localrepo: experimental support for non-zlib revlog compression
The final part of integrating the compression manager APIs into
revlog storage is the plumbing for repositories to advertise they
are using non-zlib storage and for revlogs to instantiate a non-zlib
compression engine.
The main intent of the compression manager work was to zstd all
of the things. Adding zstd to revlogs has proved to be more involved
than other places because revlogs are... special. Very small inputs
and the use of delta chains (which are themselves a form of
compression) are a completely different use case from streaming
compression, which bundles and the wire protocol employ. I've
conducted numerous experiments with zstd in revlogs and have yet
to formalize compression settings and a storage architecture that
I'm confident I won't regret later. In other words, I'm not yet
ready to commit to a new mechanism for using zstd - or any other
compression format - in revlogs.
That being said, having some support for zstd (and other compression
formats) in revlogs in core is beneficial. It can allow others to
conduct experiments.
This patch introduces *highly experimental* support for non-zlib
compression formats in revlogs. Introduced is a config option to
control which compression engine to use. Also introduced is a namespace
of "exp-compression-*" requirements to denote support for non-zlib
compression in revlogs. I've prefixed the namespace with "exp-"
(short for "experimental") because I'm not confident of the
requirements "schema" and in no way want to give the illusion of
supporting these requirements in the future. I fully intend to drop
support for these requirements once we figure out what we're doing
with zstd in revlogs.
A good portion of the patch is teaching the requirements system
about registered compression engines and passing the requested
compression engine as an opener option so revlogs can instantiate
the proper compression engine for new operations.
That's a verbose way of saying "we can now use zstd in revlogs!"
On an `hg pull` conversion of the mozilla-unified repo with no extra
redelta settings (like aggressivemergedeltas), we can see the impact
of zstd vs zlib in revlogs:
$ hg perfrevlogchunks -c
! chunk
! wall 2.032052 comb 2.040000 user 1.990000 sys 0.050000 (best of 5)
! wall 1.866360 comb 1.860000 user 1.820000 sys 0.040000 (best of 6)
! chunk batch
! wall 1.877261 comb 1.870000 user 1.860000 sys 0.010000 (best of 6)
! wall 1.705410 comb 1.710000 user 1.690000 sys 0.020000 (best of 6)
$ hg perfrevlogchunks -m
! chunk
! wall 2.721427 comb 2.720000 user 2.640000 sys 0.080000 (best of 4)
! wall 2.035076 comb 2.030000 user 1.950000 sys 0.080000 (best of 5)
! chunk batch
! wall 2.614561 comb 2.620000 user 2.580000 sys 0.040000 (best of 4)
! wall 1.910252 comb 1.910000 user 1.880000 sys 0.030000 (best of 6)
$ hg perfrevlog -c -d 1
! wall 4.812885 comb 4.820000 user 4.800000 sys 0.020000 (best of 3)
! wall 4.699621 comb 4.710000 user 4.700000 sys 0.010000 (best of 3)
$ hg perfrevlog -m -d 1000
! wall 34.252800 comb 34.250000 user 33.730000 sys 0.520000 (best of 3)
! wall 24.094999 comb 24.090000 user 23.320000 sys 0.770000 (best of 3)
Only modest wins for the changelog. But manifest reading is
significantly faster. What's going on?
One reason might be data volume. zstd decompresses faster. So given
more bytes, it will put more distance between it and zlib.
Another reason is size. In the current design, zstd revlogs are
*larger*:
debugcreatestreamclonebundle (size in bytes)
zlib: 1,638,852,492
zstd: 1,680,601,332
I haven't investigated this fully, but I reckon a significant cause of
larger revlogs is that the zstd frame/header has more bytes than
zlib's. For very small inputs or data that doesn't compress well, we'll
tend to store more uncompressed chunks than with zlib (because the
compressed size isn't smaller than original). This will make revlog
reading faster because it is doing less decompression.
Moving on to bundle performance:
$ hg bundle -a -t none-v2 (total CPU time)
zlib: 102.79s
zstd: 97.75s
So, marginal CPU decrease for reading all chunks in all revlogs
(this is somewhat disappointing).
$ hg bundle -a -t <engine>-v2 (total CPU time)
zlib: 191.59s
zstd: 115.36s
This last test effectively measures the difference between zlib->zlib
and zstd->zstd for revlogs to bundle. This is a rough approximation of
what a server does during `hg clone`.
There are some promising results for zstd. But not enough for me to
feel comfortable advertising it to users. We'll get there...
Revert interactive tests
1 add and commit file f
2 add commit file folder1/g
3 add and commit file folder2/h
4 add and commit file folder1/i
5 commit change to file f
6 commit changes to files folder1/g folder2/h
7 commit changes to files folder1/g folder2/h
8 revert interactive to commit id 2 (line 3 above), check that folder1/i is removed and
9 make workdir match 7
10 run the same test than 8 from within folder1 and check same expectations
$ cat <<EOF >> $HGRCPATH
> [ui]
> interactive = true
> [extensions]
> record =
> purge =
> EOF
$ mkdir -p a/folder1 a/folder2
$ cd a
$ hg init
>>> open('f', 'wb').write("1\n2\n3\n4\n5\n")
$ hg add f ; hg commit -m "adding f"
$ cat f > folder1/g ; hg add folder1/g ; hg commit -m "adding folder1/g"
$ cat f > folder2/h ; hg add folder2/h ; hg commit -m "adding folder2/h"
$ cat f > folder1/i ; hg add folder1/i ; hg commit -m "adding folder1/i"
>>> open('f', 'wb').write("a\n1\n2\n3\n4\n5\nb\n")
$ hg commit -m "modifying f"
>>> open('folder1/g', 'wb').write("c\n1\n2\n3\n4\n5\nd\n")
$ hg commit -m "modifying folder1/g"
>>> open('folder2/h', 'wb').write("e\n1\n2\n3\n4\n5\nf\n")
$ hg commit -m "modifying folder2/h"
$ hg tip
changeset: 6:59dd6e4ab63a
tag: tip
user: test
date: Thu Jan 01 00:00:00 1970 +0000
summary: modifying folder2/h
$ hg revert -i -r 2 --all -- << EOF
> y
> y
> y
> y
> y
> y
> n
> n
> EOF
reverting f
reverting folder1/g (glob)
removing folder1/i (glob)
reverting folder2/h (glob)
remove added file folder1/i (Yn)? y
diff --git a/f b/f
2 hunks, 2 lines changed
examine changes to 'f'? [Ynesfdaq?] y
@@ -1,5 +1,6 @@
+a
1
2
3
4
5
revert change 1/6 to 'f'? [Ynesfdaq?] y
@@ -1,5 +2,6 @@
1
2
3
4
5
+b
revert change 2/6 to 'f'? [Ynesfdaq?] y
diff --git a/folder1/g b/folder1/g
2 hunks, 2 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] y
@@ -1,5 +1,6 @@
+c
1
2
3
4
5
revert change 3/6 to 'folder1/g'? [Ynesfdaq?] y
@@ -1,5 +2,6 @@
1
2
3
4
5
+d
revert change 4/6 to 'folder1/g'? [Ynesfdaq?] n
diff --git a/folder2/h b/folder2/h
2 hunks, 2 lines changed
examine changes to 'folder2/h'? [Ynesfdaq?] n
$ cat f
1
2
3
4
5
$ cat folder1/g
1
2
3
4
5
d
$ cat folder2/h
e
1
2
3
4
5
f
Test that --interactive lift the need for --all
$ echo q | hg revert -i -r 2
reverting folder1/g (glob)
reverting folder2/h (glob)
diff --git a/folder1/g b/folder1/g
1 hunks, 1 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] q
abort: user quit
[255]
$ ls folder1/
g
Test that a noop revert doesn't do an unnecessary backup
$ (echo y; echo n) | hg revert -i -r 2 folder1/g
diff --git a/folder1/g b/folder1/g
1 hunks, 1 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] y
@@ -3,3 +3,4 @@
3
4
5
+d
revert this change to 'folder1/g'? [Ynesfdaq?] n
$ ls folder1/
g
Test --no-backup
$ (echo y; echo y) | hg revert -i -C -r 2 folder1/g
diff --git a/folder1/g b/folder1/g
1 hunks, 1 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] y
@@ -3,3 +3,4 @@
3
4
5
+d
revert this change to 'folder1/g'? [Ynesfdaq?] y
$ ls folder1/
g
>>> open('folder1/g', 'wb').write("1\n2\n3\n4\n5\nd\n")
$ hg update -C 6
3 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg revert -i -r 2 --all -- << EOF
> n
> y
> y
> y
> y
> y
> n
> n
> EOF
reverting f
reverting folder1/g (glob)
removing folder1/i (glob)
reverting folder2/h (glob)
remove added file folder1/i (Yn)? n
diff --git a/f b/f
2 hunks, 2 lines changed
examine changes to 'f'? [Ynesfdaq?] y
@@ -1,5 +1,6 @@
+a
1
2
3
4
5
revert change 1/6 to 'f'? [Ynesfdaq?] y
@@ -1,5 +2,6 @@
1
2
3
4
5
+b
revert change 2/6 to 'f'? [Ynesfdaq?] y
diff --git a/folder1/g b/folder1/g
2 hunks, 2 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] y
@@ -1,5 +1,6 @@
+c
1
2
3
4
5
revert change 3/6 to 'folder1/g'? [Ynesfdaq?] y
@@ -1,5 +2,6 @@
1
2
3
4
5
+d
revert change 4/6 to 'folder1/g'? [Ynesfdaq?] n
diff --git a/folder2/h b/folder2/h
2 hunks, 2 lines changed
examine changes to 'folder2/h'? [Ynesfdaq?] n
$ cat f
1
2
3
4
5
$ cat folder1/g
1
2
3
4
5
d
$ cat folder2/h
e
1
2
3
4
5
f
$ hg st
M f
M folder1/g
$ hg revert --interactive f << EOF
> y
> y
> n
> n
> EOF
diff --git a/f b/f
2 hunks, 2 lines changed
examine changes to 'f'? [Ynesfdaq?] y
@@ -1,6 +1,5 @@
-a
1
2
3
4
5
discard change 1/2 to 'f'? [Ynesfdaq?] y
@@ -2,6 +1,5 @@
1
2
3
4
5
-b
discard change 2/2 to 'f'? [Ynesfdaq?] n
$ hg st
M f
M folder1/g
? f.orig
$ cat f
a
1
2
3
4
5
$ cat f.orig
1
2
3
4
5
$ rm f.orig
$ hg update -C .
2 files updated, 0 files merged, 0 files removed, 0 files unresolved
Check editing files newly added by a revert
1) Create a dummy editor changing 1 to 42
$ cat > $TESTTMP/editor.sh << '__EOF__'
> cat "$1" | sed "s/1/42/g" > tt
> mv tt "$1"
> __EOF__
2) Add k
$ printf "1\n" > k
$ hg add k
$ hg commit -m "add k"
3) Use interactive revert with editing (replacing +1 with +42):
$ printf "0\n2\n" > k
$ HGEDITOR="\"sh\" \"${TESTTMP}/editor.sh\"" hg revert -i <<EOF
> y
> e
> EOF
reverting k
diff --git a/k b/k
1 hunks, 2 lines changed
examine changes to 'k'? [Ynesfdaq?] y
@@ -1,1 +1,2 @@
-1
+0
+2
discard this change to 'k'? [Ynesfdaq?] e
$ cat k
42
Check the experimental config to invert the selection:
$ cat <<EOF >> $HGRCPATH
> [experimental]
> revertalternateinteractivemode=False
> EOF
$ hg up -C .
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ printf 'firstline\nc\n1\n2\n3\n 3\n5\nd\nlastline\n' > folder1/g
$ hg diff --nodates
diff -r a3d963a027aa folder1/g
--- a/folder1/g
+++ b/folder1/g
@@ -1,7 +1,9 @@
+firstline
c
1
2
3
-4
+ 3
5
d
+lastline
$ hg revert -i <<EOF
> y
> y
> y
> n
> EOF
reverting folder1/g (glob)
diff --git a/folder1/g b/folder1/g
3 hunks, 3 lines changed
examine changes to 'folder1/g'? [Ynesfdaq?] y
@@ -1,5 +1,4 @@
-firstline
c
1
2
3
discard change 1/3 to 'folder1/g'? [Ynesfdaq?] y
@@ -2,7 +1,7 @@
c
1
2
3
- 3
+4
5
d
discard change 2/3 to 'folder1/g'? [Ynesfdaq?] y
@@ -7,3 +6,2 @@
5
d
-lastline
discard change 3/3 to 'folder1/g'? [Ynesfdaq?] n
$ hg diff --nodates
diff -r a3d963a027aa folder1/g
--- a/folder1/g
+++ b/folder1/g
@@ -5,3 +5,4 @@
4
5
d
+lastline
$ hg update -C .
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg purge
$ touch newfile
$ hg add newfile
$ hg status
A newfile
$ hg revert -i <<EOF
> n
> EOF
forgetting newfile
forget added file newfile (Yn)? n
$ hg status
A newfile
$ hg revert -i <<EOF
> y
> EOF
forgetting newfile
forget added file newfile (Yn)? y
$ hg status
? newfile