Mercurial > public > mercurial-scm > hg-stable
annotate mercurial/stabletailgraph/stabletailsort.py @ 50460:f0d2b18f0274
stabletailgraph: implement stable-tail sort
This adds the computation of the "stable-tail sort", an incremental node
sorting method. It is a stepping stone for the implementation of faster
label discovery (for example for obs markers) and more caching.
author | pacien <pacien.trangirard@pacien.net> |
---|---|
date | Thu, 30 Mar 2023 22:22:44 +0200 |
parents | |
children | e1496403b08c |
rev | line source |
---|---|
50460
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
1 # stabletailsort.py - stable ordering of revisions |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
2 # |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
3 # Copyright 2021-2023 Pacien TRAN-GIRARD <pacien.trangirard@pacien.net> |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
4 # |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
5 # This software may be used and distributed according to the terms of the |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
6 # GNU General Public License version 2 or any later version. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
7 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
8 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
9 Stable-tail sort computation. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
10 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
11 The "stable-tail sort", or STS, is a reverse topological ordering of the |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
12 ancestors of a node, which tends to share large suffixes with the stable-tail |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
13 sort of ancestors and other nodes, giving it its name. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
14 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
15 Its properties should make it suitable for making chunks of ancestors with high |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
16 reuse and incrementality for example. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
17 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
18 This module and implementation are experimental. Most functions are not yet |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
19 optimised to operate on large production graphs. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
20 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
21 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
22 import itertools |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
23 from ..node import nullrev |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
24 from .. import ancestor |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
25 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
26 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
27 def _sorted_parents(cl, p1, p2): |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
28 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
29 Chooses and returns the pair (px, pt) from (p1, p2). |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
30 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
31 Where |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
32 "px" denotes the parent starting the "exclusive" part, and |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
33 "pt" denotes the parent starting the "Tail" part. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
34 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
35 "px" is chosen as the parent with the lowest rank with the goal of |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
36 minimising the size of the exclusive part and maximise the size of the |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
37 tail part, hopefully reducing the overall complexity of the stable sort. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
38 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
39 In case of equal ranks, the stable node ID is used as a tie-breaker. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
40 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
41 r1, r2 = cl.fast_rank(p1), cl.fast_rank(p2) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
42 if r1 < r2: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
43 return (p1, p2) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
44 elif r1 > r2: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
45 return (p2, p1) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
46 elif cl.node(p1) < cl.node(p2): |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
47 return (p1, p2) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
48 else: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
49 return (p2, p1) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
50 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
51 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
52 def _nonoedipal_parent_revs(cl, rev): |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
53 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
54 Returns the non-œdipal parent pair of the given revision. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
55 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
56 An œdipal merge is a merge with parents p1, p2 with either |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
57 p1 in ancestors(p2) or p2 in ancestors(p1). |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
58 In the first case, p1 is the œdipal parent. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
59 In the second case, p2 is the œdipal parent. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
60 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
61 Œdipal edges start empty exclusive parts. They do not bring new ancestors. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
62 As such, they can be skipped when computing any topological sort or any |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
63 iteration over the ancestors of a node. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
64 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
65 The œdipal edges are eliminated here using the rank information. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
66 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
67 p1, p2 = cl.parentrevs(rev) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
68 if p1 == nullrev or cl.fast_rank(p2) == cl.fast_rank(rev) - 1: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
69 return p2, nullrev |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
70 elif p2 == nullrev or cl.fast_rank(p1) == cl.fast_rank(rev) - 1: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
71 return p1, nullrev |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
72 else: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
73 return p1, p2 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
74 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
75 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
76 def _stable_tail_sort(cl, head_rev): |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
77 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
78 Naive topological iterator of the ancestors given by the stable-tail sort. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
79 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
80 The stable-tail sort of a node "h" is defined as the sequence: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
81 sts(h) := [h] + excl(h) + sts(pt(h)) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
82 where excl(h) := u for u in sts(px(h)) if u not in ancestors(pt(h)) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
83 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
84 This implementation uses a call-stack whose size is |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
85 O(number of open merges). |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
86 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
87 As such, this implementation exists mainly as a defining reference. |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
88 """ |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
89 cursor_rev = head_rev |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
90 while cursor_rev != nullrev: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
91 yield cursor_rev |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
92 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
93 p1, p2 = _nonoedipal_parent_revs(cl, cursor_rev) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
94 if p1 == nullrev: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
95 cursor_rev = p2 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
96 elif p2 == nullrev: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
97 cursor_rev = p1 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
98 else: |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
99 px, pt = _sorted_parents(cl, p1, p2) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
100 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
101 tail_ancestors = ancestor.lazyancestors( |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
102 cl.parentrevs, (pt,), inclusive=True |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
103 ) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
104 exclusive_ancestors = ( |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
105 a for a in _stable_tail_sort(cl, px) if a not in tail_ancestors |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
106 ) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
107 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
108 excl_part_size = cl.fast_rank(cursor_rev) - cl.fast_rank(pt) - 1 |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
109 yield from itertools.islice(exclusive_ancestors, excl_part_size) |
f0d2b18f0274
stabletailgraph: implement stable-tail sort
pacien <pacien.trangirard@pacien.net>
parents:
diff
changeset
|
110 cursor_rev = pt |