> The Jonker-Volgenant algorithm was implemented to answer questions such
> as: given two different versions of a topic branch (or iterations of a
> patch series), what is the best pairing of commits/patches between the
> different versions?
>
> Signed-off-by: Johannes Schindelin <johannes.schindelin@xxxxxx>
> ---
> Makefile | 1 +
> hungarian.c | 205 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> hungarian.h | 19 +++++
(Nit: I personally don't really like these filenames, I know they will
surprise and distract me every time I notice them for years to come... :)
> +int compute_assignment(int column_count, int row_count, double *cost,
> + int *column2row, int *row2column)
> +{
> + double *v = xmalloc(sizeof(double) * column_count), *d;
> + int *free_row, free_count = 0, saved_free_count, *pred, *col;
> + int i, j, phase;
<snip>
> + for (free_count = 0; free_count < saved_free_count; free_count++) {
> + int i1 = free_row[free_count], low = 0, up = 0, last, k;
> + double min, c, u1;
<snip most of the loop's body>
> + /* augmentation */
> + do {
> + if (j < 0)
> + BUG("negative j: %d", j);
> + i = pred[j];
> + column2row[j] = i;
> + k = j;
> + j = row2column[i];
> + row2column[i] = k;
Coccinelle suggests to replace the last three lines above with:
SWAP(j, row2column[i]);
I think it's right, using the SWAP macro makes the resulting code not
only shorter and clearer, but it also saves the reader from thinking
about whether it's important to set 'k = j' (I think it's not), or 'k'
is just used here in lieu of a dedicated 'tmp' variable (I think it
is).
> + } while (i1 != i);
> + }
> +
