diff.c: allow to pass more flags to diff_populate_filespecNguyễn Thái Ngọc Duy, May 29, 2014
  15. 4/4 diff: mark any file larger than core.bigfilethreshold binaryNguyễn Thái Ngọc Duy, May 29, 2014
  16. Thomas BraunJun 19, 2014
  17. Duy NguyenJun 23, 2014
  18. Thomas BraunJun 23, 2014
  19. 1/4 wrapper.c: introduce gentle xmallocz that does not die()Nguyễn Thái Ngọc Duy, Jun 24, 2014
  20. 2/4 fsck: do not die when not enough memory to examine a pack entryNguyễn Thái Ngọc Duy, Jun 24, 2014
  21. Junio C HamanoJun 26, 2014
  22. Duy NguyenJun 29, 2014
  23. 3/4 diff.c: allow to pass more flags to diff_populate_filespecNguyễn Thái Ngọc Duy, Jun 24, 2014
  24. 4/4 diff: mark any file larger than core.bigfilethreshold binaryNguyễn Thái Ngọc Duy, Jun 24, 2014
  25. Junio C HamanoJun 26, 2014
  26. Thomas BraunJun 27, 2014
  27. Duy NguyenJun 29, 2014
  28. 0/6 Large file improvementsNguyễn Thái Ngọc Duy, Aug 13, 2014
  29. 1/6 wrapper.c: introduce gentle xmalloc(z) that does not die()Nguyễn Thái Ngọc Duy, Aug 13, 2014
  30. Junio C HamanoAug 14, 2014
  31. 2/6 sha1_file.c: do not die failing to malloc in unpack_compressed_entryNguyễn Thái Ngọc Duy, Aug 13, 2014
  32. Junio C HamanoAug 13, 2014
  33. 3/6 unpack-objects: continue when fail to malloc due to large objectsNguyễn Thái Ngọc Duy, Aug 13, 2014
  34. Junio C HamanoAug 14, 2014
  35. Duy NguyenAug 15, 2014
  36. 4/6 diff.c: allow to pass more flags to diff_populate_filespecNguyễn Thái Ngọc Duy, Aug 13, 2014
  37. 5/6 diff --stat: mark any file larger than core.bigfilethreshold binaryNguyễn Thái Ngọc Duy, Aug 13, 2014
  38. Eric SunshineAug 13, 2014
  39. 6/6 diff: shortcut for diff'ing two binary SHA-1 objectsNguyễn Thái Ngọc Duy, Aug 13, 2014
  40. Junio C HamanoAug 14, 2014
  41. Duy NguyenAug 15, 2014
  42. Junio C HamanoAug 14, 2014
  43. 0/5 Large file improvementsNguyễn Thái Ngọc Duy, Aug 16, 2014
  44. 1/5 wrapper.c: introduce gentle xmallocz that does not die()Nguyễn Thái Ngọc Duy, Aug 16, 2014
  45. 2/5 sha1_file.c: do not die failing to malloc in unpack_compressed_entryNguyễn Thái Ngọc Duy, Aug 16, 2014
  46. 3/5 diff.c: allow to pass more flags to diff_populate_filespecNguyễn Thái Ngọc Duy, Aug 16, 2014
  47. 4/5 diff --stat: mark any file larger than core.bigfilethreshold binaryNguyễn Thái Ngọc Duy, Aug 16, 2014
  48. 5/5 diff: shortcut for diff'ing two binary SHA-1 objectsNguyễn Thái Ngọc Duy, Aug 16, 2014
  49. Thomas BraunMay 28, 2014

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.