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

Re: [PATCH 1/4] xdiff: reduce size of action arrays

From
Junio C Hamano <gitster@pobox.com>
Date
Apr 2, 2026, 19:19 UTC
Message-ID
<xmqqy0j5p3ur.fsf@gitster.g>
In-Reply-To
<447b8c0af1746d61bfa26e7908a784583ab5dc2e.1775141855.git.phillip.wood@dunelm.org.uk>
Phillip Wood <phillip.wood123@gmail.com> writes:
Show 12 quoted lines
> From: Phillip Wood <phillip.wood@dunelm.org.uk>
>
> When the myers algorithm is selected the input files are pre-processed
> to remove any common prefix and suffix. Then any lines that appear
> only in one side of the diff are marked as changed and frequently
> occurring lines are marked as changed if they are adjacent to a
> changed line. This step requires a couple of temporary arrays. As as
> the common prefix and suffix have already been removed, the arrays
> only need to be big enough to hold the lines between them, not the
> whole file. Reduce the size of the arrays and adjust the loops that
> use them accordingly while taking care to keep indexing the arrays
> in xdfile_t with absolute line numbers.
"As as"???
Show 31 quoted lines
> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
> ---
>  xdiff/xprepare.c | 31 +++++++++++++++++--------------
>  1 file changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/xdiff/xprepare.c b/xdiff/xprepare.c
> index 1f2e8c6b4b9..4bb3a8ef41c 100644
> --- a/xdiff/xprepare.c
> +++ b/xdiff/xprepare.c
> @@ -273,16 +273,19 @@ static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xd
>  	uint8_t *action1 = NULL, *action2 = NULL;
>  	bool need_min = !!(cf->flags & XDF_NEED_MINIMAL);
>  	int ret = 0;
> +	ptrdiff_t off = xdf1->dstart;
> +	ptrdiff_t len1 = xdf1->dend - off + 1;
> +	ptrdiff_t len2 = xdf2->dend - off + 1;
>  
>  	/*
>  	 * Create temporary arrays that will help us decide if
>  	 * changed[i] should remain false, or become true.
>  	 */
> -	if (!XDL_CALLOC_ARRAY(action1, xdf1->nrec + 1)) {
> +	if (!XDL_CALLOC_ARRAY(action1, len1)) {
>  		ret = -1;
>  		goto cleanup;
>  	}
> -	if (!XDL_CALLOC_ARRAY(action2, xdf2->nrec + 1)) {
> +	if (!XDL_CALLOC_ARRAY(action2, len2)) {
>  		ret = -1;
>  		goto cleanup;
>  	}

OK, so we used to allocate for the whole thing, but now we only allocate for lines starting at dstart. "off" is the difference between [i], the index into these action arrays, and the true line numbers.

Show 8 quoted lines
> @@ -299,8 +302,8 @@ static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xd
>  	/*
>  	 * Initialize temporary arrays with DISCARD, KEEP, or INVESTIGATE.
>  	 */
> -	for (i = xdf1->dstart; i <= xdf1->dend; i++) {
> -		size_t mph1 = xdf1->recs[i].minimal_perfect_hash;
> +	for (i = 0; i < len1; i++) {
> +		size_t mph1 = xdf1->recs[i + off].minimal_perfect_hash;

And we iterate as many times as we have entries in the action array, but we need to offset the [i] with off when looking at the record.

Show 11 quoted lines
>  		rcrec = cf->rcrecs[mph1];
>  		nm = rcrec ? rcrec->len2 : 0;
>  		if (nm == 0)
> @@ -311,8 +314,8 @@ static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xd
>  			action1[i] = INVESTIGATE;
>  	}
>  
> -	for (i = xdf2->dstart; i <= xdf2->dend; i++) {
> -		size_t mph2 = xdf2->recs[i].minimal_perfect_hash;
> +	for (i = 0; i < len2; i++) {
> +		size_t mph2 = xdf2->recs[i + off].minimal_perfect_hash;
Likewise.
Show 9 quoted lines
> @@ -328,37 +331,37 @@ static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xd
>  	 * false, or become true.
>  	 */
>  	xdf1->nreff = 0;
> -	for (i = xdf1->dstart; i <= xdf1->dend; i++) {
> +	for (i = 0; i < len1; i++) {
>  		if (action1[i] == INVESTIGATE) {
> -			if (!xdl_clean_mmatch(action1, i, xdf1->dstart, xdf1->dend))
> +			if (!xdl_clean_mmatch(action1, i, 0, len1 - 1))

Let me think aloud to see if I can follow the logic here. Looking at the implementation of the xdl_clean_mmatch() function, it takes an action array, an offset 'i' into it (starting from dstart), the beginning 'dstart' and the end 'dend' offsets. The idea is that an index derived from 'i' is used to index the action array and the beginning and the end offsets are used to limit how much far the access can deviate from 'i'.

Now we stripped the first xdf1->dstart elements from action1[] array, 'i' in this loop runs from 0 (i.e. one beyond the initial common section) and len1 (i.e. one before the tail end of the common section), i.e., everything is consistently shifted down by xdf1->dstart in this call. So xdl_clean_mmatch() does not even need to know that it is fed a shortened action[] array.

Show 12 quoted lines
>  				action1[i] = KEEP;
>  			else
>  				action1[i] = DISCARD;
>  		}
>  
>  		if (action1[i] == KEEP) {
> -			xdf1->reference_index[xdf1->nreff++] = i;
> +			xdf1->reference_index[xdf1->nreff++] = i + off;
>  			/* changed[i] remains false */
>  		} else if (action1[i] == DISCARD)
> -			xdf1->changed[i] = true;
> +			xdf1->changed[i + off] = true;

But these two arrays are not shrunk, so we need to compensate by the 'off' offset.

And the remainder is similar but for xdf2 instead of xdf1 above.
Show 25 quoted lines
>  		else
>  			BUG("Illegal state for action1[i]");
>  	}
>  
>  	xdf2->nreff = 0;
> -	for (i = xdf2->dstart; i <= xdf2->dend; i++) {
> +	for (i = 0; i < len2; i++) {
>  		if (action2[i] == INVESTIGATE) {
> -			if (!xdl_clean_mmatch(action2, i, xdf2->dstart, xdf2->dend))
> +			if (!xdl_clean_mmatch(action2, i, 0, len2 - 1))
>  				action2[i] = KEEP;
>  			else
>  				action2[i] = DISCARD;
>  		}
>  
>  		if (action2[i] == KEEP) {
> -			xdf2->reference_index[xdf2->nreff++] = i;
> +			xdf2->reference_index[xdf2->nreff++] = i + off;
>  			/* changed[i] remains false */
>  		} else if (action2[i] == DISCARD)
> -			xdf2->changed[i] = true;
> +			xdf2->changed[i + off] = true;
>  		else
>  			BUG("Illegal state for action2[i]");
>  	}
Previous: Phillip WoodNext: Phillip Wood
Message 3 of 14 in “xdiff: reduce the size of a couple of arrays”
  1. 0/4 xdiff: reduce the size of a couple of arraysPhillip Wood, Apr 2, 2026
  2. 1/4 xdiff: reduce size of action arraysPhillip Wood, Apr 2, 2026
  3. Junio C HamanoApr 2, 2026
  4. 2/4 xdiff: cleanup xdl_clean_mmatch()Phillip Wood, Apr 2, 2026
  5. Junio C HamanoApr 2, 2026
  6. 3/4 xprepare: simplify error handlingPhillip Wood, Apr 2, 2026
  7. Junio C HamanoApr 2, 2026
  8. 4/4 xdiff: reduce the size of arrayPhillip Wood, Apr 2, 2026
  9. Junio C HamanoApr 2, 2026
  10. 0/4 xdiff: reduce the size of a couple of arraysPhillip Wood, May 4, 2026
  11. 1/4 xdiff: reduce size of action arraysPhillip Wood, May 4, 2026
  12. 2/4 xdiff: cleanup xdl_clean_mmatch()Phillip Wood, May 4, 2026
  13. 3/4 xprepare: simplify error handlingPhillip Wood, May 4, 2026
  14. 4/4 xdiff: reduce the size of arrayPhillip Wood, May 4, 2026

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.