view tests/test-rebase-newancestor.t @ 40326:fed697fa1734

sqlitestore: file storage backend using SQLite

This commit provides an extension which uses SQLite to store file
data (as opposed to revlogs).

As the inline documentation describes, there are still several
aspects to the extension that are incomplete. But it's a start.
The extension does support basic clone, checkout, and commit
workflows, which makes it suitable for simple use cases.

One notable missing feature is support for "bundlerepos." This is
probably responsible for the most test failures when the extension
is activated as part of the test suite.

All revision data is stored in SQLite. Data is stored as zstd
compressed chunks (default if zstd is available), zlib compressed
chunks (default if zstd is not available), or raw chunks (if
configured or if a compressed delta is not smaller than the raw
delta). This makes things very similar to revlogs.

Unlike revlogs, the extension doesn't yet enforce a limit on delta
chain length. This is an obvious limitation and should be addressed.
This is somewhat mitigated by the use of zstd, which is much faster
than zlib to decompress.

There is a dedicated table for storing deltas. Deltas are stored
by the SHA-1 hash of their uncompressed content. The "fileindex" table
has columns that reference the delta for each revision and the base
delta that delta should be applied against. A recursive SQL query
is used to resolve the delta chain along with the delta data.

By storing deltas by hash, we are able to de-duplicate delta storage!
With revlogs, the same deltas in different revlogs would result in
duplicate storage of that delta. In this scheme, inserting the
duplicate delta is a no-op and delta chains simply reference the
existing delta.

When initially implementing this extension, I did not have
content-indexed deltas and deltas could be duplicated across files
(just like revlogs). When I implemented content-indexed deltas, the
size of the SQLite database for a full clone of mozilla-unified
dropped:

before: 2,554,261,504 bytes
after:  2,488,754,176 bytes

Surprisingly, this is still larger than the bytes size of revlog
files:

revlog files: 2,104,861,230 bytes
du -b:        2,254,381,614

I would have expected storage to be smaller since we're not limiting
delta chain length and since we're using zstd instead of zlib. I
suspect the SQLite indexes and per-column overhead account for the
bulk of the differences. (Keep in mind that revlog uses a 64-byte
packed struct for revision index data and deltas are stored without
padding. Aside from the 12 unused bytes in the 32 byte node field,
revlogs are pretty efficient.) Another source of overhead is file
name storage. With revlogs, file names are stored in the filesystem.
But with SQLite, we need to store file names in the database. This is
roughly equivalent to the size of the fncache file, which for the
mozilla-unified repository is ~34MB.

Since the SQLite database isn't append-only and since delta chains
can reference any delta, this opens some interesting possibilities.
For example, we could store deltas in reverse, such that fulltexts
are stored for newer revisions and deltas are applied to reconstruct
older revisions. This is likely a more optimal storage strategy for
version control, as new data tends to be more frequently accessed
than old data. We would obviously need wire protocol support for
transferring revision data from newest to oldest. And we would
probably need some kind of mechanism for "re-encoding" stores. But
it should be doable.

This extension is very much experimental quality. There are a handful
of features that don't work. It probably isn't suitable for day-to-day
use. But it could be used in limited cases (e.g. read-only checkouts
like in CI). And it is also a good proving ground for alternate
storage backends. As we continue to define interfaces for all things
storage, it will be useful to have a viable alternate storage backend
to see how things shake out in practice.

test-storage.py passes on Python 2 and introduces no new test failures on
Python 3. Having the storage-level unit tests has proved to be insanely
useful when developing this extension. Those tests caught numerous bugs
during development and I'm convinced this style of testing is the way
forward for ensuring alternate storage backends work as intended. Of
course, test coverage isn't close to what it needs to be. But it is
a start. And what coverage we have gives me confidence that basic store
functionality is implemented properly.

Differential Revision: https://phab.mercurial-scm.org/D4928

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > rebase=
  > drawdag=$TESTDIR/drawdag.py
  > [alias]
  > tglog = log -G --template "{rev}: {node|short} '{desc}' {branches}\n"
  > EOF

  $ hg init repo
  $ cd repo

  $ echo A > a
  $ echo >> a
  $ hg ci -Am A
  adding a

  $ echo B > a
  $ echo >> a
  $ hg ci -m B

  $ echo C > a
  $ echo >> a
  $ hg ci -m C

  $ hg up -q -C 0

  $ echo D >> a
  $ hg ci -Am AD
  created new head

  $ hg tglog
  @  3: 3878212183bd 'AD'
  |
  | o  2: 30ae917c0e4f 'C'
  | |
  | o  1: 0f4f7cb4f549 'B'
  |/
  o  0: 1e635d440a73 'A'
  
  $ hg rebase -s 1 -d 3
  rebasing 1:0f4f7cb4f549 "B"
  merging a
  rebasing 2:30ae917c0e4f "C"
  merging a
  saved backup bundle to $TESTTMP/repo/.hg/strip-backup/0f4f7cb4f549-82b3b163-rebase.hg

  $ hg tglog
  o  3: 25773bc4b4b0 'C'
  |
  o  2: c09015405f75 'B'
  |
  @  1: 3878212183bd 'AD'
  |
  o  0: 1e635d440a73 'A'
  

  $ cd ..


Test rebasing of merges with ancestors of the rebase destination - a situation
that often happens when trying to recover from repeated merging with a mainline
branch.

The test case creates a dev branch that contains a couple of merges from the
default branch. When rebasing to the default branch, these merges would be
merges with ancestors on the same branch. The merges _could_ contain some
interesting conflict resolutions or additional changes in the merge commit, but
that is mixed up with the actual merge stuff and there is in general no way to
separate them.

Note: The dev branch contains _no_ changes to f-default. It might be unclear
how rebasing of ancestor merges should be handled, but the current behavior
with spurious prompts for conflicts in files that didn't change seems very
wrong.

  $ hg init ancestor-merge
  $ cd ancestor-merge

  $ touch f-default
  $ hg ci -Aqm 'default: create f-default'

  $ hg branch -q dev
  $ hg ci -qm 'dev: create branch'

  $ echo stuff > f-dev
  $ hg ci -Aqm 'dev: f-dev stuff'

  $ hg up -q default
  $ echo stuff > f-default
  $ hg ci -m 'default: f-default stuff'

  $ hg up -q dev
  $ hg merge -q default
  $ hg ci -m 'dev: merge default'

  $ hg up -q default
  $ hg rm f-default
  $ hg ci -m 'default: remove f-default'

  $ hg up -q dev
  $ hg merge -q default
  $ hg ci -m 'dev: merge default'

  $ hg up -q default
  $ echo stuff > f-other
  $ hg ci -Aqm 'default: f-other stuff'

  $ hg tglog
  @  7: e08089805d82 'default: f-other stuff'
  |
  | o  6: 9455ee510502 'dev: merge default' dev
  |/|
  o |  5: 462860db70a1 'default: remove f-default'
  | |
  | o  4: 4b019212aaf6 'dev: merge default' dev
  |/|
  o |  3: f157ecfd2b6b 'default: f-default stuff'
  | |
  | o  2: ec2c14fb2984 'dev: f-dev stuff' dev
  | |
  | o  1: 1d1a643d390e 'dev: create branch' dev
  |/
  o  0: e90e8eb90b6f 'default: create f-default'
  
  $ hg clone -qU . ../ancestor-merge-2

Full rebase all the way back from branching point:

  $ hg rebase -r 'only(dev,default)' -d default --config ui.interactive=True << EOF
  > c
  > EOF
  rebasing 1:1d1a643d390e "dev: create branch"
  note: rebase of 1:1d1a643d390e created no changes to commit
  rebasing 2:ec2c14fb2984 "dev: f-dev stuff"
  rebasing 4:4b019212aaf6 "dev: merge default"
  file 'f-default' was deleted in local [dest] but was modified in other [source].
  What do you want to do?
  use (c)hanged version, leave (d)eleted, or leave (u)nresolved? c
  rebasing 6:9455ee510502 "dev: merge default"
  saved backup bundle to $TESTTMP/ancestor-merge/.hg/strip-backup/1d1a643d390e-43e9e04b-rebase.hg
  $ hg tglog
  o  6: fbc098e72227 'dev: merge default'
  |
  o  5: eda7b7f46f5d 'dev: merge default'
  |
  o  4: 3e075b1c0a40 'dev: f-dev stuff'
  |
  @  3: e08089805d82 'default: f-other stuff'
  |
  o  2: 462860db70a1 'default: remove f-default'
  |
  o  1: f157ecfd2b6b 'default: f-default stuff'
  |
  o  0: e90e8eb90b6f 'default: create f-default'
  
Grafty cherry picking rebasing:

  $ cd ../ancestor-merge-2

  $ hg phase -fdr0:
  $ hg rebase -r 'children(only(dev,default))' -d default --config ui.interactive=True << EOF
  > c
  > EOF
  rebasing 2:ec2c14fb2984 "dev: f-dev stuff"
  rebasing 4:4b019212aaf6 "dev: merge default"
  file 'f-default' was deleted in local [dest] but was modified in other [source].
  What do you want to do?
  use (c)hanged version, leave (d)eleted, or leave (u)nresolved? c
  rebasing 6:9455ee510502 "dev: merge default"
  saved backup bundle to $TESTTMP/ancestor-merge-2/.hg/strip-backup/ec2c14fb2984-62d0b222-rebase.hg
  $ hg tglog
  o  7: fbc098e72227 'dev: merge default'
  |
  o  6: eda7b7f46f5d 'dev: merge default'
  |
  o  5: 3e075b1c0a40 'dev: f-dev stuff'
  |
  o  4: e08089805d82 'default: f-other stuff'
  |
  o  3: 462860db70a1 'default: remove f-default'
  |
  o  2: f157ecfd2b6b 'default: f-default stuff'
  |
  | o  1: 1d1a643d390e 'dev: create branch' dev
  |/
  o  0: e90e8eb90b6f 'default: create f-default'
  
  $ cd ..


Test order of parents of rebased merged with un-rebased changes as p1.

  $ hg init parentorder
  $ cd parentorder
  $ touch f
  $ hg ci -Aqm common
  $ touch change
  $ hg ci -Aqm change
  $ touch target
  $ hg ci -Aqm target
  $ hg up -qr 0
  $ touch outside
  $ hg ci -Aqm outside
  $ hg merge -qr 1
  $ hg ci -m 'merge p1 3=outside p2 1=ancestor'
  $ hg par
  changeset:   4:6990226659be
  tag:         tip
  parent:      3:f59da8fc0fcf
  parent:      1:dd40c13f7a6f
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     merge p1 3=outside p2 1=ancestor
  
  $ hg up -qr 1
  $ hg merge -qr 3
  $ hg ci -qm 'merge p1 1=ancestor p2 3=outside'
  $ hg par
  changeset:   5:a57575f79074
  tag:         tip
  parent:      1:dd40c13f7a6f
  parent:      3:f59da8fc0fcf
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     merge p1 1=ancestor p2 3=outside
  
  $ hg tglog
  @    5: a57575f79074 'merge p1 1=ancestor p2 3=outside'
  |\
  +---o  4: 6990226659be 'merge p1 3=outside p2 1=ancestor'
  | |/
  | o  3: f59da8fc0fcf 'outside'
  | |
  +---o  2: a60552eb93fb 'target'
  | |
  o |  1: dd40c13f7a6f 'change'
  |/
  o  0: 02f0f58d5300 'common'
  
  $ hg rebase -r 4 -d 2
  rebasing 4:6990226659be "merge p1 3=outside p2 1=ancestor"
  saved backup bundle to $TESTTMP/parentorder/.hg/strip-backup/6990226659be-4d67a0d3-rebase.hg
  $ hg tip
  changeset:   5:cca50676b1c5
  tag:         tip
  parent:      2:a60552eb93fb
  parent:      3:f59da8fc0fcf
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     merge p1 3=outside p2 1=ancestor
  
  $ hg rebase -r 4 -d 2
  rebasing 4:a57575f79074 "merge p1 1=ancestor p2 3=outside"
  saved backup bundle to $TESTTMP/parentorder/.hg/strip-backup/a57575f79074-385426e5-rebase.hg
  $ hg tip
  changeset:   5:f9daf77ffe76
  tag:         tip
  parent:      2:a60552eb93fb
  parent:      3:f59da8fc0fcf
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     merge p1 1=ancestor p2 3=outside
  
  $ hg tglog
  @    5: f9daf77ffe76 'merge p1 1=ancestor p2 3=outside'
  |\
  +---o  4: cca50676b1c5 'merge p1 3=outside p2 1=ancestor'
  | |/
  | o  3: f59da8fc0fcf 'outside'
  | |
  o |  2: a60552eb93fb 'target'
  | |
  o |  1: dd40c13f7a6f 'change'
  |/
  o  0: 02f0f58d5300 'common'
  
rebase of merge of ancestors

  $ hg up -qr 2
  $ hg merge -qr 3
  $ echo 'other change while merging future "rebase ancestors"' > other
  $ hg ci -Aqm 'merge rebase ancestors'
  $ hg rebase -d 5 -v
  rebasing 6:4c5f12f25ebe "merge rebase ancestors" (tip)
  resolving manifests
  removing other
  note: merging f9daf77ffe76+ and 4c5f12f25ebe using bids from ancestors a60552eb93fb and f59da8fc0fcf
  
  calculating bids for ancestor a60552eb93fb
  resolving manifests
  
  calculating bids for ancestor f59da8fc0fcf
  resolving manifests
  
  auction for merging merge bids
   other: consensus for g
  end of auction
  
  getting other
  committing files:
  other
  committing manifest
  committing changelog
  rebase merging completed
  1 changesets found
  uncompressed size of bundle content:
       199 (changelog)
       216 (manifests)
       182  other
  saved backup bundle to $TESTTMP/parentorder/.hg/strip-backup/4c5f12f25ebe-f46990e5-rebase.hg
  1 changesets found
  uncompressed size of bundle content:
       254 (changelog)
       167 (manifests)
       182  other
  adding branch
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  rebase completed
  $ hg tglog
  @  6: 113755df812b 'merge rebase ancestors'
  |
  o    5: f9daf77ffe76 'merge p1 1=ancestor p2 3=outside'
  |\
  +---o  4: cca50676b1c5 'merge p1 3=outside p2 1=ancestor'
  | |/
  | o  3: f59da8fc0fcf 'outside'
  | |
  o |  2: a60552eb93fb 'target'
  | |
  o |  1: dd40c13f7a6f 'change'
  |/
  o  0: 02f0f58d5300 'common'
  
Due to the limitation of 3-way merge algorithm (1 merge base), rebasing a merge
may include unwanted content:

  $ hg init $TESTTMP/dual-merge-base1
  $ cd $TESTTMP/dual-merge-base1
  $ hg debugdrawdag <<'EOS'
  >   F
  >  /|
  > D E
  > | |
  > B C
  > |/
  > A Z
  > |/
  > R
  > EOS
  $ hg rebase -r D+E+F -d Z
  rebasing 5:5f2c926dfecf "D" (D)
  rebasing 6:b296604d9846 "E" (E)
  rebasing 7:caa9781e507d "F" (F tip)
  abort: rebasing 7:caa9781e507d will include unwanted changes from 4:d6003a550c2c or 3:c1e6b162678d
  [255]

The warning does not get printed if there is no unwanted change detected:

  $ hg init $TESTTMP/dual-merge-base2
  $ cd $TESTTMP/dual-merge-base2
  $ hg debugdrawdag <<'EOS'
  >   D
  >  /|
  > B C
  > |/
  > A Z
  > |/
  > R
  > EOS
  $ hg rebase -r B+C+D -d Z
  rebasing 3:c1e6b162678d "B" (B)
  rebasing 4:d6003a550c2c "C" (C)
  rebasing 5:c8f78076273e "D" (D tip)
  saved backup bundle to $TESTTMP/dual-merge-base2/.hg/strip-backup/d6003a550c2c-6f1424b6-rebase.hg
  $ hg manifest -r 'desc(D)'
  B
  C
  R
  Z

The merge base could be different from old p1 (changed parent becomes new p1):

  $ hg init $TESTTMP/chosen-merge-base1
  $ cd $TESTTMP/chosen-merge-base1
  $ hg debugdrawdag <<'EOS'
  >   F
  >  /|
  > D E
  > | |
  > B C Z
  > EOS
  $ hg rebase -r D+F -d Z
  rebasing 3:004dc1679908 "D" (D)
  rebasing 5:4be4cbf6f206 "F" (F tip)
  saved backup bundle to $TESTTMP/chosen-merge-base1/.hg/strip-backup/004dc1679908-06a66a3c-rebase.hg
  $ hg manifest -r 'desc(F)'
  C
  D
  E
  Z
  $ hg log -r `hg log -r 'desc(F)' -T '{p1node}'` -T '{desc}\n'
  D

  $ hg init $TESTTMP/chosen-merge-base2
  $ cd $TESTTMP/chosen-merge-base2
  $ hg debugdrawdag <<'EOS'
  >   F
  >  /|
  > D E
  > | |
  > B C Z
  > EOS
  $ hg rebase -r E+F -d Z
  rebasing 4:974e4943c210 "E" (E)
  rebasing 5:4be4cbf6f206 "F" (F tip)
  saved backup bundle to $TESTTMP/chosen-merge-base2/.hg/strip-backup/974e4943c210-b2874da5-rebase.hg
  $ hg manifest -r 'desc(F)'
  B
  D
  E
  Z
  $ hg log -r `hg log -r 'desc(F)' -T '{p1node}'` -T '{desc}\n'
  E