view tests/test-status.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

  $ hg init repo1
  $ cd repo1
  $ mkdir a b a/1 b/1 b/2
  $ touch in_root a/in_a b/in_b a/1/in_a_1 b/1/in_b_1 b/2/in_b_2

hg status in repo root:

  $ hg status
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root

hg status . in repo root:

  $ hg status .
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root

  $ hg status --cwd a
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root
  $ hg status --cwd a .
  ? 1/in_a_1
  ? in_a
  $ hg status --cwd a ..
  ? 1/in_a_1
  ? in_a
  ? ../b/1/in_b_1
  ? ../b/2/in_b_2
  ? ../b/in_b
  ? ../in_root

  $ hg status --cwd b
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root
  $ hg status --cwd b .
  ? 1/in_b_1
  ? 2/in_b_2
  ? in_b
  $ hg status --cwd b ..
  ? ../a/1/in_a_1
  ? ../a/in_a
  ? 1/in_b_1
  ? 2/in_b_2
  ? in_b
  ? ../in_root

  $ hg status --cwd a/1
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root
  $ hg status --cwd a/1 .
  ? in_a_1
  $ hg status --cwd a/1 ..
  ? in_a_1
  ? ../in_a

  $ hg status --cwd b/1
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root
  $ hg status --cwd b/1 .
  ? in_b_1
  $ hg status --cwd b/1 ..
  ? in_b_1
  ? ../2/in_b_2
  ? ../in_b

  $ hg status --cwd b/2
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root
  $ hg status --cwd b/2 .
  ? in_b_2
  $ hg status --cwd b/2 ..
  ? ../1/in_b_1
  ? in_b_2
  ? ../in_b

combining patterns with root and patterns without a root works

  $ hg st a/in_a re:.*b$
  ? a/in_a
  ? b/in_b

tweaking defaults works
  $ hg status --cwd a --config ui.tweakdefaults=yes
  ? 1/in_a_1
  ? in_a
  ? ../b/1/in_b_1
  ? ../b/2/in_b_2
  ? ../b/in_b
  ? ../in_root
  $ HGPLAIN=1 hg status --cwd a --config ui.tweakdefaults=yes
  ? a/1/in_a_1 (glob)
  ? a/in_a (glob)
  ? b/1/in_b_1 (glob)
  ? b/2/in_b_2 (glob)
  ? b/in_b (glob)
  ? in_root
  $ HGPLAINEXCEPT=tweakdefaults hg status --cwd a --config ui.tweakdefaults=yes
  ? 1/in_a_1
  ? in_a
  ? ../b/1/in_b_1
  ? ../b/2/in_b_2
  ? ../b/in_b
  ? ../in_root (glob)

relative paths can be requested

  $ cat >> $HGRCPATH <<EOF
  > [commands]
  > status.relative = True
  > EOF
  $ hg status --cwd a
  ? 1/in_a_1
  ? in_a
  ? ../b/1/in_b_1
  ? ../b/2/in_b_2
  ? ../b/in_b
  ? ../in_root
  $ HGPLAIN=1 hg status --cwd a
  ? a/1/in_a_1 (glob)
  ? a/in_a (glob)
  ? b/1/in_b_1 (glob)
  ? b/2/in_b_2 (glob)
  ? b/in_b (glob)
  ? in_root

if relative paths are explicitly off, tweakdefaults doesn't change it
  $ cat >> $HGRCPATH <<EOF
  > [commands]
  > status.relative = False
  > EOF
  $ hg status --cwd a --config ui.tweakdefaults=yes
  ? a/1/in_a_1
  ? a/in_a
  ? b/1/in_b_1
  ? b/2/in_b_2
  ? b/in_b
  ? in_root

  $ cd ..

  $ hg init repo2
  $ cd repo2
  $ touch modified removed deleted ignored
  $ echo "^ignored$" > .hgignore
  $ hg ci -A -m 'initial checkin'
  adding .hgignore
  adding deleted
  adding modified
  adding removed
  $ touch modified added unknown ignored
  $ hg add added
  $ hg remove removed
  $ rm deleted

hg status:

  $ hg status
  A added
  R removed
  ! deleted
  ? unknown

hg status modified added removed deleted unknown never-existed ignored:

  $ hg status modified added removed deleted unknown never-existed ignored
  never-existed: * (glob)
  A added
  R removed
  ! deleted
  ? unknown

  $ hg copy modified copied

hg status -C:

  $ hg status -C
  A added
  A copied
    modified
  R removed
  ! deleted
  ? unknown

hg status -A:

  $ hg status -A
  A added
  A copied
    modified
  R removed
  ! deleted
  ? unknown
  I ignored
  C .hgignore
  C modified

  $ hg status -A -T '{status} {path} {node|shortest}\n'
  A added ffff
  A copied ffff
  R removed ffff
  ! deleted ffff
  ? unknown ffff
  I ignored ffff
  C .hgignore ffff
  C modified ffff

  $ hg status -A -Tjson
  [
   {
    "path": "added",
    "status": "A"
   },
   {
    "path": "copied",
    "source": "modified",
    "status": "A"
   },
   {
    "path": "removed",
    "status": "R"
   },
   {
    "path": "deleted",
    "status": "!"
   },
   {
    "path": "unknown",
    "status": "?"
   },
   {
    "path": "ignored",
    "status": "I"
   },
   {
    "path": ".hgignore",
    "status": "C"
   },
   {
    "path": "modified",
    "status": "C"
   }
  ]

  $ hg status -A -Tpickle > pickle
  >>> from __future__ import print_function
  >>> import pickle
  >>> data = sorted((x[b'status'].decode(), x[b'path'].decode()) for x in pickle.load(open("pickle", r"rb")))
  >>> for s, p in data: print("%s %s" % (s, p))
  ! deleted
  ? pickle
  ? unknown
  A added
  A copied
  C .hgignore
  C modified
  I ignored
  R removed
  $ rm pickle

  $ echo "^ignoreddir$" > .hgignore
  $ mkdir ignoreddir
  $ touch ignoreddir/file

Test templater support:

  $ hg status -AT "[{status}]\t{if(source, '{source} -> ')}{path}\n"
  [M]	.hgignore
  [A]	added
  [A]	modified -> copied
  [R]	removed
  [!]	deleted
  [?]	ignored
  [?]	unknown
  [I]	ignoreddir/file
  [C]	modified
  $ hg status -AT default
  M .hgignore
  A added
  A copied
    modified
  R removed
  ! deleted
  ? ignored
  ? unknown
  I ignoreddir/file
  C modified
  $ hg status -T compact
  abort: "status" not in template map
  [255]

hg status ignoreddir/file:

  $ hg status ignoreddir/file

hg status -i ignoreddir/file:

  $ hg status -i ignoreddir/file
  I ignoreddir/file
  $ cd ..

Check 'status -q' and some combinations

  $ hg init repo3
  $ cd repo3
  $ touch modified removed deleted ignored
  $ echo "^ignored$" > .hgignore
  $ hg commit -A -m 'initial checkin'
  adding .hgignore
  adding deleted
  adding modified
  adding removed
  $ touch added unknown ignored
  $ hg add added
  $ echo "test" >> modified
  $ hg remove removed
  $ rm deleted
  $ hg copy modified copied

Specify working directory revision explicitly, that should be the same as
"hg status"

  $ hg status --change "wdir()"
  M modified
  A added
  A copied
  R removed
  ! deleted
  ? unknown

Run status with 2 different flags.
Check if result is the same or different.
If result is not as expected, raise error

  $ assert() {
  >     hg status $1 > ../a
  >     hg status $2 > ../b
  >     if diff ../a ../b > /dev/null; then
  >         out=0
  >     else
  >         out=1
  >     fi
  >     if [ $3 -eq 0 ]; then
  >         df="same"
  >     else
  >         df="different"
  >     fi
  >     if [ $out -ne $3 ]; then
  >         echo "Error on $1 and $2, should be $df."
  >     fi
  > }

Assert flag1 flag2 [0-same | 1-different]

  $ assert "-q" "-mard"      0
  $ assert "-A" "-marduicC"  0
  $ assert "-qA" "-mardcC"   0
  $ assert "-qAui" "-A"      0
  $ assert "-qAu" "-marducC" 0
  $ assert "-qAi" "-mardicC" 0
  $ assert "-qu" "-u"        0
  $ assert "-q" "-u"         1
  $ assert "-m" "-a"         1
  $ assert "-r" "-d"         1
  $ cd ..

  $ hg init repo4
  $ cd repo4
  $ touch modified removed deleted
  $ hg ci -q -A -m 'initial checkin'
  $ touch added unknown
  $ hg add added
  $ hg remove removed
  $ rm deleted
  $ echo x > modified
  $ hg copy modified copied
  $ hg ci -m 'test checkin' -d "1000001 0"
  $ rm *
  $ touch unrelated
  $ hg ci -q -A -m 'unrelated checkin' -d "1000002 0"

hg status --change 1:

  $ hg status --change 1
  M modified
  A added
  A copied
  R removed

hg status --change 1 unrelated:

  $ hg status --change 1 unrelated

hg status -C --change 1 added modified copied removed deleted:

  $ hg status -C --change 1 added modified copied removed deleted
  M modified
  A added
  A copied
    modified
  R removed

hg status -A --change 1 and revset:

  $ hg status -A --change '1|1'
  M modified
  A added
  A copied
    modified
  R removed
  C deleted

  $ cd ..

hg status with --rev and reverted changes:

  $ hg init reverted-changes-repo
  $ cd reverted-changes-repo
  $ echo a > file
  $ hg add file
  $ hg ci -m a
  $ echo b > file
  $ hg ci -m b

reverted file should appear clean

  $ hg revert -r 0 .
  reverting file
  $ hg status -A --rev 0
  C file

#if execbit
reverted file with changed flag should appear modified

  $ chmod +x file
  $ hg status -A --rev 0
  M file

  $ hg revert -r 0 .
  reverting file

reverted and committed file with changed flag should appear modified

  $ hg co -C .
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ chmod +x file
  $ hg ci -m 'change flag'
  $ hg status -A --rev 1 --rev 2
  M file
  $ hg diff -r 1 -r 2

#endif

  $ cd ..

hg status of binary file starting with '\1\n', a separator for metadata:

  $ hg init repo5
  $ cd repo5
  >>> open("010a", r"wb").write(b"\1\nfoo") and None
  $ hg ci -q -A -m 'initial checkin'
  $ hg status -A
  C 010a

  >>> open("010a", r"wb").write(b"\1\nbar") and None
  $ hg status -A
  M 010a
  $ hg ci -q -m 'modify 010a'
  $ hg status -A --rev 0:1
  M 010a

  $ touch empty
  $ hg ci -q -A -m 'add another file'
  $ hg status -A --rev 1:2 010a
  C 010a

  $ cd ..

test "hg status" with "directory pattern" which matches against files
only known on target revision.

  $ hg init repo6
  $ cd repo6

  $ echo a > a.txt
  $ hg add a.txt
  $ hg commit -m '#0'
  $ mkdir -p 1/2/3/4/5
  $ echo b > 1/2/3/4/5/b.txt
  $ hg add 1/2/3/4/5/b.txt
  $ hg commit -m '#1'

  $ hg update -C 0 > /dev/null
  $ hg status -A
  C a.txt

the directory matching against specified pattern should be removed,
because directory existence prevents 'dirstate.walk()' from showing
warning message about such pattern.

  $ test ! -d 1
  $ hg status -A --rev 1 1/2/3/4/5/b.txt
  R 1/2/3/4/5/b.txt
  $ hg status -A --rev 1 1/2/3/4/5
  R 1/2/3/4/5/b.txt
  $ hg status -A --rev 1 1/2/3
  R 1/2/3/4/5/b.txt
  $ hg status -A --rev 1 1
  R 1/2/3/4/5/b.txt

  $ hg status --config ui.formatdebug=True --rev 1 1
  status = [
      {
          'path': '1/2/3/4/5/b.txt',
          'status': 'R'
      },
  ]

#if windows
  $ hg --config ui.slash=false status -A --rev 1 1
  R 1\2\3\4\5\b.txt
#endif

  $ cd ..

Status after move overwriting a file (issue4458)
=================================================


  $ hg init issue4458
  $ cd issue4458
  $ echo a > a
  $ echo b > b
  $ hg commit -Am base
  adding a
  adding b


with --force

  $ hg mv b --force a
  $ hg st --copies
  M a
    b
  R b
  $ hg revert --all
  reverting a
  undeleting b
  $ rm *.orig

without force

  $ hg rm a
  $ hg st --copies
  R a
  $ hg mv b a
  $ hg st --copies
  M a
    b
  R b

using ui.statuscopies setting
  $ hg st --config ui.statuscopies=true
  M a
    b
  R b
  $ hg st --config ui.statuscopies=false
  M a
  R b
  $ hg st --config ui.tweakdefaults=yes
  M a
    b
  R b

using log status template (issue5155)
  $ hg log -Tstatus -r 'wdir()' -C
  changeset:   2147483647:ffffffffffff
  parent:      0:8c55c58b4c0e
  user:        test
  date:        * (glob)
  files:
  M a
    b
  R b
  

Other "bug" highlight, the revision status does not report the copy information.
This is buggy behavior.

  $ hg commit -m 'blah'
  $ hg st --copies --change .
  M a
  R b

using log status template, the copy information is displayed correctly.
  $ hg log -Tstatus -r. -C
  changeset:   1:6685fde43d21
  tag:         tip
  user:        test
  date:        * (glob)
  summary:     blah
  files:
  M a
    b
  R b
  

  $ cd ..