view tests/test-git-export.t @ 30155:b7a966ce89ed

changelog: disable delta chains

This patch disables delta chains on changelogs. After this patch, new
entries on changelogs - including existing changelogs - will be stored
as the fulltext of that data (likely compressed). No delta computation
will be performed.

An overview of delta chains and data justifying this change follows.

Revlogs try to store entries as a delta against a previous entry (either
a parent revision in the case of generaldelta or the previous physical
revision when not using generaldelta). Most of the time this is the
correct thing to do: it frequently results in less CPU usage and smaller
storage.

Delta chains are most effective when the base revision being deltad
against is similar to the current data. This tends to occur naturally
for manifests and file data, since only small parts of each tend to
change with each revision. Changelogs, however, are a different story.

Changelog entries represent changesets/commits. And unless commits in a
repository are homogonous (same author, changing same files, similar
commit messages, etc), a delta from one entry to the next tends to be
relatively large compared to the size of the entry. This means that
delta chains tend to be short. How short? Here is the full vs delta
revision breakdown on some real world repos:

Repo             % Full    % Delta   Max Length
hg                45.8       54.2        6
mozilla-central   42.4       57.6        8
mozilla-unified   42.5       57.5       17
pypy              46.1       53.9        6
python-zstandard  46.1       53.9        3

(I threw in python-zstandard as an example of a repo that is homogonous.
It contains a small Python project with changes all from the same
author.)

Contrast this with the manifest revlog for these repos, where 99+% of
revisions are deltas and delta chains run into the thousands.

So delta chains aren't as useful on changelogs. But even a short delta
chain may provide benefits. Let's measure that.

Delta chains may require less CPU to read revisions if the CPU time
spent reading smaller deltas is less than the CPU time used to
decompress larger individual entries. We can measure this via
`hg perfrevlog -c -d 1` to iterate a revlog to resolve each revision's
fulltext. Here are the results of that command on a repo using delta
chains in its changelog and on a repo without delta chains:

hg (forward)
! wall 0.407008 comb 0.410000 user 0.410000 sys 0.000000 (best of 25)
! wall 0.390061 comb 0.390000 user 0.390000 sys 0.000000 (best of 26)

hg (reverse)
! wall 0.515221 comb 0.520000 user 0.520000 sys 0.000000 (best of 19)
! wall 0.400018 comb 0.400000 user 0.390000 sys 0.010000 (best of 25)

mozilla-central (forward)
! wall 4.508296 comb 4.490000 user 4.490000 sys 0.000000 (best of 3)
! wall 4.370222 comb 4.370000 user 4.350000 sys 0.020000 (best of 3)

mozilla-central (reverse)
! wall 5.758995 comb 5.760000 user 5.720000 sys 0.040000 (best of 3)
! wall 4.346503 comb 4.340000 user 4.320000 sys 0.020000 (best of 3)

mozilla-unified (forward)
! wall 4.957088 comb 4.950000 user 4.940000 sys 0.010000 (best of 3)
! wall 4.660528 comb 4.650000 user 4.630000 sys 0.020000 (best of 3)

mozilla-unified (reverse)
! wall 6.119827 comb 6.110000 user 6.090000 sys 0.020000 (best of 3)
! wall 4.675136 comb 4.670000 user 4.670000 sys 0.000000 (best of 3)

pypy (forward)
! wall 1.231122 comb 1.240000 user 1.230000 sys 0.010000 (best of 8)
! wall 1.164896 comb 1.160000 user 1.160000 sys 0.000000 (best of 9)

pypy (reverse)
! wall 1.467049 comb 1.460000 user 1.460000 sys 0.000000 (best of 7)
! wall 1.160200 comb 1.170000 user 1.160000 sys 0.010000 (best of 9)

The data clearly shows that it takes less wall and CPU time to resolve
revisions when there are no delta chains in the changelogs, regardless
of the direction of traversal. Furthermore, not using a delta chain
means that fulltext resolution in reverse is as fast as iterating
forward. So not using delta chains on the changelog is a clear CPU win
for reading operations.

An example of a user-visible operation showing this speed-up is revset
evaluation. Here are results for
`hg perfrevset 'author(gps) or author(mpm)'`:

hg
! wall 1.655506 comb 1.660000 user 1.650000 sys 0.010000 (best of 6)
! wall 1.612723 comb 1.610000 user 1.600000 sys 0.010000 (best of 7)

mozilla-central
! wall 17.629826 comb 17.640000 user 17.600000 sys 0.040000 (best of 3)
! wall 17.311033 comb 17.300000 user 17.260000 sys 0.040000 (best of 3)

What about 00changelog.i size?

Repo                Delta Chains     No Delta Chains
hg                    7,033,250         6,976,771
mozilla-central      82,978,748        81,574,623
mozilla-unified      88,112,349        86,702,162
pypy                 20,740,699        20,659,741

The data shows that removing delta chains from the changelog makes the
changelog smaller.

Delta chains are also used during changegroup generation. This
operation essentially converts a series of revisions to one large
delta chain. And changegroup generation is smart: if the delta in
the revlog matches what the changegroup is emitting, it will reuse
the delta instead of recalculating it. We can measure the impact
removing changelog delta chains has on changegroup generation via
`hg perfchangegroupchangelog`:

hg
! wall 1.589245 comb 1.590000 user 1.590000 sys 0.000000 (best of 7)
! wall 1.788060 comb 1.790000 user 1.790000 sys 0.000000 (best of 6)

mozilla-central
! wall 17.382585 comb 17.380000 user 17.340000 sys 0.040000 (best of 3)
! wall 20.161357 comb 20.160000 user 20.120000 sys 0.040000 (best of 3)

mozilla-unified
! wall 18.722839 comb 18.720000 user 18.680000 sys 0.040000 (best of 3)
! wall 21.168075 comb 21.170000 user 21.130000 sys 0.040000 (best of 3)

pypy
! wall 4.828317 comb 4.830000 user 4.820000 sys 0.010000 (best of 3)
! wall 5.415455 comb 5.420000 user 5.410000 sys 0.010000 (best of 3)

The data shows eliminating delta chains makes the changelog part of
changegroup generation slower. This is expected since we now have to
compute deltas for revisions where we could recycle the delta before.

It is worth putting this regression into context of overall changegroup
times. Here is the rough total CPU time spent in changegroup generation
for various repos while using delta chains on the changelog:

Repo              CPU Time (s)    CPU Time w/ compression
hg                  4.50              7.05
mozilla-central   111.1             222.0
pypy               28.68             75.5

Before compression, removing delta chains from the changegroup adds
~4.4% overhead to hg changegroup generation, 1.3% to mozilla-central,
and 2.0% to pypy. When you factor in zlib compression, these percentages
are roughly divided by 2.

While the increased CPU usage for changegroup generation is unfortunate,
I think it is acceptable because the percentage is small, server
operators (those likely impacted most by this) have other mechanisms
to mitigate CPU consumption (namely reducing zlib compression level and
pre-generated clone bundles), and because there is room to optimize this
in the future. For example, we could use the nullid as the base revision,
effectively encoding the full revision for each entry in the changegroup.
When doing this, `hg perfchangegroupchangelog` nearly halves:

mozilla-unified
! wall 21.168075 comb 21.170000 user 21.130000 sys 0.040000 (best of 3)
! wall 11.196461 comb 11.200000 user 11.190000 sys 0.010000 (best of 3)

This looks very promising as a future optimization opportunity.

It's worth that the changes in test-acl.t to the changegroup part size.
This is because revision 6 in the changegroup had a delta chain of
length 2 before and after this patch the base revision is nullrev.
When the base revision is nullrev, cg2packer.deltaparent() hardcodes
the *previous* revision from the changegroup as the delta parent.
This caused the delta in the changegroup to switch base revisions,
the delta to change, and the size to change accordingly. While the
size increased in this case, I think sizes will remain the same
on average, as the delta base for changelog revisions doesn't matter
too much (as this patch shows). So, I don't consider this a regression.

  $ hg init
  $ echo start > start
  $ hg ci -Amstart
  adding start

New file:

  $ mkdir dir1
  $ echo new > dir1/new
  $ hg ci -Amnew
  adding dir1/new
  $ hg diff --git -r 0
  diff --git a/dir1/new b/dir1/new
  new file mode 100644
  --- /dev/null
  +++ b/dir1/new
  @@ -0,0 +1,1 @@
  +new

Copy:

  $ mkdir dir2
  $ hg cp dir1/new dir1/copy
  $ echo copy1 >> dir1/copy
  $ hg cp dir1/new dir2/copy
  $ echo copy2 >> dir2/copy
  $ hg ci -mcopy
  $ hg diff --git -r 1:tip
  diff --git a/dir1/new b/dir1/copy
  copy from dir1/new
  copy to dir1/copy
  --- a/dir1/new
  +++ b/dir1/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  diff --git a/dir1/new b/dir2/copy
  copy from dir1/new
  copy to dir2/copy
  --- a/dir1/new
  +++ b/dir2/copy
  @@ -1,1 +1,2 @@
   new
  +copy2

Cross and same-directory copies with a relative root:

  $ hg diff --git --root .. -r 1:tip
  abort: .. not under root '$TESTTMP'
  [255]
  $ hg diff --git --root doesnotexist -r 1:tip
  $ hg diff --git --root . -r 1:tip
  diff --git a/dir1/new b/dir1/copy
  copy from dir1/new
  copy to dir1/copy
  --- a/dir1/new
  +++ b/dir1/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  diff --git a/dir1/new b/dir2/copy
  copy from dir1/new
  copy to dir2/copy
  --- a/dir1/new
  +++ b/dir2/copy
  @@ -1,1 +1,2 @@
   new
  +copy2
  $ hg diff --git --root dir1 -r 1:tip
  diff --git a/new b/copy
  copy from new
  copy to copy
  --- a/new
  +++ b/copy
  @@ -1,1 +1,2 @@
   new
  +copy1

  $ hg diff --git --root dir2/ -r 1:tip
  diff --git a/copy b/copy
  new file mode 100644
  --- /dev/null
  +++ b/copy
  @@ -0,0 +1,2 @@
  +new
  +copy2

  $ hg diff --git --root dir1 -r 1:tip -I '**/copy'
  diff --git a/new b/copy
  copy from new
  copy to copy
  --- a/new
  +++ b/copy
  @@ -1,1 +1,2 @@
   new
  +copy1

  $ hg diff --git --root dir1 -r 1:tip dir2
  warning: dir2 not inside relative root dir1

  $ hg diff --git --root dir1 -r 1:tip 'dir2/{copy}'
  warning: dir2/{copy} not inside relative root dir1 (glob)

  $ cd dir1
  $ hg diff --git --root .. -r 1:tip
  diff --git a/dir1/new b/dir1/copy
  copy from dir1/new
  copy to dir1/copy
  --- a/dir1/new
  +++ b/dir1/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  diff --git a/dir1/new b/dir2/copy
  copy from dir1/new
  copy to dir2/copy
  --- a/dir1/new
  +++ b/dir2/copy
  @@ -1,1 +1,2 @@
   new
  +copy2

  $ hg diff --git --root ../.. -r 1:tip
  abort: ../.. not under root '$TESTTMP'
  [255]
  $ hg diff --git --root ../doesnotexist -r 1:tip
  $ hg diff --git --root .. -r 1:tip
  diff --git a/dir1/new b/dir1/copy
  copy from dir1/new
  copy to dir1/copy
  --- a/dir1/new
  +++ b/dir1/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  diff --git a/dir1/new b/dir2/copy
  copy from dir1/new
  copy to dir2/copy
  --- a/dir1/new
  +++ b/dir2/copy
  @@ -1,1 +1,2 @@
   new
  +copy2

  $ hg diff --git --root . -r 1:tip
  diff --git a/new b/copy
  copy from new
  copy to copy
  --- a/new
  +++ b/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  $ hg diff --git --root . -r 1:tip copy
  diff --git a/new b/copy
  copy from new
  copy to copy
  --- a/new
  +++ b/copy
  @@ -1,1 +1,2 @@
   new
  +copy1
  $ hg diff --git --root . -r 1:tip ../dir2
  warning: ../dir2 not inside relative root . (glob)
  $ hg diff --git --root . -r 1:tip '../dir2/*'
  warning: ../dir2/* not inside relative root . (glob)
  $ cd ..

Rename:

  $ hg mv dir1/copy dir1/rename1
  $ echo rename1 >> dir1/rename1
  $ hg mv dir2/copy dir1/rename2
  $ echo rename2 >> dir1/rename2
  $ hg ci -mrename
  $ hg diff --git -r 2:tip
  diff --git a/dir1/copy b/dir1/rename1
  rename from dir1/copy
  rename to dir1/rename1
  --- a/dir1/copy
  +++ b/dir1/rename1
  @@ -1,2 +1,3 @@
   new
   copy1
  +rename1
  diff --git a/dir2/copy b/dir1/rename2
  rename from dir2/copy
  rename to dir1/rename2
  --- a/dir2/copy
  +++ b/dir1/rename2
  @@ -1,2 +1,3 @@
   new
   copy2
  +rename2

Cross and same-directory renames with a relative root:

  $ hg diff --root dir1 --git -r 2:tip
  diff --git a/copy b/rename1
  rename from copy
  rename to rename1
  --- a/copy
  +++ b/rename1
  @@ -1,2 +1,3 @@
   new
   copy1
  +rename1
  diff --git a/rename2 b/rename2
  new file mode 100644
  --- /dev/null
  +++ b/rename2
  @@ -0,0 +1,3 @@
  +new
  +copy2
  +rename2

  $ hg diff --root dir2 --git -r 2:tip
  diff --git a/copy b/copy
  deleted file mode 100644
  --- a/copy
  +++ /dev/null
  @@ -1,2 +0,0 @@
  -new
  -copy2

  $ hg diff --root dir1 --git -r 2:tip -I '**/copy'
  diff --git a/copy b/copy
  deleted file mode 100644
  --- a/copy
  +++ /dev/null
  @@ -1,2 +0,0 @@
  -new
  -copy1

  $ hg diff --root dir1 --git -r 2:tip -I '**/rename*'
  diff --git a/copy b/rename1
  copy from copy
  copy to rename1
  --- a/copy
  +++ b/rename1
  @@ -1,2 +1,3 @@
   new
   copy1
  +rename1
  diff --git a/rename2 b/rename2
  new file mode 100644
  --- /dev/null
  +++ b/rename2
  @@ -0,0 +1,3 @@
  +new
  +copy2
  +rename2

Delete:

  $ hg rm dir1/*
  $ hg ci -mdelete
  $ hg diff --git -r 3:tip
  diff --git a/dir1/new b/dir1/new
  deleted file mode 100644
  --- a/dir1/new
  +++ /dev/null
  @@ -1,1 +0,0 @@
  -new
  diff --git a/dir1/rename1 b/dir1/rename1
  deleted file mode 100644
  --- a/dir1/rename1
  +++ /dev/null
  @@ -1,3 +0,0 @@
  -new
  -copy1
  -rename1
  diff --git a/dir1/rename2 b/dir1/rename2
  deleted file mode 100644
  --- a/dir1/rename2
  +++ /dev/null
  @@ -1,3 +0,0 @@
  -new
  -copy2
  -rename2

  $ cat > src <<EOF
  > 1
  > 2
  > 3
  > 4
  > 5
  > EOF
  $ hg ci -Amsrc
  adding src

#if execbit

chmod 644:

  $ chmod +x src
  $ hg ci -munexec
  $ hg diff --git -r 5:tip
  diff --git a/src b/src
  old mode 100644
  new mode 100755

Rename+mod+chmod:

  $ hg mv src dst
  $ chmod -x dst
  $ echo a >> dst
  $ hg ci -mrenamemod
  $ hg diff --git -r 6:tip
  diff --git a/src b/dst
  old mode 100755
  new mode 100644
  rename from src
  rename to dst
  --- a/src
  +++ b/dst
  @@ -3,3 +3,4 @@
   3
   4
   5
  +a

Nonexistent in tip+chmod:

  $ hg diff --git -r 5:6
  diff --git a/src b/src
  old mode 100644
  new mode 100755

#else

Dummy changes when no exec bit, mocking the execbit commit structure

  $ echo change >> src
  $ hg ci -munexec
  $ hg mv src dst
  $ hg ci -mrenamemod

#endif

Binary diff:

  $ cp "$TESTDIR/binfile.bin" .
  $ hg add binfile.bin
  $ hg diff --git > b.diff
  $ cat b.diff
  diff --git a/binfile.bin b/binfile.bin
  new file mode 100644
  index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..37ba3d1c6f17137d9c5f5776fa040caf5fe73ff9
  GIT binary patch
  literal 593
  zc$@)I0<QguP)<h;3K|Lk000e1NJLTq000mG000mO0ssI2kdbIM00009a7bBm000XU
  z000XU0RWnu7ytkO2XskIMF-Uh9TW;VpMjwv0005-Nkl<ZD9@FWPs=e;7{<>W$NUkd
  zX$nnYLt$-$V!?uy+1V%`z&Eh=ah|duER<4|QWhju3gb^nF*8iYobxWG-qqXl=2~5M
  z*IoDB)sG^CfNuoBmqLTVU^<;@nwHP!1wrWd`{(mHo6VNXWtyh{alzqmsH*yYzpvLT
  zLdY<T=ks|woh-`&01!ej#(xbV1f|pI*=%;d-%F*E*X#ZH`4I%6SS+$EJDE&ct=8po
  ziN#{?_j|kD%Cd|oiqds`xm@;oJ-^?NG3Gdqrs?5u*zI;{nogxsx~^|Fn^Y?Gdc6<;
  zfMJ+iF1J`LMx&A2?dEwNW8ClebzPTbIh{@$hS6*`kH@1d%Lo7fA#}N1)oN7`gm$~V
  z+wDx#)OFqMcE{s!JN0-xhG8ItAjVkJwEcb`3WWlJfU2r?;Pd%dmR+q@mSri5q9_W-
  zaR2~ECX?B2w+zELozC0s*6Z~|QG^f{3I#<`?)Q7U-JZ|q5W;9Q8i_=pBuSzunx=U;
  z9C)5jBoYw9^?EHyQl(M}1OlQcCX>lXB*ODN003Z&P17_@)3Pi=i0wb04<W?v-u}7K
  zXmmQA+wDgE!qR9o8jr`%=ab_&uh(l?R=r;Tjiqon91I2-hIu?57~@*4h7h9uORK#=
  fQItJW-{SoTm)8|5##k|m00000NkvXXu0mjf{mKw4
  

Import binary diff:

  $ hg revert binfile.bin
  $ rm binfile.bin
  $ hg import -mfoo b.diff
  applying b.diff
  $ cmp binfile.bin "$TESTDIR/binfile.bin"

Rename binary file:

  $ hg mv binfile.bin renamed.bin
  $ hg diff --git
  diff --git a/binfile.bin b/renamed.bin
  rename from binfile.bin
  rename to renamed.bin

Diff across many revisions:

  $ hg mv dst dst2
  $ hg ci -m 'mv dst dst2'

  $ echo >> start
  $ hg ci -m 'change start'

  $ hg revert -r -2 start
  $ hg mv dst2 dst3
  $ hg ci -m 'mv dst2 dst3; revert start'

  $ hg diff --git -r 9:11
  diff --git a/dst2 b/dst3
  rename from dst2
  rename to dst3

Reversed:

  $ hg diff --git -r 11:9
  diff --git a/dst3 b/dst2
  rename from dst3
  rename to dst2


  $ echo a >> foo
  $ hg add foo
  $ hg ci -m 'add foo'
  $ echo b >> foo
  $ hg ci -m 'change foo'
  $ hg mv foo bar
  $ hg ci -m 'mv foo bar'
  $ echo c >> bar
  $ hg ci -m 'change bar'

File created before r1 and renamed before r2:

  $ hg diff --git -r -3:-1
  diff --git a/foo b/bar
  rename from foo
  rename to bar
  --- a/foo
  +++ b/bar
  @@ -1,2 +1,3 @@
   a
   b
  +c

Reversed:

  $ hg diff --git -r -1:-3
  diff --git a/bar b/foo
  rename from bar
  rename to foo
  --- a/bar
  +++ b/foo
  @@ -1,3 +1,2 @@
   a
   b
  -c

File created in r1 and renamed before r2:

  $ hg diff --git -r -4:-1
  diff --git a/foo b/bar
  rename from foo
  rename to bar
  --- a/foo
  +++ b/bar
  @@ -1,1 +1,3 @@
   a
  +b
  +c

Reversed:

  $ hg diff --git -r -1:-4
  diff --git a/bar b/foo
  rename from bar
  rename to foo
  --- a/bar
  +++ b/foo
  @@ -1,3 +1,1 @@
   a
  -b
  -c

File created after r1 and renamed before r2:

  $ hg diff --git -r -5:-1
  diff --git a/bar b/bar
  new file mode 100644
  --- /dev/null
  +++ b/bar
  @@ -0,0 +1,3 @@
  +a
  +b
  +c

Reversed:

  $ hg diff --git -r -1:-5
  diff --git a/bar b/bar
  deleted file mode 100644
  --- a/bar
  +++ /dev/null
  @@ -1,3 +0,0 @@
  -a
  -b
  -c


Comparing with the working dir:

  $ echo >> start
  $ hg ci -m 'change start again'

  $ echo > created
  $ hg add created
  $ hg ci -m 'add created'

  $ hg mv created created2
  $ hg ci -m 'mv created created2'

  $ hg mv created2 created3

There's a copy in the working dir:

  $ hg diff --git
  diff --git a/created2 b/created3
  rename from created2
  rename to created3

There's another copy between the original rev and the wd:

  $ hg diff --git -r -2
  diff --git a/created b/created3
  rename from created
  rename to created3

The source of the copy was created after the original rev:

  $ hg diff --git -r -3
  diff --git a/created3 b/created3
  new file mode 100644
  --- /dev/null
  +++ b/created3
  @@ -0,0 +1,1 @@
  +
  $ hg ci -m 'mv created2 created3'


  $ echo > brand-new
  $ hg add brand-new
  $ hg ci -m 'add brand-new'
  $ hg mv brand-new brand-new2

Created in parent of wd; renamed in the wd:

  $ hg diff --git
  diff --git a/brand-new b/brand-new2
  rename from brand-new
  rename to brand-new2

Created between r1 and parent of wd; renamed in the wd:

  $ hg diff --git -r -2
  diff --git a/brand-new2 b/brand-new2
  new file mode 100644
  --- /dev/null
  +++ b/brand-new2
  @@ -0,0 +1,1 @@
  +
  $ hg ci -m 'mv brand-new brand-new2'

One file is copied to many destinations and removed:

  $ hg cp brand-new2 brand-new3
  $ hg mv brand-new2 brand-new3-2
  $ hg ci -m 'multiple renames/copies'
  $ hg diff --git -r -2 -r -1
  diff --git a/brand-new2 b/brand-new3
  rename from brand-new2
  rename to brand-new3
  diff --git a/brand-new2 b/brand-new3-2
  copy from brand-new2
  copy to brand-new3-2

Reversed:

  $ hg diff --git -r -1 -r -2
  diff --git a/brand-new3-2 b/brand-new2
  rename from brand-new3-2
  rename to brand-new2
  diff --git a/brand-new3 b/brand-new3
  deleted file mode 100644
  --- a/brand-new3
  +++ /dev/null
  @@ -1,1 +0,0 @@
  -

There should be a trailing TAB if there are spaces in the file name:

  $ echo foo > 'with spaces'
  $ hg add 'with spaces'
  $ hg diff --git
  diff --git a/with spaces b/with spaces
  new file mode 100644
  --- /dev/null
  +++ b/with spaces	
  @@ -0,0 +1,1 @@
  +foo
  $ hg ci -m 'add filename with spaces'

Additions should be properly marked even in the middle of a merge

  $ hg up -r -2
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ echo "New File" >> inmerge
  $ hg add inmerge
  $ hg ci -m "file in merge"
  created new head
  $ hg up 23
  1 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg merge
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)
  $ hg diff -g
  diff --git a/inmerge b/inmerge
  new file mode 100644
  --- /dev/null
  +++ b/inmerge
  @@ -0,0 +1,1 @@
  +New File