Merge branch 'cc/bisect' (early part)
[gitweb.git] / builtin-rev-list.c
index aa3c962e56bd08e8e2a632b7170283d05efb69a2..4ba1c12e0b6fedb056aebed6ffc8f8bb8d6cfd18 100644 (file)
@@ -1,20 +1,12 @@
 #include "cache.h"
-#include "refs.h"
-#include "tag.h"
 #include "commit.h"
-#include "tree.h"
-#include "blob.h"
-#include "tree-walk.h"
 #include "diff.h"
 #include "revision.h"
 #include "list-objects.h"
 #include "builtin.h"
 #include "log-tree.h"
 #include "graph.h"
-
-/* bits #0-15 in revision.h */
-
-#define COUNTED                (1u<<16)
+#include "bisect.h"
 
 static const char rev_list_usage[] =
 "git rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
@@ -50,74 +42,69 @@ static const char rev_list_usage[] =
 "    --bisect-all"
 ;
 
-static struct rev_info revs;
-
-static int bisect_list;
-static int show_timestamp;
-static int hdr_termination;
-static const char *header_prefix;
-
-static void finish_commit(struct commit *commit);
-static void show_commit(struct commit *commit)
+static void finish_commit(struct commit *commit, void *data);
+static void show_commit(struct commit *commit, void *data)
 {
-       graph_show_commit(revs.graph);
+       struct rev_list_info *info = data;
+       struct rev_info *revs = info->revs;
+
+       graph_show_commit(revs->graph);
 
-       if (show_timestamp)
+       if (info->show_timestamp)
                printf("%lu ", commit->date);
-       if (header_prefix)
-               fputs(header_prefix, stdout);
+       if (info->header_prefix)
+               fputs(info->header_prefix, stdout);
 
-       if (!revs.graph) {
+       if (!revs->graph) {
                if (commit->object.flags & BOUNDARY)
                        putchar('-');
                else if (commit->object.flags & UNINTERESTING)
                        putchar('^');
-               else if (revs.left_right) {
+               else if (revs->left_right) {
                        if (commit->object.flags & SYMMETRIC_LEFT)
                                putchar('<');
                        else
                                putchar('>');
                }
        }
-       if (revs.abbrev_commit && revs.abbrev)
-               fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
+       if (revs->abbrev_commit && revs->abbrev)
+               fputs(find_unique_abbrev(commit->object.sha1, revs->abbrev),
                      stdout);
        else
                fputs(sha1_to_hex(commit->object.sha1), stdout);
-       if (revs.print_parents) {
+       if (revs->print_parents) {
                struct commit_list *parents = commit->parents;
                while (parents) {
                        printf(" %s", sha1_to_hex(parents->item->object.sha1));
                        parents = parents->next;
                }
        }
-       if (revs.children.name) {
+       if (revs->children.name) {
                struct commit_list *children;
 
-               children = lookup_decoration(&revs.children, &commit->object);
+               children = lookup_decoration(&revs->children, &commit->object);
                while (children) {
                        printf(" %s", sha1_to_hex(children->item->object.sha1));
                        children = children->next;
                }
        }
-       show_decorations(commit);
-       if (revs.commit_format == CMIT_FMT_ONELINE)
+       show_decorations(revs, commit);
+       if (revs->commit_format == CMIT_FMT_ONELINE)
                putchar(' ');
        else
                putchar('\n');
 
-       if (revs.verbose_header && commit->buffer) {
-               struct strbuf buf;
-               strbuf_init(&buf, 0);
-               pretty_print_commit(revs.commit_format, commit,
-                                   &buf, revs.abbrev, NULL, NULL,
-                                   revs.date_mode, 0);
-               if (revs.graph) {
+       if (revs->verbose_header && commit->buffer) {
+               struct strbuf buf = STRBUF_INIT;
+               pretty_print_commit(revs->commit_format, commit,
+                                   &buf, revs->abbrev, NULL, NULL,
+                                   revs->date_mode, 0);
+               if (revs->graph) {
                        if (buf.len) {
-                               if (revs.commit_format != CMIT_FMT_ONELINE)
-                                       graph_show_oneline(revs.graph);
+                               if (revs->commit_format != CMIT_FMT_ONELINE)
+                                       graph_show_oneline(revs->graph);
 
-                               graph_show_commit_msg(revs.graph, &buf);
+                               graph_show_commit_msg(revs->graph, &buf);
 
                                /*
                                 * Add a newline after the commit message.
@@ -135,7 +122,7 @@ static void show_commit(struct commit *commit)
                                 * format doesn't explicitly end in a newline.)
                                 */
                                if (buf.len && buf.buf[buf.len - 1] == '\n')
-                                       graph_show_padding(revs.graph);
+                                       graph_show_padding(revs->graph);
                                putchar('\n');
                        } else {
                                /*
@@ -143,23 +130,23 @@ static void show_commit(struct commit *commit)
                                 * the rest of the graph output for this
                                 * commit.
                                 */
-                               if (graph_show_remainder(revs.graph))
+                               if (graph_show_remainder(revs->graph))
                                        putchar('\n');
                        }
                } else {
                        if (buf.len)
-                               printf("%s%c", buf.buf, hdr_termination);
+                               printf("%s%c", buf.buf, info->hdr_termination);
                }
                strbuf_release(&buf);
        } else {
-               if (graph_show_remainder(revs.graph))
+               if (graph_show_remainder(revs->graph))
                        putchar('\n');
        }
        maybe_flush_or_die(stdout, "stdout");
-       finish_commit(commit);
+       finish_commit(commit, data);
 }
 
-static void finish_commit(struct commit *commit)
+static void finish_commit(struct commit *commit, void *data)
 {
        if (commit->parents) {
                free_commit_list(commit->parents);
@@ -199,389 +186,127 @@ static void show_edge(struct commit *commit)
        printf("-%s\n", sha1_to_hex(commit->object.sha1));
 }
 
-/*
- * This is a truly stupid algorithm, but it's only
- * used for bisection, and we just don't care enough.
- *
- * We care just barely enough to avoid recursing for
- * non-merge entries.
- */
-static int count_distance(struct commit_list *entry)
+static inline int log2i(int n)
 {
-       int nr = 0;
-
-       while (entry) {
-               struct commit *commit = entry->item;
-               struct commit_list *p;
-
-               if (commit->object.flags & (UNINTERESTING | COUNTED))
-                       break;
-               if (!(commit->object.flags & TREESAME))
-                       nr++;
-               commit->object.flags |= COUNTED;
-               p = commit->parents;
-               entry = p;
-               if (p) {
-                       p = p->next;
-                       while (p) {
-                               nr += count_distance(p);
-                               p = p->next;
-                       }
-               }
-       }
+       int log2 = 0;
 
-       return nr;
-}
+       for (; n > 1; n >>= 1)
+               log2++;
 
-static void clear_distance(struct commit_list *list)
-{
-       while (list) {
-               struct commit *commit = list->item;
-               commit->object.flags &= ~COUNTED;
-               list = list->next;
-       }
+       return log2;
 }
 
-#define DEBUG_BISECT 0
-
-static inline int weight(struct commit_list *elem)
+static inline int exp2i(int n)
 {
-       return *((int*)(elem->item->util));
+       return 1 << n;
 }
 
-static inline void weight_set(struct commit_list *elem, int weight)
+/*
+ * Estimate the number of bisect steps left (after the current step)
+ *
+ * For any x between 0 included and 2^n excluded, the probability for
+ * n - 1 steps left looks like:
+ *
+ * P(2^n + x) == (2^n - x) / (2^n + x)
+ *
+ * and P(2^n + x) < 0.5 means 2^n < 3x
+ */
+int estimate_bisect_steps(int all)
 {
-       *((int*)(elem->item->util)) = weight;
-}
+       int n, x, e;
 
-static int count_interesting_parents(struct commit *commit)
-{
-       struct commit_list *p;
-       int count;
+       if (all < 3)
+               return 0;
 
-       for (count = 0, p = commit->parents; p; p = p->next) {
-               if (p->item->object.flags & UNINTERESTING)
-                       continue;
-               count++;
-       }
-       return count;
-}
+       n = log2i(all);
+       e = exp2i(n);
+       x = all - e;
 
-static inline int halfway(struct commit_list *p, int nr)
-{
-       /*
-        * Don't short-cut something we are not going to return!
-        */
-       if (p->item->object.flags & TREESAME)
-               return 0;
-       if (DEBUG_BISECT)
-               return 0;
-       /*
-        * 2 and 3 are halfway of 5.
-        * 3 is halfway of 6 but 2 and 4 are not.
-        */
-       switch (2 * weight(p) - nr) {
-       case -1: case 0: case 1:
-               return 1;
-       default:
-               return 0;
-       }
+       return (e < 3 * x) ? n : n - 1;
 }
 
-#if !DEBUG_BISECT
-#define show_list(a,b,c,d) do { ; } while (0)
-#else
-static void show_list(const char *debug, int counted, int nr,
-                     struct commit_list *list)
+void print_commit_list(struct commit_list *list,
+                      const char *format_cur,
+                      const char *format_last)
 {
-       struct commit_list *p;
-
-       fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
-
-       for (p = list; p; p = p->next) {
-               struct commit_list *pp;
-               struct commit *commit = p->item;
-               unsigned flags = commit->object.flags;
-               enum object_type type;
-               unsigned long size;
-               char *buf = read_sha1_file(commit->object.sha1, &type, &size);
-               char *ep, *sp;
-
-               fprintf(stderr, "%c%c%c ",
-                       (flags & TREESAME) ? ' ' : 'T',
-                       (flags & UNINTERESTING) ? 'U' : ' ',
-                       (flags & COUNTED) ? 'C' : ' ');
-               if (commit->util)
-                       fprintf(stderr, "%3d", weight(p));
-               else
-                       fprintf(stderr, "---");
-               fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
-               for (pp = commit->parents; pp; pp = pp->next)
-                       fprintf(stderr, " %.*s", 8,
-                               sha1_to_hex(pp->item->object.sha1));
-
-               sp = strstr(buf, "\n\n");
-               if (sp) {
-                       sp += 2;
-                       for (ep = sp; *ep && *ep != '\n'; ep++)
-                               ;
-                       fprintf(stderr, " %.*s", (int)(ep - sp), sp);
-               }
-               fprintf(stderr, "\n");
+       for ( ; list; list = list->next) {
+               const char *format = list->next ? format_cur : format_last;
+               printf(format, sha1_to_hex(list->item->object.sha1));
        }
 }
-#endif /* DEBUG_BISECT */
 
-static struct commit_list *best_bisection(struct commit_list *list, int nr)
+static void show_tried_revs(struct commit_list *tried)
 {
-       struct commit_list *p, *best;
-       int best_distance = -1;
-
-       best = list;
-       for (p = list; p; p = p->next) {
-               int distance;
-               unsigned flags = p->item->object.flags;
-
-               if (flags & TREESAME)
-                       continue;
-               distance = weight(p);
-               if (nr - distance < distance)
-                       distance = nr - distance;
-               if (distance > best_distance) {
-                       best = p;
-                       best_distance = distance;
-               }
-       }
-
-       return best;
+       printf("bisect_tried='");
+       print_commit_list(tried, "%s|", "%s");
+       printf("'\n");
 }
 
-struct commit_dist {
-       struct commit *commit;
-       int distance;
-};
-
-static int compare_commit_dist(const void *a_, const void *b_)
+static void print_var_str(const char *var, const char *val)
 {
-       struct commit_dist *a, *b;
-
-       a = (struct commit_dist *)a_;
-       b = (struct commit_dist *)b_;
-       if (a->distance != b->distance)
-               return b->distance - a->distance; /* desc sort */
-       return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
+       printf("%s='%s'\n", var, val);
 }
 
-static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
+static void print_var_int(const char *var, int val)
 {
-       struct commit_list *p;
-       struct commit_dist *array = xcalloc(nr, sizeof(*array));
-       int cnt, i;
-
-       for (p = list, cnt = 0; p; p = p->next) {
-               int distance;
-               unsigned flags = p->item->object.flags;
-
-               if (flags & TREESAME)
-                       continue;
-               distance = weight(p);
-               if (nr - distance < distance)
-                       distance = nr - distance;
-               array[cnt].commit = p->item;
-               array[cnt].distance = distance;
-               cnt++;
-       }
-       qsort(array, cnt, sizeof(*array), compare_commit_dist);
-       for (p = list, i = 0; i < cnt; i++) {
-               struct name_decoration *r = xmalloc(sizeof(*r) + 100);
-               struct object *obj = &(array[i].commit->object);
-
-               sprintf(r->name, "dist=%d", array[i].distance);
-               r->next = add_decoration(&name_decoration, obj, r);
-               p->item = array[i].commit;
-               p = p->next;
-       }
-       if (p)
-               p->next = NULL;
-       free(array);
-       return list;
+       printf("%s=%d\n", var, val);
 }
 
-/*
- * zero or positive weight is the number of interesting commits it can
- * reach, including itself.  Especially, weight = 0 means it does not
- * reach any tree-changing commits (e.g. just above uninteresting one
- * but traversal is with pathspec).
- *
- * weight = -1 means it has one parent and its distance is yet to
- * be computed.
- *
- * weight = -2 means it has more than one parent and its distance is
- * unknown.  After running count_distance() first, they will get zero
- * or positive distance.
- */
-static struct commit_list *do_find_bisection(struct commit_list *list,
-                                            int nr, int *weights,
-                                            int find_all)
+int show_bisect_vars(struct rev_list_info *info, int reaches, int all)
 {
-       int n, counted;
-       struct commit_list *p;
-
-       counted = 0;
-
-       for (n = 0, p = list; p; p = p->next) {
-               struct commit *commit = p->item;
-               unsigned flags = commit->object.flags;
-
-               p->item->util = &weights[n++];
-               switch (count_interesting_parents(commit)) {
-               case 0:
-                       if (!(flags & TREESAME)) {
-                               weight_set(p, 1);
-                               counted++;
-                               show_list("bisection 2 count one",
-                                         counted, nr, list);
-                       }
-                       /*
-                        * otherwise, it is known not to reach any
-                        * tree-changing commit and gets weight 0.
-                        */
-                       break;
-               case 1:
-                       weight_set(p, -1);
-                       break;
-               default:
-                       weight_set(p, -2);
-                       break;
-               }
-       }
+       int cnt, flags = info->bisect_show_flags;
+       char hex[41] = "";
+       struct commit_list *tried;
+       struct rev_info *revs = info->revs;
+
+       if (!revs->commits && !(flags & BISECT_SHOW_TRIED))
+               return 1;
 
-       show_list("bisection 2 initialize", counted, nr, list);
+       revs->commits = filter_skipped(revs->commits, &tried,
+                                      flags & BISECT_SHOW_ALL,
+                                      NULL, NULL);
 
        /*
-        * If you have only one parent in the resulting set
-        * then you can reach one commit more than that parent
-        * can reach.  So we do not have to run the expensive
-        * count_distance() for single strand of pearls.
-        *
-        * However, if you have more than one parents, you cannot
-        * just add their distance and one for yourself, since
-        * they usually reach the same ancestor and you would
-        * end up counting them twice that way.
-        *
-        * So we will first count distance of merges the usual
-        * way, and then fill the blanks using cheaper algorithm.
+        * revs->commits can reach "reaches" commits among
+        * "all" commits.  If it is good, then there are
+        * (all-reaches) commits left to be bisected.
+        * On the other hand, if it is bad, then the set
+        * to bisect is "reaches".
+        * A bisect set of size N has (N-1) commits further
+        * to test, as we already know one bad one.
         */
-       for (p = list; p; p = p->next) {
-               if (p->item->object.flags & UNINTERESTING)
-                       continue;
-               if (weight(p) != -2)
-                       continue;
-               weight_set(p, count_distance(p));
-               clear_distance(list);
-
-               /* Does it happen to be at exactly half-way? */
-               if (!find_all && halfway(p, nr))
-                       return p;
-               counted++;
-       }
-
-       show_list("bisection 2 count_distance", counted, nr, list);
+       cnt = all - reaches;
+       if (cnt < reaches)
+               cnt = reaches;
 
-       while (counted < nr) {
-               for (p = list; p; p = p->next) {
-                       struct commit_list *q;
-                       unsigned flags = p->item->object.flags;
+       if (revs->commits)
+               strcpy(hex, sha1_to_hex(revs->commits->item->object.sha1));
 
-                       if (0 <= weight(p))
-                               continue;
-                       for (q = p->item->parents; q; q = q->next) {
-                               if (q->item->object.flags & UNINTERESTING)
-                                       continue;
-                               if (0 <= weight(q))
-                                       break;
-                       }
-                       if (!q)
-                               continue;
-
-                       /*
-                        * weight for p is unknown but q is known.
-                        * add one for p itself if p is to be counted,
-                        * otherwise inherit it from q directly.
-                        */
-                       if (!(flags & TREESAME)) {
-                               weight_set(p, weight(q)+1);
-                               counted++;
-                               show_list("bisection 2 count one",
-                                         counted, nr, list);
-                       }
-                       else
-                               weight_set(p, weight(q));
-
-                       /* Does it happen to be at exactly half-way? */
-                       if (!find_all && halfway(p, nr))
-                               return p;
-               }
+       if (flags & BISECT_SHOW_ALL) {
+               traverse_commit_list(revs, show_commit, show_object, info);
+               printf("------\n");
        }
 
-       show_list("bisection 2 counted all", counted, nr, list);
+       if (flags & BISECT_SHOW_TRIED)
+               show_tried_revs(tried);
 
-       if (!find_all)
-               return best_bisection(list, nr);
-       else
-               return best_bisection_sorted(list, nr);
-}
-
-static struct commit_list *find_bisection(struct commit_list *list,
-                                         int *reaches, int *all,
-                                         int find_all)
-{
-       int nr, on_list;
-       struct commit_list *p, *best, *next, *last;
-       int *weights;
+       print_var_str("bisect_rev", hex);
+       print_var_int("bisect_nr", cnt - 1);
+       print_var_int("bisect_good", all - reaches - 1);
+       print_var_int("bisect_bad", reaches - 1);
+       print_var_int("bisect_all", all);
+       print_var_int("bisect_steps", estimate_bisect_steps(all));
 
-       show_list("bisection 2 entry", 0, 0, list);
-
-       /*
-        * Count the number of total and tree-changing items on the
-        * list, while reversing the list.
-        */
-       for (nr = on_list = 0, last = NULL, p = list;
-            p;
-            p = next) {
-               unsigned flags = p->item->object.flags;
-
-               next = p->next;
-               if (flags & UNINTERESTING)
-                       continue;
-               p->next = last;
-               last = p;
-               if (!(flags & TREESAME))
-                       nr++;
-               on_list++;
-       }
-       list = last;
-       show_list("bisection 2 sorted", 0, nr, list);
-
-       *all = nr;
-       weights = xcalloc(on_list, sizeof(*weights));
-
-       /* Do the real work of finding bisection commit. */
-       best = do_find_bisection(list, nr, weights, find_all);
-       if (best) {
-               if (!find_all)
-                       best->next = NULL;
-               *reaches = weight(best);
-       }
-       free(weights);
-       return best;
+       return 0;
 }
 
 int cmd_rev_list(int argc, const char **argv, const char *prefix)
 {
-       struct commit_list *list;
+       struct rev_info revs;
+       struct rev_list_info info;
        int i;
        int read_from_stdin = 0;
+       int bisect_list = 0;
        int bisect_show_vars = 0;
        int bisect_find_all = 0;
        int quiet = 0;
@@ -592,6 +317,9 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
        revs.commit_format = CMIT_FMT_UNSPECIFIED;
        argc = setup_revisions(argc, argv, &revs, NULL);
 
+       memset(&info, 0, sizeof(info));
+       info.revs = &revs;
+
        quiet = DIFF_OPT_TST(&revs.diffopt, QUIET);
        for (i = 1 ; i < argc; i++) {
                const char *arg = argv[i];
@@ -601,7 +329,7 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
                        continue;
                }
                if (!strcmp(arg, "--timestamp")) {
-                       show_timestamp = 1;
+                       info.show_timestamp = 1;
                        continue;
                }
                if (!strcmp(arg, "--bisect")) {
@@ -611,6 +339,8 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
                if (!strcmp(arg, "--bisect-all")) {
                        bisect_list = 1;
                        bisect_find_all = 1;
+                       info.bisect_show_flags = BISECT_SHOW_ALL;
+                       revs.show_decorations = 1;
                        continue;
                }
                if (!strcmp(arg, "--bisect-vars")) {
@@ -629,19 +359,17 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
        }
        if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
                /* The command line has a --pretty  */
-               hdr_termination = '\n';
+               info.hdr_termination = '\n';
                if (revs.commit_format == CMIT_FMT_ONELINE)
-                       header_prefix = "";
+                       info.header_prefix = "";
                else
-                       header_prefix = "commit ";
+                       info.header_prefix = "commit ";
        }
        else if (revs.verbose_header)
                /* Only --header was specified */
                revs.commit_format = CMIT_FMT_RAW;
 
-       list = revs.commits;
-
-       if ((!list &&
+       if ((!revs.commits &&
             (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
              !revs.pending.nr)) ||
            revs.diff)
@@ -662,47 +390,15 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
 
                revs.commits = find_bisection(revs.commits, &reaches, &all,
                                              bisect_find_all);
-               if (bisect_show_vars) {
-                       int cnt;
-                       char hex[41];
-                       if (!revs.commits)
-                               return 1;
-                       /*
-                        * revs.commits can reach "reaches" commits among
-                        * "all" commits.  If it is good, then there are
-                        * (all-reaches) commits left to be bisected.
-                        * On the other hand, if it is bad, then the set
-                        * to bisect is "reaches".
-                        * A bisect set of size N has (N-1) commits further
-                        * to test, as we already know one bad one.
-                        */
-                       cnt = all - reaches;
-                       if (cnt < reaches)
-                               cnt = reaches;
-                       strcpy(hex, sha1_to_hex(revs.commits->item->object.sha1));
-
-                       if (bisect_find_all) {
-                               traverse_commit_list(&revs, show_commit, show_object);
-                               printf("------\n");
-                       }
 
-                       printf("bisect_rev=%s\n"
-                              "bisect_nr=%d\n"
-                              "bisect_good=%d\n"
-                              "bisect_bad=%d\n"
-                              "bisect_all=%d\n",
-                              hex,
-                              cnt - 1,
-                              all - reaches - 1,
-                              reaches - 1,
-                              all);
-                       return 0;
-               }
+               if (bisect_show_vars)
+                       return show_bisect_vars(&info, reaches, all);
        }
 
        traverse_commit_list(&revs,
-               quiet ? finish_commit : show_commit,
-               quiet ? finish_object : show_object);
+                            quiet ? finish_commit : show_commit,
+                            quiet ? finish_object : show_object,
+                            &info);
 
        return 0;
 }