Mercurial > hg
annotate mercurial/help/internals/revlogs.txt @ 41266:92cf293fd93d
remotefilelog: import Queue on Python 2, and queue on Python 3
Differential Revision: https://phab.mercurial-scm.org/D5599
author | Augie Fackler <augie@google.com> |
---|---|
date | Wed, 16 Jan 2019 10:56:39 -0500 |
parents | e7a2cc84dbc0 |
children | bfd65b5e070b |
rev | line source |
---|---|
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
1 Revision logs - or *revlogs* - are an append only data structure for |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
2 storing discrete entries, or *revisions*. They are the primary storage |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
3 mechanism of repository data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
4 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
5 Revlogs effectively model a directed acyclic graph (DAG). Each node |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
6 has edges to 1 or 2 *parent* nodes. Each node contains metadata and |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
7 the raw value for that node. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
8 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
9 Revlogs consist of entries which have metadata and revision data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
10 Metadata includes the hash of the revision's content, sizes, and |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
11 links to its *parent* entries. The collective metadata is referred |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
12 to as the *index* and the revision data is the *data*. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
13 |
41199
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
14 Revision data is stored as a series of compressed deltas against |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
15 ancestor revisions. |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
16 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
17 Revlogs are written in an append-only fashion. We never need to rewrite |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
18 a file to insert nor do we need to remove data. Rolling back in-progress |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
19 writes can be performed by truncating files. Read locks can be avoided |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
20 using simple techniques. This means that references to other data in |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
21 the same revlog *always* refer to a previous entry. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
22 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
23 Revlogs can be modeled as 0-indexed arrays. The first revision is |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
24 revision #0 and the second is revision #1. The revision -1 is typically |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
25 used to mean *does not exist* or *not defined*. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
26 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
27 File Format |
29747
aba2bb2a6d0f
help: don't try to render a section on sub-topics
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29094
diff
changeset
|
28 =========== |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
29 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
30 A revlog begins with a 32-bit big endian integer holding version info |
28590
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
31 and feature flags. This integer is shared with the first revision |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
32 entry. |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
33 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
34 This integer is logically divided into 2 16-bit shorts. The least |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
35 significant half of the integer is the format/version short. The other |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
36 short holds feature flags that dictate behavior of the revlog. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
37 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
38 The following values for the format/version short are defined: |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
39 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
40 0 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
41 The original revlog version. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
42 1 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
43 RevlogNG (*next generation*). It replaced version 0 when it was |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
44 implemented in 2006. |
32697
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
45 2 |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
46 In-development version incorporating accumulated knowledge and |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
47 missing features from 10+ years of revlog version 1. |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
48 57005 (0xdead) |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
49 Reserved for internal testing of new versions. No defined format |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
50 beyond 32-bit header. |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
51 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
52 The feature flags short consists of bit flags. Where 0 is the least |
41199
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
53 significant bit. The bit flags vary by revlog version. |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
54 |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
55 Version 0 revlogs have no defined flags and the presence of a flag |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
56 is considered an error. |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
57 |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
58 Version 1 revlogs have the following flags at the specified bit offsets: |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
59 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
60 0 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
61 Store revision data inline. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
62 1 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
63 Generaldelta encoding. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
64 |
41199
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
65 Version 2 revlogs have the following flags at the specified bit offsets: |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
66 |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
67 0 |
d8fe67db5234
internals: minor rewriting of revlogs documentation
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32697
diff
changeset
|
68 Store revision data inline. |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
69 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
70 The following header values are common: |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
71 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
72 00 00 00 01 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
73 v1 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
74 00 01 00 01 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
75 v1 + inline |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
76 00 02 00 01 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
77 v1 + generaldelta |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
78 00 03 00 01 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
79 v1 + inline + generaldelta |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
80 |
28590
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
81 Following the 32-bit header is the remainder of the first index entry. |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
82 Following that are remaining *index* data. Inlined revision data is |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
83 possibly located between index entries. More on this layout is described |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
84 below. |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
85 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
86 Version 1 Format |
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
87 ================ |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
88 |
32393
d47b62368f3a
revlog: remove some revlogNG terminology
Gregory Szorc <gregory.szorc@gmail.com>
parents:
31214
diff
changeset
|
89 Version 1 (RevlogNG) begins with an index describing the revisions in |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
90 the revlog. If the ``inline`` flag is set, revision data is stored inline, |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
91 or between index entries (as opposed to in a separate container). |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
92 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
93 Each index entry is 64 bytes. The byte layout of each entry is as |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
94 follows, with byte 0 being the first byte (all data stored as big endian): |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
95 |
28590
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
96 0-3 (4 bytes) (rev 0 only) |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
97 Revlog header |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
98 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
99 0-5 (6 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
100 Absolute offset of revision data from beginning of revlog. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
101 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
102 6-7 (2 bytes) |
30523
726d30a6d89b
censor: flag internal documentation
Remi Chaintron <remi@fb.com>
parents:
30499
diff
changeset
|
103 Bit flags impacting revision behavior. The following bit offsets define: |
30828
0b792507ea15
help: don't let tools reflow revlog flags list
Martin von Zweigbergk <martinvonz@google.com>
parents:
30827
diff
changeset
|
104 |
30658
c49be208ae34
documentation: better censor flag documentation
Remi Chaintron <remi@fb.com>
parents:
30523
diff
changeset
|
105 0: REVIDX_ISCENSORED revision has censor metadata, must be verified. |
30828
0b792507ea15
help: don't let tools reflow revlog flags list
Martin von Zweigbergk <martinvonz@google.com>
parents:
30827
diff
changeset
|
106 |
30829
08b34c3a6f74
revlog: give EXTSTORED flag value to narrowhg
Martin von Zweigbergk <martinvonz@google.com>
parents:
30828
diff
changeset
|
107 1: REVIDX_ELLIPSIS revision hash does not match its data. Used by |
08b34c3a6f74
revlog: give EXTSTORED flag value to narrowhg
Martin von Zweigbergk <martinvonz@google.com>
parents:
30828
diff
changeset
|
108 narrowhg |
08b34c3a6f74
revlog: give EXTSTORED flag value to narrowhg
Martin von Zweigbergk <martinvonz@google.com>
parents:
30828
diff
changeset
|
109 |
08b34c3a6f74
revlog: give EXTSTORED flag value to narrowhg
Martin von Zweigbergk <martinvonz@google.com>
parents:
30828
diff
changeset
|
110 2: REVIDX_EXTSTORED revision data is stored externally. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
111 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
112 8-11 (4 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
113 Compressed length of revision data / chunk as stored in revlog. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
114 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
115 12-15 (4 bytes) |
30499
22d05b53b0e8
help: clarify contents of revlog index
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29747
diff
changeset
|
116 Uncompressed length of revision data. This is the size of the full |
22d05b53b0e8
help: clarify contents of revlog index
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29747
diff
changeset
|
117 revision data, not the size of the chunk post decompression. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
118 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
119 16-19 (4 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
120 Base or previous revision this revision's delta was produced against. |
31214
167b69ccc62c
help: align description of 'base rev' with reality [issue5488]
Kim Alvefur <zash@zash.se>
parents:
30829
diff
changeset
|
121 This revision holds full text (as opposed to a delta) if it points to |
167b69ccc62c
help: align description of 'base rev' with reality [issue5488]
Kim Alvefur <zash@zash.se>
parents:
30829
diff
changeset
|
122 itself. For generaldelta repos, this is the previous revision in the |
167b69ccc62c
help: align description of 'base rev' with reality [issue5488]
Kim Alvefur <zash@zash.se>
parents:
30829
diff
changeset
|
123 delta chain. For non-generaldelta repos, this is the base or first |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
124 revision in the delta chain. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
125 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
126 20-23 (4 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
127 A revision this revision is *linked* to. This allows a revision in |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
128 one revlog to be forever associated with a revision in another |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
129 revlog. For example, a file's revlog may point to the changelog |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
130 revision that introduced it. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
131 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
132 24-27 (4 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
133 Revision of 1st parent. -1 indicates no parent. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
134 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
135 28-31 (4 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
136 Revision of 2nd parent. -1 indicates no 2nd parent. |
30827
e997e4826459
help: format revlog.txt more closely to result
Martin von Zweigbergk <martinvonz@google.com>
parents:
30746
diff
changeset
|
137 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
138 32-63 (32 bytes) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
139 Hash of revision's full text. Currently, SHA-1 is used and only |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
140 the first 20 bytes of this field are used. The rest of the bytes |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
141 are ignored and should be stored as \0. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
142 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
143 If inline revision data is being stored, the compressed revision data |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
144 (of length from bytes offset 8-11 from the index entry) immediately |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
145 follows the index entry. There is no header on the revision data. There |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
146 is no padding between it and the index entries before and after. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
147 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
148 If revision data is not inline, then raw revision data is stored in a |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
149 separate byte container. The offsets from bytes 0-5 and the compressed |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
150 length from bytes 8-11 define how to access this data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
151 |
28590
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
152 The first 4 bytes of the revlog are shared between the revlog header |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
153 and the 6 byte absolute offset field from the first revlog entry. |
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
154 |
32697
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
155 Version 2 Format |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
156 ================ |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
157 |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
158 (In development. Format not finalized or stable.) |
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
159 |
41202
e7a2cc84dbc0
revlog: always enable generaldelta on version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
41199
diff
changeset
|
160 Version 2 is identical to version 2 with the following differences. |
e7a2cc84dbc0
revlog: always enable generaldelta on version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
41199
diff
changeset
|
161 |
e7a2cc84dbc0
revlog: always enable generaldelta on version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
41199
diff
changeset
|
162 There is no dedicated *generaldelta* revlog format flag. Instead, |
e7a2cc84dbc0
revlog: always enable generaldelta on version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
41199
diff
changeset
|
163 the feature is implied enabled by default. |
32697
19b9fc40cc51
revlog: skeleton support for version 2 revlogs
Gregory Szorc <gregory.szorc@gmail.com>
parents:
32393
diff
changeset
|
164 |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
165 Delta Chains |
29747
aba2bb2a6d0f
help: don't try to render a section on sub-topics
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29094
diff
changeset
|
166 ============ |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
167 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
168 Revision data is encoded as a chain of *chunks*. Each chain begins with |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
169 the compressed original full text for that revision. Each subsequent |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
170 *chunk* is a *delta* against the previous revision. We therefore call |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
171 these chains of chunks/deltas *delta chains*. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
172 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
173 The full text for a revision is reconstructed by loading the original |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
174 full text for the base revision of a *delta chain* and then applying |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
175 *deltas* until the target revision is reconstructed. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
176 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
177 *Delta chains* are limited in length so lookup time is bound. They are |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
178 limited to ~2x the length of the revision's data. The linear distance |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
179 between the base chunk and the final chunk is also limited so the |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
180 amount of read I/O to load all chunks in the delta chain is bound. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
181 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
182 Deltas and delta chains are either computed against the previous |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
183 revision in the revlog or another revision (almost certainly one of |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
184 the parents of the revision). Historically, deltas were computed against |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
185 the previous revision. The *generaldelta* revlog feature flag (enabled |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
186 by default in Mercurial 3.7) activates the mode where deltas are |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
187 computed against an arbitrary revision (almost certainly a parent revision). |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
188 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
189 File Storage |
29747
aba2bb2a6d0f
help: don't try to render a section on sub-topics
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29094
diff
changeset
|
190 ============ |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
191 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
192 Revlogs logically consist of an index (metadata of entries) and |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
193 revision data. This data may be stored together in a single file or in |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
194 separate files. The mechanism used is indicated by the ``inline`` feature |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
195 flag on the revlog. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
196 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
197 Mercurial's behavior is to use inline storage until a revlog reaches a |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
198 certain size, at which point it will be converted to non-inline. The |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
199 reason there is a size limit on inline storage is to establish an upper |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
200 bound on how much data must be read to load the index. It would be a waste |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
201 to read tens or hundreds of extra megabytes of data just to access the |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
202 index data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
203 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
204 The actual layout of revlog files on disk is governed by the repository's |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
205 *store format*. Typically, a ``.i`` file represents the index revlog |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
206 (possibly containing inline data) and a ``.d`` file holds the revision data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
207 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
208 Revision Entries |
29747
aba2bb2a6d0f
help: don't try to render a section on sub-topics
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29094
diff
changeset
|
209 ================ |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
210 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
211 Revision entries consist of an optional 1 byte header followed by an |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
212 encoding of the revision data. The headers are as follows: |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
213 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
214 \0 (0x00) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
215 Revision data is the entirety of the entry, including this header. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
216 u (0x75) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
217 Raw revision data follows. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
218 x (0x78) |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
219 zlib (RFC 1950) data. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
220 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
221 The 0x78 value is actually the first byte of the zlib header (CMF byte). |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
222 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
223 Hash Computation |
29747
aba2bb2a6d0f
help: don't try to render a section on sub-topics
Gregory Szorc <gregory.szorc@gmail.com>
parents:
29094
diff
changeset
|
224 ================ |
27631
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
225 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
226 The hash of the revision is stored in the index and is used both as a primary |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
227 key and for data integrity verification. |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
228 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
229 Currently, SHA-1 is the only supported hashing algorithm. To obtain the SHA-1 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
230 hash of a revision: |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
231 |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
232 1. Hash the parent nodes |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
233 2. Hash the fulltext of the revision |
c18292a6ff54
internals: document revlog format
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
234 |
28590
b0b9f6b0a777
help: document sharing of revlog header with revision 0
Gregory Szorc <gregory.szorc@gmail.com>
parents:
27631
diff
changeset
|
235 The 20 byte node ids of the parents are fed into the hasher in ascending order. |