view tests/test-histedit-arguments.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

Test argument handling and various data parsing
==================================================


Enable extensions used by this test.
  $ cat >>$HGRCPATH <<EOF
  > [extensions]
  > histedit=
  > EOF

Repo setup.
  $ hg init foo
  $ cd foo
  $ echo alpha >> alpha
  $ hg addr
  adding alpha
  $ hg ci -m one
  $ echo alpha >> alpha
  $ hg ci -m two
  $ echo alpha >> alpha
  $ hg ci -m three
  $ echo alpha >> alpha
  $ hg ci -m four
  $ echo alpha >> alpha
  $ hg ci -m five

  $ hg log --style compact --graph
  @  4[tip]   08d98a8350f3   1970-01-01 00:00 +0000   test
  |    five
  |
  o  3   c8e68270e35a   1970-01-01 00:00 +0000   test
  |    four
  |
  o  2   eb57da33312f   1970-01-01 00:00 +0000   test
  |    three
  |
  o  1   579e40513370   1970-01-01 00:00 +0000   test
  |    two
  |
  o  0   6058cbb6cfd7   1970-01-01 00:00 +0000   test
       one
  

histedit --continue/--abort with no existing state
--------------------------------------------------

  $ hg histedit --continue
  abort: no histedit in progress
  [255]
  $ hg histedit --abort
  abort: no histedit in progress
  [255]

Run a dummy edit to make sure we get tip^^ correctly via revsingle.
--------------------------------------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^"
  pick eb57da33312f 2 three
  pick c8e68270e35a 3 four
  pick 08d98a8350f3 4 five
  
  # Edit history between eb57da33312f and 08d98a8350f3
  #
  # Commits are listed from least to most recent
  #
  # You can reorder changesets by reordering the lines
  #
  # Commands:
  #
  #  e, edit = use commit, but stop for amending
  #  m, mess = edit commit message without changing commit content
  #  p, pick = use commit
  #  b, base = checkout changeset and apply further changesets from there
  #  d, drop = remove commit from history
  #  f, fold = use commit, but combine it with the one above
  #  r, roll = like fold, but discard this commit's description and date
  #

Run on a revision not ancestors of the current working directory.
--------------------------------------------------------------------

  $ hg up 2
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg histedit -r 4
  abort: 08d98a8350f3 is not an ancestor of working directory
  [255]
  $ hg up --quiet


Test that we pick the minimum of a revrange
---------------------------------------

  $ HGEDITOR=cat hg histedit '2::' --commands - << EOF
  > pick eb57da33312f 2 three
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  $ hg up --quiet

  $ HGEDITOR=cat hg histedit 'tip:2' --commands - << EOF
  > pick eb57da33312f 2 three
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  $ hg up --quiet

Test config specified default
-----------------------------

  $ HGEDITOR=cat hg histedit --config "histedit.defaultrev=only(.) - ::eb57da33312f" --commands - << EOF
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF

Run on a revision not descendants of the initial parent
--------------------------------------------------------------------

Test the message shown for inconsistent histedit state, which may be
created (and forgotten) by Mercurial earlier than 2.7. This emulates
Mercurial earlier than 2.7 by renaming ".hg/histedit-state"
temporarily.

  $ hg log -G -T '{rev} {shortest(node)} {desc}\n' -r 2::
  @  4 08d9 five
  |
  o  3 c8e6 four
  |
  o  2 eb57 three
  |
  ~
  $ HGEDITOR=cat hg histedit -r 4 --commands - << EOF
  > edit 08d98a8350f3 4 five
  > EOF
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  reverting alpha
  Editing (08d98a8350f3), you may commit or record as needed now.
  (hg histedit --continue to resume)
  [1]

  $ hg graft --continue
  abort: no graft in progress
  (continue: hg histedit --continue)
  [255]

  $ mv .hg/histedit-state .hg/histedit-state.back
  $ hg update --quiet --clean 2
  $ echo alpha >> alpha
  $ mv .hg/histedit-state.back .hg/histedit-state

  $ hg histedit --continue
  saved backup bundle to $TESTTMP/foo/.hg/strip-backup/08d98a8350f3-02594089-histedit.hg
  $ hg log -G -T '{rev} {shortest(node)} {desc}\n' -r 2::
  @  4 f5ed five
  |
  | o  3 c8e6 four
  |/
  o  2 eb57 three
  |
  ~

  $ hg unbundle -q $TESTTMP/foo/.hg/strip-backup/08d98a8350f3-02594089-histedit.hg
  $ hg strip -q -r f5ed --config extensions.strip=
  $ hg up -q 08d98a8350f3

Test that missing revisions are detected
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick eb57da33312f 2 three
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: missing rules for changeset c8e68270e35a
  (use "drop c8e68270e35a" to discard, see also: 'hg help -e histedit.config')
  [255]

Test that extra revisions are detected
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick 6058cbb6cfd7 0 one
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: pick "6058cbb6cfd7" changeset was not a candidate
  (only use listed changesets)
  [255]

Test malformed line
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pickeb57da33312f2three
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: malformed line "pickeb57da33312f2three"
  [255]

Test unknown changeset
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick 0123456789ab 2 three
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: unknown changeset 0123456789ab listed
  [255]

Test unknown command
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > coin eb57da33312f 2 three
  > pick c8e68270e35a 3 four
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: unknown action "coin"
  [255]

Test duplicated changeset
---------------------------------------

So one is missing and one appear twice.

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick eb57da33312f 2 three
  > pick eb57da33312f 2 three
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: duplicated command for changeset eb57da33312f
  [255]

Test bogus rev
---------------------------------------

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick eb57da33312f 2 three
  > pick 0u98
  > pick 08d98a8350f3 4 five
  > EOF
  hg: parse error: invalid changeset 0u98
  [255]

Test short version of command
---------------------------------------

Note: we use varying amounts of white space between command name and changeset
short hash. This tests issue3893.

  $ HGEDITOR=cat hg histedit "tip^^" --commands - << EOF
  > pick eb57da33312f 2 three
  > p    c8e68270e35a 3 four
  > f 08d98a8350f3 4 five
  > EOF
  four
  ***
  five
  
  
  
  HG: Enter commit message.  Lines beginning with 'HG:' are removed.
  HG: Leave message empty to abort commit.
  HG: --
  HG: user: test
  HG: branch 'default'
  HG: changed alpha
  saved backup bundle to $TESTTMP/foo/.hg/strip-backup/c8e68270e35a-63d8b8d8-histedit.hg

  $ hg update -q 2
  $ echo x > x
  $ hg add x
  $ hg commit -m'x' x
  created new head
  $ hg histedit -r 'heads(all())'
  abort: The specified revisions must have exactly one common root
  [255]

Test that trimming description using multi-byte characters
--------------------------------------------------------------------

  $ "$PYTHON" <<EOF
  > fp = open('logfile', 'wb')
  > fp.write(b'12345678901234567890123456789012345678901234567890' +
  >          b'12345') # there are 5 more columns for 80 columns
  > 
  > # 2 x 4 = 8 columns, but 3 x 4 = 12 bytes
  > fp.write(u'\u3042\u3044\u3046\u3048'.encode('utf-8'))
  > 
  > fp.close()
  > EOF
  $ echo xx >> x
  $ hg --encoding utf-8 commit --logfile logfile

  $ HGEDITOR=cat hg --encoding utf-8 histedit tip
  pick 3d3ea1f3a10b 5 1234567890123456789012345678901234567890123456789012345\xe3\x81\x82... (esc)
  
  # Edit history between 3d3ea1f3a10b and 3d3ea1f3a10b
  #
  # Commits are listed from least to most recent
  #
  # You can reorder changesets by reordering the lines
  #
  # Commands:
  #
  #  e, edit = use commit, but stop for amending
  #  m, mess = edit commit message without changing commit content
  #  p, pick = use commit
  #  b, base = checkout changeset and apply further changesets from there
  #  d, drop = remove commit from history
  #  f, fold = use commit, but combine it with the one above
  #  r, roll = like fold, but discard this commit's description and date
  #

Test --continue with --keep

  $ hg strip -q -r . --config extensions.strip=
  $ hg histedit '.^' -q --keep --commands - << EOF
  > edit eb57da33312f 2 three
  > pick f3cfcca30c44 4 x
  > EOF
  Editing (eb57da33312f), you may commit or record as needed now.
  (hg histedit --continue to resume)
  [1]
  $ echo edit >> alpha
  $ hg histedit -q --continue
  $ hg log -G -T '{rev}:{node|short} {desc}'
  @  6:8fda0c726bf2 x
  |
  o  5:63379946892c three
  |
  | o  4:f3cfcca30c44 x
  | |
  | | o  3:2a30f3cfee78 four
  | |/   ***
  | |    five
  | o  2:eb57da33312f three
  |/
  o  1:579e40513370 two
  |
  o  0:6058cbb6cfd7 one
  

Test that abort fails gracefully on exception
----------------------------------------------
  $ hg histedit . -q --commands - << EOF
  > edit 8fda0c726bf2 6 x
  > EOF
  Editing (8fda0c726bf2), you may commit or record as needed now.
  (hg histedit --continue to resume)
  [1]
Corrupt histedit state file
  $ sed 's/8fda0c726bf2/123456789012/' .hg/histedit-state > ../corrupt-histedit
  $ mv ../corrupt-histedit .hg/histedit-state
  $ hg histedit --abort
  warning: encountered an exception during histedit --abort; the repository may not have been completely cleaned up
  abort: $TESTTMP/foo/.hg/strip-backup/*-histedit.hg: $ENOENT$ (glob) (windows !)
  abort: $ENOENT$: $TESTTMP/foo/.hg/strip-backup/*-histedit.hg (glob) (no-windows !)
  [255]
Histedit state has been exited
  $ hg summary -q
  parent: 5:63379946892c 
  commit: 1 added, 1 unknown (new branch head)
  update: 4 new changesets (update)

  $ cd ..

Set up default base revision tests

  $ hg init defaultbase
  $ cd defaultbase
  $ touch foo
  $ hg -q commit -A -m root
  $ echo 1 > foo
  $ hg commit -m 'public 1'
  $ hg phase --force --public -r .
  $ echo 2 > foo
  $ hg commit -m 'draft after public'
  $ hg -q up -r 1
  $ echo 3 > foo
  $ hg commit -m 'head 1 public'
  created new head
  $ hg phase --force --public -r .
  $ echo 4 > foo
  $ hg commit -m 'head 1 draft 1'
  $ echo 5 > foo
  $ hg commit -m 'head 1 draft 2'
  $ hg -q up -r 2
  $ echo 6 > foo
  $ hg commit -m 'head 2 commit 1'
  $ echo 7 > foo
  $ hg commit -m 'head 2 commit 2'
  $ hg -q up -r 2
  $ echo 8 > foo
  $ hg commit -m 'head 3'
  created new head
  $ hg -q up -r 2
  $ echo 9 > foo
  $ hg commit -m 'head 4'
  created new head
  $ hg merge --tool :local -r 8
  0 files updated, 1 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)
  $ hg commit -m 'merge head 3 into head 4'
  $ echo 11 > foo
  $ hg commit -m 'commit 1 after merge'
  $ echo 12 > foo
  $ hg commit -m 'commit 2 after merge'

  $ hg log -G -T '{rev}:{node|short} {phase} {desc}\n'
  @  12:8cde254db839 draft commit 2 after merge
  |
  o  11:6f2f0241f119 draft commit 1 after merge
  |
  o    10:90506cc76b00 draft merge head 3 into head 4
  |\
  | o  9:f8607a373a97 draft head 4
  | |
  o |  8:0da92be05148 draft head 3
  |/
  | o  7:4c35cdf97d5e draft head 2 commit 2
  | |
  | o  6:931820154288 draft head 2 commit 1
  |/
  | o  5:8cdc02b9bc63 draft head 1 draft 2
  | |
  | o  4:463b8c0d2973 draft head 1 draft 1
  | |
  | o  3:23a0c4eefcbf public head 1 public
  | |
  o |  2:4117331c3abb draft draft after public
  |/
  o  1:4426d359ea59 public public 1
  |
  o  0:54136a8ddf32 public root
  

Default base revision should stop at public changesets

  $ hg -q up 8cdc02b9bc63
  $ hg histedit --commands - <<EOF
  > pick 463b8c0d2973
  > pick 8cdc02b9bc63
  > EOF

Default base revision should stop at branchpoint

  $ hg -q up 4c35cdf97d5e
  $ hg histedit --commands - <<EOF
  > pick 931820154288
  > pick 4c35cdf97d5e
  > EOF

Default base revision should stop at merge commit

  $ hg -q up 8cde254db839
  $ hg histedit --commands - <<EOF
  > pick 6f2f0241f119
  > pick 8cde254db839
  > EOF

commit --amend should abort if histedit is in progress
(issue4800) and markers are not being created.
Eventually, histedit could perhaps look at `source` extra,
in which case this test should be revisited.

  $ hg -q up 8cde254db839
  $ hg histedit 6f2f0241f119 --commands - <<EOF
  > pick 8cde254db839
  > edit 6f2f0241f119
  > EOF
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  merging foo
  warning: conflicts while merging foo! (edit, then use 'hg resolve --mark')
  Fix up the change (pick 8cde254db839)
  (hg histedit --continue to resume)
  [1]
  $ hg resolve -m --all
  (no more unresolved files)
  continue: hg histedit --continue
  $ hg histedit --cont
  merging foo
  warning: conflicts while merging foo! (edit, then use 'hg resolve --mark')
  Editing (6f2f0241f119), you may commit or record as needed now.
  (hg histedit --continue to resume)
  [1]
  $ hg resolve -m --all
  (no more unresolved files)
  continue: hg histedit --continue
  $ hg commit --amend -m 'reject this fold'
  abort: histedit in progress
  (use 'hg histedit --continue' or 'hg histedit --abort')
  [255]

With markers enabled, histedit does not get confused, and
amend should not be blocked by the ongoing histedit.

  $ cat >>$HGRCPATH <<EOF
  > [experimental]
  > evolution.createmarkers=True
  > evolution.allowunstable=True
  > EOF
  $ hg commit --amend -m 'allow this fold'
  $ hg histedit --continue

  $ cd ..

Test autoverb feature

  $ hg init autoverb
  $ cd autoverb
  $ echo alpha >> alpha
  $ hg ci -qAm one
  $ echo alpha >> alpha
  $ hg ci -qm two
  $ echo beta >> beta
  $ hg ci -qAm "roll! one"

  $ hg log --style compact --graph
  @  2[tip]   4f34d0f8b5fa   1970-01-01 00:00 +0000   test
  |    roll! one
  |
  o  1   579e40513370   1970-01-01 00:00 +0000   test
  |    two
  |
  o  0   6058cbb6cfd7   1970-01-01 00:00 +0000   test
       one
  

Check that 'roll' is selected by default

  $ HGEDITOR=cat hg histedit 0 --config experimental.histedit.autoverb=True
  pick 6058cbb6cfd7 0 one
  roll 4f34d0f8b5fa 2 roll! one
  pick 579e40513370 1 two
  
  # Edit history between 6058cbb6cfd7 and 4f34d0f8b5fa
  #
  # Commits are listed from least to most recent
  #
  # You can reorder changesets by reordering the lines
  #
  # Commands:
  #
  #  e, edit = use commit, but stop for amending
  #  m, mess = edit commit message without changing commit content
  #  p, pick = use commit
  #  b, base = checkout changeset and apply further changesets from there
  #  d, drop = remove commit from history
  #  f, fold = use commit, but combine it with the one above
  #  r, roll = like fold, but discard this commit's description and date
  #

  $ cd ..

Check that histedit's commands accept revsets
  $ hg init bar
  $ cd bar
  $ echo w >> a
  $ hg ci -qAm "adds a"
  $ echo x >> b
  $ hg ci -qAm "adds b"
  $ echo y >> c
  $ hg ci -qAm "adds c"
  $ echo z >> d
  $ hg ci -qAm "adds d"
  $ hg log -G -T '{rev} {desc}\n'
  @  3 adds d
  |
  o  2 adds c
  |
  o  1 adds b
  |
  o  0 adds a
  
  $ HGEDITOR=cat hg histedit "2" --commands - << EOF
  > base -4 adds c
  > pick 2 adds c
  > pick tip adds d
  > EOF
  $ hg log -G -T '{rev} {desc}\n'
  @  5 adds d
  |
  o  4 adds c
  |
  | o  1 adds b
  |/
  o  0 adds a