view tests/test-mq-subrepo.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 <<EOF >> $HGRCPATH
  > [ui]
  > commitsubrepos = Yes
  > [extensions]
  > mq =
  > record =
  > [diff]
  > nodates = 1
  > EOF

  $ stdin=`pwd`/stdin.tmp

fn to create new repository w/dirty subrepo, and cd into it
  $ mkrepo() {
  >     hg init $1
  >     cd $1
  >     hg qinit
  > }

fn to create dirty subrepo
  $ mksubrepo() {
  >     hg init $1
  >     cd $1
  >     echo a > a
  >     hg add
  >     cd ..
  > }

  $ testadd() {
  >     cat - > "$stdin"
  >     mksubrepo sub
  >     echo sub = sub >> .hgsub
  >     hg add .hgsub
  >     echo % abort when adding .hgsub w/dirty subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     echo [$?]
  >     hg -R sub ci -m0sub
  >     echo % update substate when adding .hgsub w/clean updated subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     hg debugsub
  > }

  $ testmod() {
  >     cat - > "$stdin"
  >     mksubrepo sub2
  >     echo sub2 = sub2 >> .hgsub
  >     echo % abort when modifying .hgsub w/dirty subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     echo [$?]
  >     hg -R sub2 ci -m0sub2
  >     echo % update substate when modifying .hgsub w/clean updated subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     hg debugsub
  > }

  $ testrm1() {
  >     cat - > "$stdin"
  >     mksubrepo sub3
  >     echo sub3 = sub3 >> .hgsub
  >     hg ci -Aqmsub3
  >     $EXTRA
  >     echo b >> sub3/a
  >     hg rm .hgsub
  >     echo % update substate when removing .hgsub w/dirty subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     echo % debugsub should be empty
  >     hg debugsub
  > }

  $ testrm2() {
  >     cat - > "$stdin"
  >     mksubrepo sub4
  >     echo sub4 = sub4 >> .hgsub
  >     hg ci -Aqmsub4
  >     $EXTRA
  >     hg rm .hgsub
  >     echo % update substate when removing .hgsub w/clean updated subrepo
  >     hg status -S
  >     echo '%' $*
  >     cat "$stdin" | hg $*
  >     echo % debugsub should be empty
  >     hg debugsub
  > }


handle subrepos safely on qnew

  $ mkrepo repo-2499-qnew
  $ testadd qnew -X path:no-effect -m0 0.diff
  adding a
  % abort when adding .hgsub w/dirty subrepo
  A .hgsub
  A sub/a
  % qnew -X path:no-effect -m0 0.diff
  abort: uncommitted changes in subrepository "sub"
  [255]
  % update substate when adding .hgsub w/clean updated subrepo
  A .hgsub
  A sub/a
  % qnew -X path:no-effect -m0 0.diff
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31

  $ testmod qnew --cwd .. -R repo-2499-qnew -X path:no-effect -m1 1.diff
  adding a
  % abort when modifying .hgsub w/dirty subrepo
  M .hgsub
  A sub2/a
  % qnew --cwd .. -R repo-2499-qnew -X path:no-effect -m1 1.diff
  abort: uncommitted changes in subrepository "sub2"
  [255]
  % update substate when modifying .hgsub w/clean updated subrepo
  M .hgsub
  A sub2/a
  % qnew --cwd .. -R repo-2499-qnew -X path:no-effect -m1 1.diff
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31
  path sub2
   source   sub2
   revision 1f94c7611cc6b74f5a17b16121a1170d44776845

  $ hg qpop -qa
  patch queue now empty
  $ testrm1 qnew -m2 2.diff
  adding a
  % update substate when removing .hgsub w/dirty subrepo
  M sub3/a
  R .hgsub
  % qnew -m2 2.diff
  % debugsub should be empty

  $ hg qpop -qa
  patch queue now empty
  $ testrm2 qnew -m3 3.diff
  adding a
  % update substate when removing .hgsub w/clean updated subrepo
  R .hgsub
  % qnew -m3 3.diff
  % debugsub should be empty

  $ cd ..


handle subrepos safely on qrefresh

  $ mkrepo repo-2499-qrefresh
  $ hg qnew -m0 0.diff
  $ testadd qrefresh
  adding a
  % abort when adding .hgsub w/dirty subrepo
  A .hgsub
  A sub/a
  % qrefresh
  abort: uncommitted changes in subrepository "sub"
  [255]
  % update substate when adding .hgsub w/clean updated subrepo
  A .hgsub
  A sub/a
  % qrefresh
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31

  $ hg qnew -m1 1.diff
  $ testmod qrefresh
  adding a
  % abort when modifying .hgsub w/dirty subrepo
  M .hgsub
  A sub2/a
  % qrefresh
  abort: uncommitted changes in subrepository "sub2"
  [255]
  % update substate when modifying .hgsub w/clean updated subrepo
  M .hgsub
  A sub2/a
  % qrefresh
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31
  path sub2
   source   sub2
   revision 1f94c7611cc6b74f5a17b16121a1170d44776845

  $ hg qpop -qa
  patch queue now empty
  $ EXTRA='hg qnew -m2 2.diff'
  $ testrm1 qrefresh
  adding a
  % update substate when removing .hgsub w/dirty subrepo
  M sub3/a
  R .hgsub
  % qrefresh
  % debugsub should be empty

  $ hg qpop -qa
  patch queue now empty
  $ EXTRA='hg qnew -m3 3.diff'
  $ testrm2 qrefresh
  adding a
  % update substate when removing .hgsub w/clean updated subrepo
  R .hgsub
  % qrefresh
  % debugsub should be empty
  $ EXTRA=

  $ cd ..


handle subrepos safely on qpush/qpop
(and we cannot qpop / qpush with a modified subrepo)

  $ mkrepo repo-2499-qpush
  $ mksubrepo sub
  adding a
  $ hg -R sub ci -m0sub
  $ echo sub = sub > .hgsub
  $ hg add .hgsub
  $ hg commit -m0
  $ hg debugsub
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31
  $ echo foo > ./sub/a
  $ hg -R sub commit -m foo
  $ hg commit -m1
  $ hg qimport -r "0:tip"
  $ hg -R sub id --id
  aa037b301eba

qpop
  $ hg -R sub update 0000
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg qpop
  abort: local changed subrepos found, qrefresh first
  [255]
  $ hg revert sub
  reverting subrepo sub
  adding sub/a
  $ hg qpop
  popping 1
  now at: 0
  $ hg status -AS
  C .hgsub
  C .hgsubstate
  C sub/a
  $ hg -R sub id --id
  b2fdb12cd82b

qpush
  $ hg -R sub update 0000
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg qpush
  abort: local changed subrepos found, qrefresh first
  [255]
  $ hg revert sub
  reverting subrepo sub
  adding sub/a
  $ hg qpush
  applying 1
   subrepository sub diverged (local revision: b2fdb12cd82b, remote revision: aa037b301eba)
  (M)erge, keep (l)ocal or keep (r)emote? m
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  now at: 1
  $ hg status -AS
  C .hgsub
  C .hgsubstate
  C sub/a
  $ hg -R sub id --id
  aa037b301eba

  $ cd ..


handle subrepos safely on qrecord

  $ mkrepo repo-2499-qrecord
  $ testadd qrecord --config ui.interactive=1 -m0 0.diff <<EOF
  > y
  > y
  > EOF
  adding a
  % abort when adding .hgsub w/dirty subrepo
  A .hgsub
  A sub/a
  % qrecord --config ui.interactive=1 -m0 0.diff
  diff --git a/.hgsub b/.hgsub
  new file mode 100644
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  @@ -0,0 +1,1 @@
  +sub = sub
  record this change to '.hgsub'? [Ynesfdaq?] y
  
  warning: subrepo spec file '.hgsub' not found
  warning: subrepo spec file '.hgsub' not found
  abort: uncommitted changes in subrepository "sub"
  [255]
  % update substate when adding .hgsub w/clean updated subrepo
  A .hgsub
  A sub/a
  % qrecord --config ui.interactive=1 -m0 0.diff
  diff --git a/.hgsub b/.hgsub
  new file mode 100644
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  @@ -0,0 +1,1 @@
  +sub = sub
  record this change to '.hgsub'? [Ynesfdaq?] y
  
  warning: subrepo spec file '.hgsub' not found
  warning: subrepo spec file '.hgsub' not found
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31
  $ testmod qrecord --config ui.interactive=1 -m1 1.diff <<EOF
  > y
  > y
  > EOF
  adding a
  % abort when modifying .hgsub w/dirty subrepo
  M .hgsub
  A sub2/a
  % qrecord --config ui.interactive=1 -m1 1.diff
  diff --git a/.hgsub b/.hgsub
  1 hunks, 1 lines changed
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  @@ -1,1 +1,2 @@
   sub = sub
  +sub2 = sub2
  record this change to '.hgsub'? [Ynesfdaq?] y
  
  abort: uncommitted changes in subrepository "sub2"
  [255]
  % update substate when modifying .hgsub w/clean updated subrepo
  M .hgsub
  A sub2/a
  % qrecord --config ui.interactive=1 -m1 1.diff
  diff --git a/.hgsub b/.hgsub
  1 hunks, 1 lines changed
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  @@ -1,1 +1,2 @@
   sub = sub
  +sub2 = sub2
  record this change to '.hgsub'? [Ynesfdaq?] y
  
  path sub
   source   sub
   revision b2fdb12cd82b021c3b7053d67802e77b6eeaee31
  path sub2
   source   sub2
   revision 1f94c7611cc6b74f5a17b16121a1170d44776845

  $ hg qpop -qa
  patch queue now empty
  $ testrm1 qrecord --config ui.interactive=1 -m2 2.diff <<EOF
  > y
  > y
  > EOF
  adding a
  % update substate when removing .hgsub w/dirty subrepo
  M sub3/a
  R .hgsub
  % qrecord --config ui.interactive=1 -m2 2.diff
  diff --git a/.hgsub b/.hgsub
  deleted file mode 100644
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  % debugsub should be empty

  $ hg qpop -qa
  patch queue now empty
  $ testrm2 qrecord --config ui.interactive=1 -m3 3.diff <<EOF
  > y
  > y
  > EOF
  adding a
  % update substate when removing .hgsub w/clean updated subrepo
  R .hgsub
  % qrecord --config ui.interactive=1 -m3 3.diff
  diff --git a/.hgsub b/.hgsub
  deleted file mode 100644
  examine changes to '.hgsub'? [Ynesfdaq?] y
  
  % debugsub should be empty

  $ cd ..


correctly handle subrepos with patch queues
  $ mkrepo repo-subrepo-with-queue
  $ mksubrepo sub
  adding a
  $ hg -R sub qnew sub0.diff
  $ echo sub = sub >> .hgsub
  $ hg add .hgsub
  $ hg qnew 0.diff

  $ cd ..

check whether MQ operations can import updated .hgsubstate correctly
both into 'revision' and 'patch file under .hg/patches':

  $ hg init importing-hgsubstate
  $ cd importing-hgsubstate

  $ echo a > a
  $ hg commit -u test -d '0 0' -Am '#0 in parent'
  adding a
  $ hg init sub
  $ echo sa > sub/sa
  $ hg -R sub commit -u test -d '0 0' -Am '#0 in sub'
  adding sa
  $ echo 'sub = sub' > .hgsub
  $ touch .hgsubstate
  $ hg add .hgsub .hgsubstate

  $ hg qnew -u test -d '0 0' import-at-qnew
  $ hg -R sub parents --template '{node} sub\n'
  b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  $ cat .hgsubstate
  b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  $ hg diff -c tip
  diff -r f499373e340c -r f69e96d86e75 .hgsub
  --- /dev/null
  +++ b/.hgsub
  @@ -0,0 +1,1 @@
  +sub = sub
  diff -r f499373e340c -r f69e96d86e75 .hgsubstate
  --- /dev/null
  +++ b/.hgsubstate
  @@ -0,0 +1,1 @@
  +b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  $ cat .hg/patches/import-at-qnew
  # HG changeset patch
  # User test
  # Date 0 0
  # Parent  f499373e340cdca5d01dee904aeb42dd2a325e71
  
  diff -r f499373e340c -r f69e96d86e75 .hgsub
  --- /dev/null
  +++ b/.hgsub
  @@ -0,0 +1,1 @@
  +sub = sub
  diff -r f499373e340c -r f69e96d86e75 .hgsubstate
  --- /dev/null
  +++ b/.hgsubstate
  @@ -0,0 +1,1 @@
  +b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  $ hg parents --template '{node}\n'
  f69e96d86e75a6d4fd88285dc9697acb23951041
  $ hg parents --template '{files}\n'
  .hgsub .hgsubstate

check also whether qnew not including ".hgsubstate" explicitly causes
as same result (in node hash) as one including it.

  $ hg qpop -a -q
  patch queue now empty
  $ hg qdelete import-at-qnew
  $ echo 'sub = sub' > .hgsub
  $ hg add .hgsub
  $ rm -f .hgsubstate
  $ hg qnew -u test -d '0 0' import-at-qnew
  $ hg parents --template '{node}\n'
  f69e96d86e75a6d4fd88285dc9697acb23951041
  $ hg parents --template '{files}\n'
  .hgsub .hgsubstate

check whether qrefresh imports updated .hgsubstate correctly

  $ hg qpop
  popping import-at-qnew
  patch queue now empty
  $ hg qpush
  applying import-at-qnew
  now at: import-at-qnew
  $ hg parents --template '{files}\n'
  .hgsub .hgsubstate

  $ hg qnew import-at-qrefresh
  $ echo sb > sub/sb
  $ hg -R sub commit -u test -d '0 0' -Am '#1 in sub'
  adding sb
  $ hg qrefresh -u test -d '0 0'
  $ hg -R sub parents --template '{node} sub\n'
  88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ cat .hgsubstate
  88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg diff -c tip
  diff -r 05b056bb9c8c -r d987bec230f4 .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ cat .hg/patches/import-at-qrefresh
  # HG changeset patch
  # User test
  # Date 0 0
  # Parent  05b056bb9c8c05ff15258b84fd42ab3527271033
  
  diff -r 05b056bb9c8c .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg parents --template '{files}\n'
  .hgsubstate

  $ hg qrefresh -u test -d '0 0'
  $ cat .hgsubstate
  88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg diff -c tip
  diff -r 05b056bb9c8c -r d987bec230f4 .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ cat .hg/patches/import-at-qrefresh
  # HG changeset patch
  # User test
  # Date 0 0
  # Parent  05b056bb9c8c05ff15258b84fd42ab3527271033
  
  diff -r 05b056bb9c8c .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg parents --template '{files}\n'
  .hgsubstate

  $ hg update -C tip
  0 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg qpop -a
  popping import-at-qrefresh
  popping import-at-qnew
  patch queue now empty

  $ hg -R sub update -C 0
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ echo 'sub = sub' > .hgsub
  $ hg commit -Am '#1 in parent'
  adding .hgsub
  $ hg -R sub update -C 1
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg commit -Am '#2 in parent (but will be rolled back soon)'
  $ hg rollback
  repository tip rolled back to revision 1 (undo commit)
  working directory now based on revision 1
  $ hg status
  M .hgsubstate
  $ hg qnew -u test -d '0 0' checkstate-at-qnew
  $ hg -R sub parents --template '{node} sub\n'
  88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ cat .hgsubstate
  88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg diff -c tip
  diff -r 4d91eb2fa1d1 -r 1259c112d884 .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ cat .hg/patches/checkstate-at-qnew
  # HG changeset patch
  # User test
  # Date 0 0
  # Parent  4d91eb2fa1d1b22ec513347b9cd06f6b49d470fa
  
  diff -r 4d91eb2fa1d1 -r 1259c112d884 .hgsubstate
  --- a/.hgsubstate
  +++ b/.hgsubstate
  @@ -1,1 +1,1 @@
  -b6f6e9c41f3dfd374a6d2ed4535c87951cf979cf sub
  +88ac1bef5ed43b689d1d200b59886b675dec474b sub
  $ hg parents --template '{files}\n'
  .hgsubstate

check whether qrefresh not including ".hgsubstate" explicitly causes
as same result (in node hash) as one including it.

  $ hg update -C -q 0
  $ hg qpop -a -q
  patch queue now empty
  $ hg qnew -u test -d '0 0' add-hgsub-at-qrefresh
  $ echo 'sub = sub' > .hgsub
  $ echo > .hgsubstate
  $ hg add .hgsub .hgsubstate
  $ hg qrefresh -u test -d '0 0'
  $ hg parents --template '{node}\n'
  7c48c35501aae6770ed9c2517014628615821a8e
  $ hg parents --template '{files}\n'
  .hgsub .hgsubstate

  $ hg qpop -a -q
  patch queue now empty
  $ hg qdelete add-hgsub-at-qrefresh
  $ hg qnew -u test -d '0 0' add-hgsub-at-qrefresh
  $ echo 'sub = sub' > .hgsub
  $ hg add .hgsub
  $ rm -f .hgsubstate
  $ hg qrefresh -u test -d '0 0'
  $ hg parents --template '{node}\n'
  7c48c35501aae6770ed9c2517014628615821a8e
  $ hg parents --template '{files}\n'
  .hgsub .hgsubstate

  $ cd ..

  $ cd ..