Mercurial > hg
comparison tests/test-revlog-delta-find.t @ 49764:e300f445ca77
delta-find: add a test checking various simple behavior
There are enough work happening in this area that it is worth having a dedicated
test for it. So far we add to small test checking that the "best" parent is
picked as the delta base and that this behavior can be controlled during commit
and unbundle.
author | Pierre-Yves David <pierre-yves.david@octobus.net> |
---|---|
date | Mon, 05 Dec 2022 03:23:46 +0100 |
parents | |
children | f1887500f3ec |
comparison
equal
deleted
inserted
replaced
49762:c4d587fa161c | 49764:e300f445ca77 |
---|---|
1 ========================================================== | |
2 Test various things around delta computation within revlog | |
3 ========================================================== | |
4 | |
5 | |
6 basic setup | |
7 ----------- | |
8 | |
9 $ cat << EOF >> $HGRCPATH | |
10 > [debug] | |
11 > revlog.debug-delta=yes | |
12 > EOF | |
13 $ cat << EOF >> sha256line.py | |
14 > # a way to quickly produce file of significant size and poorly compressable content. | |
15 > import hashlib | |
16 > import sys | |
17 > for line in sys.stdin: | |
18 > print(hashlib.sha256(line.encode('utf8')).hexdigest()) | |
19 > EOF | |
20 | |
21 $ hg init base-repo | |
22 $ cd base-repo | |
23 | |
24 create a "large" file | |
25 | |
26 $ $TESTDIR/seq.py 1000 | $PYTHON $TESTTMP/sha256line.py > my-file.txt | |
27 $ hg add my-file.txt | |
28 $ hg commit -m initial-commit | |
29 DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob) | |
30 DBG-DELTAS: MANIFESTLOG: * (glob) | |
31 DBG-DELTAS: CHANGELOG: * (glob) | |
32 | |
33 Add more change at the end of the file | |
34 | |
35 $ $TESTDIR/seq.py 1001 1200 | $PYTHON $TESTTMP/sha256line.py >> my-file.txt | |
36 $ hg commit -m "large-change" | |
37 DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob) | |
38 DBG-DELTAS: MANIFESTLOG: * (glob) | |
39 DBG-DELTAS: CHANGELOG: * (glob) | |
40 | |
41 Add small change at the start | |
42 | |
43 $ hg up 'desc("initial-commit")' --quiet | |
44 $ mv my-file.txt foo | |
45 $ echo "small change at the start" > my-file.txt | |
46 $ cat foo >> my-file.txt | |
47 $ rm foo | |
48 $ hg commit -m "small-change" | |
49 DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob) | |
50 DBG-DELTAS: MANIFESTLOG: * (glob) | |
51 DBG-DELTAS: CHANGELOG: * (glob) | |
52 created new head | |
53 | |
54 | |
55 $ hg log -r 'head()' -T '{node}\n' >> ../base-heads.nodes | |
56 $ hg log -r 'desc("initial-commit")' -T '{node}\n' >> ../initial.node | |
57 $ hg log -r 'desc("small-change")' -T '{node}\n' >> ../small.node | |
58 $ hg log -r 'desc("large-change")' -T '{node}\n' >> ../large.node | |
59 $ cd .. | |
60 | |
61 Check delta find policy and result for merge on commit | |
62 ====================================================== | |
63 | |
64 Check that delta of merge pick best of the two parents | |
65 ------------------------------------------------------ | |
66 | |
67 As we check against both parents, the one with the largest change should | |
68 produce the smallest delta and be picked. | |
69 | |
70 $ hg clone base-repo test-parents --quiet | |
71 $ hg -R test-parents update 'nodefromfile("small.node")' --quiet | |
72 $ hg -R test-parents merge 'nodefromfile("large.node")' --quiet | |
73 | |
74 The delta base is the "large" revision as it produce a smaller delta. | |
75 | |
76 $ hg -R test-parents commit -m "merge from small change" | |
77 DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=1 * (glob) | |
78 DBG-DELTAS: MANIFESTLOG: * (glob) | |
79 DBG-DELTAS: CHANGELOG: * (glob) | |
80 | |
81 Check that the behavior tested above can we disabled | |
82 ---------------------------------------------------- | |
83 | |
84 We disable the checking of both parent at the same time. The `small` change, | |
85 that produce a less optimal delta, should be picked first as it is "closer" to | |
86 the new commit. | |
87 | |
88 $ hg clone base-repo test-no-parents --quiet | |
89 $ hg -R test-no-parents update 'nodefromfile("small.node")' --quiet | |
90 $ hg -R test-no-parents merge 'nodefromfile("large.node")' --quiet | |
91 | |
92 The delta base is the "large" revision as it produce a smaller delta. | |
93 | |
94 $ hg -R test-no-parents commit -m "merge from small change" \ | |
95 > --config storage.revlog.optimize-delta-parent-choice=no | |
96 DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob) | |
97 DBG-DELTAS: MANIFESTLOG: * (glob) | |
98 DBG-DELTAS: CHANGELOG: * (glob) | |
99 | |
100 | |
101 Check delta-find policy and result when unbundling | |
102 ================================================== | |
103 | |
104 Build a bundle with all delta built against p1 | |
105 | |
106 $ hg bundle -R test-parents --all --config devel.bundle.delta=p1 all-p1.hg | |
107 4 changesets found | |
108 | |
109 Default policy of trusting delta from the bundle | |
110 ------------------------------------------------ | |
111 | |
112 Keeping the `p1` delta used in the bundle is sub-optimal for storage, but | |
113 strusting in-bundle delta is faster to apply. | |
114 | |
115 $ hg init bundle-default | |
116 $ hg -R bundle-default unbundle all-p1.hg --quiet | |
117 DBG-DELTAS: CHANGELOG: * (glob) | |
118 DBG-DELTAS: CHANGELOG: * (glob) | |
119 DBG-DELTAS: CHANGELOG: * (glob) | |
120 DBG-DELTAS: CHANGELOG: * (glob) | |
121 DBG-DELTAS: MANIFESTLOG: * (glob) | |
122 DBG-DELTAS: MANIFESTLOG: * (glob) | |
123 DBG-DELTAS: MANIFESTLOG: * (glob) | |
124 DBG-DELTAS: MANIFESTLOG: * (glob) | |
125 DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob) | |
126 DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob) | |
127 DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob) | |
128 DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob) | |
129 | |
130 (confirm the file revision are in the same order, 2 should be smaller than 1) | |
131 | |
132 $ hg -R bundle-default debugdata my-file.txt 2 | wc -l | |
133 \s*1001 (re) | |
134 $ hg -R bundle-default debugdata my-file.txt 1 | wc -l | |
135 \s*1200 (re) | |
136 | |
137 explicitly enabled | |
138 ------------------ | |
139 | |
140 Keeping the `p1` delta used in the bundle is sub-optimal for storage, but | |
141 strusting in-bundle delta is faster to apply. | |
142 | |
143 $ hg init bundle-reuse-enabled | |
144 $ hg -R bundle-reuse-enabled unbundle all-p1.hg --quiet \ | |
145 > --config storage.revlog.reuse-external-delta-parent=yes | |
146 DBG-DELTAS: CHANGELOG: * (glob) | |
147 DBG-DELTAS: CHANGELOG: * (glob) | |
148 DBG-DELTAS: CHANGELOG: * (glob) | |
149 DBG-DELTAS: CHANGELOG: * (glob) | |
150 DBG-DELTAS: MANIFESTLOG: * (glob) | |
151 DBG-DELTAS: MANIFESTLOG: * (glob) | |
152 DBG-DELTAS: MANIFESTLOG: * (glob) | |
153 DBG-DELTAS: MANIFESTLOG: * (glob) | |
154 DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob) | |
155 DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob) | |
156 DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob) | |
157 DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob) | |
158 | |
159 (confirm the file revision are in the same order, 2 should be smaller than 1) | |
160 | |
161 $ hg -R bundle-reuse-enabled debugdata my-file.txt 2 | wc -l | |
162 \s*1001 (re) | |
163 $ hg -R bundle-reuse-enabled debugdata my-file.txt 1 | wc -l | |
164 \s*1200 (re) | |
165 | |
166 explicitly disabled | |
167 ------------------- | |
168 | |
169 Not reusing the delta-base from the parent means we the delta will be made | |
170 against the "best" parent. (so not the same as the previous two) | |
171 | |
172 $ hg init bundle-reuse-disabled | |
173 $ hg -R bundle-reuse-disabled unbundle all-p1.hg --quiet \ | |
174 > --config storage.revlog.reuse-external-delta-parent=no | |
175 DBG-DELTAS: CHANGELOG: * (glob) | |
176 DBG-DELTAS: CHANGELOG: * (glob) | |
177 DBG-DELTAS: CHANGELOG: * (glob) | |
178 DBG-DELTAS: CHANGELOG: * (glob) | |
179 DBG-DELTAS: MANIFESTLOG: * (glob) | |
180 DBG-DELTAS: MANIFESTLOG: * (glob) | |
181 DBG-DELTAS: MANIFESTLOG: * (glob) | |
182 DBG-DELTAS: MANIFESTLOG: * (glob) | |
183 DBG-DELTAS: FILELOG:my-file.txt: rev=0: delta-base=0 * (glob) | |
184 DBG-DELTAS: FILELOG:my-file.txt: rev=1: delta-base=0 * (glob) | |
185 DBG-DELTAS: FILELOG:my-file.txt: rev=2: delta-base=0 * (glob) | |
186 DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=1 * (glob) | |
187 | |
188 (confirm the file revision are in the same order, 2 should be smaller than 1) | |
189 | |
190 $ hg -R bundle-reuse-disabled debugdata my-file.txt 2 | wc -l | |
191 \s*1001 (re) | |
192 $ hg -R bundle-reuse-disabled debugdata my-file.txt 1 | wc -l | |
193 \s*1200 (re) |