tests/test-split.t

changegroup: remove reordering control (BC)

This logic - including the experimental bundle.reorder option -
was originally added in a8e3931e3fb5 in 2011 and then later ported
to changegroup.py.

The intent of this option and associated logic is to control
the ordering of revisions in deltagroups in changegroups. At the
time it was implemented, only changegroup version 1 existed
and generaldelta revlogs were just coming into the world. Changegroup
version 1 requires that deltas be made against the last revision
sent over the wire. Used with generaldelta, this created an
impedance mismatch of sorts and resulted in changegroup producers
spending a lot of time recomputing deltas.

Revision reordering was introduced so outgoing revisions would be
sent in "generaldelta order" and producers would be able to
reuse internal deltas from storage.

Later on, we introduced changegroup version 2. It supported denoting
which revision a delta was against. So we no longer needed to
sort outgoing revisions to ensure optimal delta generation from the
producer. So, subsequent changegroup versions disabled reordering.

We also later made the changelog not store deltas by default. And
we also made the changelog send out deltas in storage order. Why we
do this for changelog, I'm not sure. Maybe we want to preserve revision
order across clones? It doesn't really matter for this commit.

Fast forward to 2018. We want to abstract storage backends. And having
changegroup code require knowledge about how deltas are stored
internally interferes with that goal.

This commit removes reordering control from changegroup generation.
After this commit, the reordering behavior is:

* The changelog is always sent out in storage order (no behavior
  change).
* Non-changelog generaldelta revlogs are reordered to always be in DAG
  topological order (previously, generaldelta revlogs would be emitted
  in storage order for version 2 and 3 changegroups).
* Non-changelog non-generaldelta revlogs are sent in storage order (no
  behavior change).
* There exists no config option to override behavior.

The big difference here is that generaldelta revlogs now *always* have
their revisions sorted in DAG order before going out over the wire. This
behavior was previously only done for changegroup version 1. Version 2
and version 3 changegroups disabled reordering because the interchange
format supported encoding arbitrary delta parents, so reordering wasn't
strictly necessary.

I can think of a few significant implications for this change.

Because changegroup receivers will now see non-changelog revisions
in DAG order instead of storage order, the internal storage order of
manifests and files may differ substantially between producer and
consumer. I don't think this matters that much, since the storage
order of manifests and files is largely hidden from users. Only
the storage order of changelog matters (because `hg log` shows the
changelog in storage order). I don't think there should be any
controversy here.

The reordering of revisions has implications for changegroup producers.
Previously, generaldelta revlogs would be emitted in storage order.
And in the common case, the internally-stored delta could effectively
be copied from disk into the deltagroup delta. This meant that emitting
delta groups for generaldelta revlogs would be mostly linear read I/O.
This is desirable for performance. With us now reordering generaldelta
revlog revisions in DAG order, the read operations may use more random
I/O instead of sequential I/O. This could result in performance
loss. But with the prevalence of SSDs and fast random I/O, I'm not
too worried. (Note: the optimal emission order for revlogs is actually
delta encoding order. But the changegroup code wasn't doing that before
or after this change. We could potentially implement that in a later
commit.)

Changegroups in DAG order will have implications for receivers.
Previously, receiving storage order might mean seeing a number of
interleaved branches. This would mean long delta chains, sparse
I/O, and possibly more fulltext revisions instead of deltas, blowing
up storage storage. (This is the same set of problems that sparse
revlogs aims to address.) With the producer now sending revisions in DAG
order, the receiver also stores revisions in DAG order. That means
revisions for the same DAG branch are all grouped together. And this
should yield better storage outcomes. In other words, sending the
reordered changegroup allows the receiver to have better storage
order and for the producer to not propagate its (possibly sub-optimal)
internal storage order.

On the mozilla-unified repository, this change influences bundle
generation:

$ hg bundle -t none-v2 -a
before: time: real 355.680 secs (user 256.790+0.000 sys 16.820+0.000)
after:  time: real 382.950 secs (user 281.700+0.000 sys 17.690+0.000)

before: 7,150,228,967 bytes (uncompressed)
after:  7,041,556,273 bytes (uncompressed)

before: 1,669,063,234 bytes (zstd l=3)
after:  1,628,598,830 bytes (zstd l=3)

$ hg unbundle
before: time: real 511.910 secs (user 466.750+0.000 sys 32.680+0.000)
after:  time: real 487.790 secs (user 443.940+0.000 sys 30.840+0.000)

00manifest.d size:
source: 274,924,292 bytes
before: 304,741,626 bytes
after:  245,252,087 bytes

.hg/store total file size:
source: 2,649,133,490
before: 2,680,888,130
after:  2,627,875,673

We see the bundle size drop. That's probably because if a revlog
internally isn't storing a delta, it will choose to delta against
the last emitted revision. And on repos with interleaved branches
(like mozilla-unified), the previous revision could be an unrelated
branch and therefore be a large delta. But with this patch, the
previous revision is likely p1 or p2 and a delta should be small.

We also see the manifest size drop by ~50 MB. It's worth noting that
the manifest actually *increased* in size by ~25 MB in the old
strategy and decreased ~25 MB from its source in the new strategy.
Again, my explanation for this is that the DAG ordering in the
changegroup is resulting in better grouping of revisions in the
receiver, which results in more compact delta chains and higher
storage efficiency.

Unbundle time also dropped. I suspect this is due to the revlog having
to work less to compute deltas since the incoming deltas are more
optimal. i.e. the receiver spends less time resolving fulltext
revisions as incoming deltas bounce around between DAG branches and
delta chains.

We also see bundle generation time increase. This is not desirable.
However, the regression is only significant on the original repository:
if we generate a bundle from the repository created from the new,
always reordered bundles, we're close to baseline (if not at it with
expected noise):

$ hg bundle -t none-v2 -a
before (original): time: real 355.680 secs (user 256.790+0.000 sys 16.820+0.000)
after (original):  time: real 382.950 secs (user 281.700+0.000 sys 17.690+0.000)
after (new repo):  time: real 362.280 secs (user 260.300+0.000 sys 17.700+0.000)

This regression is a bit worrying because it will impact serving
canonical repositories (that don't have optimal internal storage
unless they are reordered - possibly as part of running
`hg debugupgraderepo`). However, this regression will only be
noticed by very large changegroups. And I'm guessing/hoping that
any repository that large is using clonebundles to mitigate server
load.

Again, sending DAG order isn't the optimal send order for servers:
sending in storage-delta order is. But in order to enable
storage-optimal send order, we'll need a storage API that handles
sorting. Future commits will introduce such an API.

Differential Revision: https://phab.mercurial-scm.org/D4721

#testcases obsstore-on obsstore-off

  $ cat > $TESTTMP/editor.py <<EOF
  > #!"$PYTHON"
  > import os
  > import sys
  > path = os.path.join(os.environ['TESTTMP'], 'messages')
  > messages = open(path).read().split('--\n')
  > prompt = open(sys.argv[1]).read()
  > sys.stdout.write(''.join('EDITOR: %s' % l for l in prompt.splitlines(True)))
  > sys.stdout.flush()
  > with open(sys.argv[1], 'w') as f:
  >    f.write(messages[0])
  > with open(path, 'w') as f:
  >    f.write('--\n'.join(messages[1:]))
  > EOF

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > drawdag=$TESTDIR/drawdag.py
  > split=
  > [ui]
  > interactive=1
  > color=no
  > paginate=never
  > [diff]
  > git=1
  > unified=0
  > [alias]
  > glog=log -G -T '{rev}:{node|short} {desc} {bookmarks}\n'
  > EOF

#if obsstore-on
  $ cat >> $HGRCPATH <<EOF
  > [experimental]
  > evolution=all
  > EOF
#endif

  $ hg init a
  $ cd a

Nothing to split

  $ hg split
  nothing to split
  [1]

  $ hg commit -m empty --config ui.allowemptycommit=1
  $ hg split
  abort: cannot split an empty revision
  [255]

  $ rm -rf .hg
  $ hg init

Cannot split working directory

  $ hg split -r 'wdir()'
  abort: cannot split working directory
  [255]

Generate some content.  The sed filter drop CR on Windows, which is dropped in
the a > b line.

  $ $TESTDIR/seq.py 1 5 | sed 's/\r$//' >> a
  $ hg ci -m a1 -A a -q
  $ hg bookmark -i r1
  $ sed 's/1/11/;s/3/33/;s/5/55/' a > b
  $ mv b a
  $ hg ci -m a2 -q
  $ hg bookmark -i r2

Cannot split a public changeset

  $ hg phase --public -r 'all()'
  $ hg split .
  abort: cannot split public changeset
  (see 'hg help phases' for details)
  [255]

  $ hg phase --draft -f -r 'all()'

Cannot split while working directory is dirty

  $ touch dirty
  $ hg add dirty
  $ hg split .
  abort: uncommitted changes
  [255]
  $ hg forget dirty
  $ rm dirty

Make a clean directory for future tests to build off of

  $ cp -R . ../clean

Split a head

  $ hg bookmark r3

  $ hg split 'all()'
  abort: cannot split multiple revisions
  [255]

  $ runsplit() {
  > cat > $TESTTMP/messages <<EOF
  > split 1
  > --
  > split 2
  > --
  > split 3
  > EOF
  > cat <<EOF | hg split "$@"
  > y
  > y
  > y
  > y
  > y
  > y
  > EOF
  > }

  $ HGEDITOR=false runsplit
  diff --git a/a b/a
  1 hunks, 1 lines changed
  examine changes to 'a'? [Ynesfdaq?] y
  
  @@ -5,1 +5,1 @@ 4
  -5
  +55
  record this change to 'a'? [Ynesfdaq?] y
  
  transaction abort!
  rollback completed
  abort: edit failed: false exited with status 1
  [255]
  $ hg status

  $ HGEDITOR="\"$PYTHON\" $TESTTMP/editor.py"
  $ runsplit
  diff --git a/a b/a
  1 hunks, 1 lines changed
  examine changes to 'a'? [Ynesfdaq?] y
  
  @@ -5,1 +5,1 @@ 4
  -5
  +55
  record this change to 'a'? [Ynesfdaq?] y
  
  EDITOR: HG: Splitting 1df0d5c5a3ab. Write commit message for the first split changeset.
  EDITOR: a2
  EDITOR: 
  EDITOR: 
  EDITOR: HG: Enter commit message.  Lines beginning with 'HG:' are removed.
  EDITOR: HG: Leave message empty to abort commit.
  EDITOR: HG: --
  EDITOR: HG: user: test
  EDITOR: HG: branch 'default'
  EDITOR: HG: changed a
  created new head
  diff --git a/a b/a
  1 hunks, 1 lines changed
  examine changes to 'a'? [Ynesfdaq?] y
  
  @@ -3,1 +3,1 @@ 2
  -3
  +33
  record this change to 'a'? [Ynesfdaq?] y
  
  EDITOR: HG: Splitting 1df0d5c5a3ab. So far it has been split into:
  EDITOR: HG: - e704349bd21b: split 1
  EDITOR: HG: Write commit message for the next split changeset.
  EDITOR: a2
  EDITOR: 
  EDITOR: 
  EDITOR: HG: Enter commit message.  Lines beginning with 'HG:' are removed.
  EDITOR: HG: Leave message empty to abort commit.
  EDITOR: HG: --
  EDITOR: HG: user: test
  EDITOR: HG: branch 'default'
  EDITOR: HG: changed a
  diff --git a/a b/a
  1 hunks, 1 lines changed
  examine changes to 'a'? [Ynesfdaq?] y
  
  @@ -1,1 +1,1 @@
  -1
  +11
  record this change to 'a'? [Ynesfdaq?] y
  
  EDITOR: HG: Splitting 1df0d5c5a3ab. So far it has been split into:
  EDITOR: HG: - e704349bd21b: split 1
  EDITOR: HG: - a09ad58faae3: split 2
  EDITOR: HG: Write commit message for the next split changeset.
  EDITOR: a2
  EDITOR: 
  EDITOR: 
  EDITOR: HG: Enter commit message.  Lines beginning with 'HG:' are removed.
  EDITOR: HG: Leave message empty to abort commit.
  EDITOR: HG: --
  EDITOR: HG: user: test
  EDITOR: HG: branch 'default'
  EDITOR: HG: changed a
  saved backup bundle to $TESTTMP/a/.hg/strip-backup/1df0d5c5a3ab-8341b760-split.hg (obsstore-off !)

#if obsstore-off
  $ hg bookmark
     r1                        0:a61bcde8c529
     r2                        3:00eebaf8d2e2
   * r3                        3:00eebaf8d2e2
  $ hg glog -p
  @  3:00eebaf8d2e2 split 3 r2 r3
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -1,1 +1,1 @@
  |  -1
  |  +11
  |
  o  2:a09ad58faae3 split 2
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -3,1 +3,1 @@
  |  -3
  |  +33
  |
  o  1:e704349bd21b split 1
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -5,1 +5,1 @@
  |  -5
  |  +55
  |
  o  0:a61bcde8c529 a1 r1
     diff --git a/a b/a
     new file mode 100644
     --- /dev/null
     +++ b/a
     @@ -0,0 +1,5 @@
     +1
     +2
     +3
     +4
     +5
  
#else
  $ hg bookmark
     r1                        0:a61bcde8c529
     r2                        4:00eebaf8d2e2
   * r3                        4:00eebaf8d2e2
  $ hg glog
  @  4:00eebaf8d2e2 split 3 r2 r3
  |
  o  3:a09ad58faae3 split 2
  |
  o  2:e704349bd21b split 1
  |
  o  0:a61bcde8c529 a1 r1
  
#endif

Split a head while working parent is not that head

  $ cp -R $TESTTMP/clean $TESTTMP/b
  $ cd $TESTTMP/b

  $ hg up 0 -q
  $ hg bookmark r3

  $ runsplit tip >/dev/null

#if obsstore-off
  $ hg bookmark
     r1                        0:a61bcde8c529
     r2                        3:00eebaf8d2e2
   * r3                        0:a61bcde8c529
  $ hg glog
  o  3:00eebaf8d2e2 split 3 r2
  |
  o  2:a09ad58faae3 split 2
  |
  o  1:e704349bd21b split 1
  |
  @  0:a61bcde8c529 a1 r1 r3
  
#else
  $ hg bookmark
     r1                        0:a61bcde8c529
     r2                        4:00eebaf8d2e2
   * r3                        0:a61bcde8c529
  $ hg glog
  o  4:00eebaf8d2e2 split 3 r2
  |
  o  3:a09ad58faae3 split 2
  |
  o  2:e704349bd21b split 1
  |
  @  0:a61bcde8c529 a1 r1 r3
  
#endif

Split a non-head

  $ cp -R $TESTTMP/clean $TESTTMP/c
  $ cd $TESTTMP/c
  $ echo d > d
  $ hg ci -m d1 -A d
  $ hg bookmark -i d1
  $ echo 2 >> d
  $ hg ci -m d2
  $ echo 3 >> d
  $ hg ci -m d3
  $ hg bookmark -i d3
  $ hg up '.^' -q
  $ hg bookmark d2
  $ cp -R . ../d

  $ runsplit -r 1 | grep rebasing
  rebasing 2:b5c5ea414030 "d1" (d1)
  rebasing 3:f4a0a8d004cc "d2" (d2)
  rebasing 4:777940761eba "d3" (d3)
#if obsstore-off
  $ hg bookmark
     d1                        4:c4b449ef030e
   * d2                        5:c9dd00ab36a3
     d3                        6:19f476bc865c
     r1                        0:a61bcde8c529
     r2                        3:00eebaf8d2e2
  $ hg glog -p
  o  6:19f476bc865c d3 d3
  |  diff --git a/d b/d
  |  --- a/d
  |  +++ b/d
  |  @@ -2,0 +3,1 @@
  |  +3
  |
  @  5:c9dd00ab36a3 d2 d2
  |  diff --git a/d b/d
  |  --- a/d
  |  +++ b/d
  |  @@ -1,0 +2,1 @@
  |  +2
  |
  o  4:c4b449ef030e d1 d1
  |  diff --git a/d b/d
  |  new file mode 100644
  |  --- /dev/null
  |  +++ b/d
  |  @@ -0,0 +1,1 @@
  |  +d
  |
  o  3:00eebaf8d2e2 split 3 r2
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -1,1 +1,1 @@
  |  -1
  |  +11
  |
  o  2:a09ad58faae3 split 2
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -3,1 +3,1 @@
  |  -3
  |  +33
  |
  o  1:e704349bd21b split 1
  |  diff --git a/a b/a
  |  --- a/a
  |  +++ b/a
  |  @@ -5,1 +5,1 @@
  |  -5
  |  +55
  |
  o  0:a61bcde8c529 a1 r1
     diff --git a/a b/a
     new file mode 100644
     --- /dev/null
     +++ b/a
     @@ -0,0 +1,5 @@
     +1
     +2
     +3
     +4
     +5
  
#else
  $ hg bookmark
     d1                        8:c4b449ef030e
   * d2                        9:c9dd00ab36a3
     d3                        10:19f476bc865c
     r1                        0:a61bcde8c529
     r2                        7:00eebaf8d2e2
  $ hg glog
  o  10:19f476bc865c d3 d3
  |
  @  9:c9dd00ab36a3 d2 d2
  |
  o  8:c4b449ef030e d1 d1
  |
  o  7:00eebaf8d2e2 split 3 r2
  |
  o  6:a09ad58faae3 split 2
  |
  o  5:e704349bd21b split 1
  |
  o  0:a61bcde8c529 a1 r1
  
#endif

Split a non-head without rebase

  $ cd $TESTTMP/d
#if obsstore-off
  $ runsplit -r 1 --no-rebase
  abort: cannot split changeset with children without rebase
  [255]
#else
  $ runsplit -r 1 --no-rebase >/dev/null
  3 new orphan changesets
  $ hg bookmark
     d1                        2:b5c5ea414030
   * d2                        3:f4a0a8d004cc
     d3                        4:777940761eba
     r1                        0:a61bcde8c529
     r2                        7:00eebaf8d2e2

  $ hg glog
  o  7:00eebaf8d2e2 split 3 r2
  |
  o  6:a09ad58faae3 split 2
  |
  o  5:e704349bd21b split 1
  |
  | *  4:777940761eba d3 d3
  | |
  | @  3:f4a0a8d004cc d2 d2
  | |
  | *  2:b5c5ea414030 d1 d1
  | |
  | x  1:1df0d5c5a3ab a2
  |/
  o  0:a61bcde8c529 a1 r1
  
#endif

Split a non-head with obsoleted descendants

#if obsstore-on
  $ hg init $TESTTMP/e
  $ cd $TESTTMP/e
  $ hg debugdrawdag <<'EOS'
  >   H I   J
  >   | |   |
  >   F G1 G2  # amend: G1 -> G2
  >   | |  /   # prune: F
  >   C D E
  >    \|/
  >     B
  >     |
  >     A
  > EOS
  2 new orphan changesets
  $ eval `hg tags -T '{tag}={node}\n'`
  $ rm .hg/localtags
  $ hg split $B --config experimental.evolution=createmarkers
  abort: split would leave orphaned changesets behind
  [255]
  $ cat > $TESTTMP/messages <<EOF
  > Split B
  > EOF
  $ cat <<EOF | hg split $B
  > y
  > y
  > EOF
  diff --git a/B b/B
  new file mode 100644
  examine changes to 'B'? [Ynesfdaq?] y
  
  @@ -0,0 +1,1 @@
  +B
  \ No newline at end of file
  record this change to 'B'? [Ynesfdaq?] y
  
  EDITOR: HG: Splitting 112478962961. Write commit message for the first split changeset.
  EDITOR: B
  EDITOR: 
  EDITOR: 
  EDITOR: HG: Enter commit message.  Lines beginning with 'HG:' are removed.
  EDITOR: HG: Leave message empty to abort commit.
  EDITOR: HG: --
  EDITOR: HG: user: test
  EDITOR: HG: branch 'default'
  EDITOR: HG: added B
  created new head
  rebasing 2:26805aba1e60 "C"
  rebasing 3:be0ef73c17ad "D"
  rebasing 4:49cb92066bfd "E"
  rebasing 7:97a6268cc7ef "G2"
  rebasing 10:e2f1e425c0db "J"
  $ hg glog -r 'sort(all(), topo)'
  o  16:556c085f8b52 J
  |
  o  15:8761f6c9123f G2
  |
  o  14:a7aeffe59b65 E
  |
  | o  13:e1e914ede9ab D
  |/
  | o  12:01947e9b98aa C
  |/
  o  11:0947baa74d47 Split B
  |
  | *  9:88ede1d5ee13 I
  | |
  | x  6:af8cbf225b7b G1
  | |
  | x  3:be0ef73c17ad D
  | |
  | | *  8:74863e5b5074 H
  | | |
  | | x  5:ee481a2a1e69 F
  | | |
  | | x  2:26805aba1e60 C
  | |/
  | x  1:112478962961 B
  |/
  o  0:426bada5c675 A
  
#endif

Preserve secret phase in split

  $ cp -R $TESTTMP/clean $TESTTMP/phases1
  $ cd $TESTTMP/phases1
  $ hg phase --secret -fr tip
  $ hg log -T '{short(node)} {phase}\n'
  1df0d5c5a3ab secret
  a61bcde8c529 draft
  $ runsplit tip >/dev/null
  $ hg log -T '{short(node)} {phase}\n'
  00eebaf8d2e2 secret
  a09ad58faae3 secret
  e704349bd21b secret
  a61bcde8c529 draft

Do not move things to secret even if phases.new-commit=secret

  $ cp -R $TESTTMP/clean $TESTTMP/phases2
  $ cd $TESTTMP/phases2
  $ cat >> .hg/hgrc <<EOF
  > [phases]
  > new-commit=secret
  > EOF
  $ hg log -T '{short(node)} {phase}\n'
  1df0d5c5a3ab draft
  a61bcde8c529 draft
  $ runsplit tip >/dev/null
  $ hg log -T '{short(node)} {phase}\n'
  00eebaf8d2e2 draft
  a09ad58faae3 draft
  e704349bd21b draft
  a61bcde8c529 draft