view tests/test-convert-svn-sink.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

#require svn13

  $ svnupanddisplay()
  > {
  >     (
  >        cd $1;
  >        svn up -q;
  >        svn st -v | sed 's/  */ /g' | sort
  >        limit=''
  >        if [ $2 -gt 0 ]; then
  >            limit="--limit=$2"
  >        fi
  >        svn log --xml -v $limit | "$PYTHON" "$TESTDIR/svnxml.py"
  >     )
  > }

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > convert =
  > EOF

  $ hg init a

Add

  $ echo a > a/a
  $ mkdir -p a/d1/d2
  $ echo b > a/d1/d2/b
  $ hg --cwd a ci -d '0 0' -A -m 'add a file'
  adding a
  adding d1/d2/b

Modify

  $ svn-safe-append.py a a/a
  $ hg --cwd a ci -d '1 0' -m 'modify a file'
  $ hg --cwd a tip -q
  1:e0e2b8a9156b

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn repository 'a-hg'
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  1 add a file
  0 modify a file
  $ svnupanddisplay a-hg-wc 2
   2 1 test d1
   2 1 test d1/d2
   2 1 test d1/d2/b
   2 2 test .
   2 2 test a
  revision: 2
  author: test
  msg: modify a file
   M /a
  revision: 1
  author: test
  msg: add a file
   A /a
   A /d1
   A /d1/d2
   A /d1/d2/b
  $ ls a a-hg-wc
  a:
  a
  d1
  
  a-hg-wc:
  a
  d1
  $ cmp a/a a-hg-wc/a

Rename

  $ hg --cwd a mv a b
  $ hg --cwd a ci -d '2 0' -m 'rename a file'
  $ hg --cwd a tip -q
  2:eb5169441d43

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 rename a file
  $ svnupanddisplay a-hg-wc 1
   3 1 test d1
   3 1 test d1/d2
   3 1 test d1/d2/b
   3 3 test .
   3 3 test b
  revision: 3
  author: test
  msg: rename a file
   D /a
   A /b (from /a@2)
  $ ls a a-hg-wc
  a:
  b
  d1
  
  a-hg-wc:
  b
  d1

Copy

  $ hg --cwd a cp b c

  $ hg --cwd a ci -d '3 0' -m 'copy a file'
  $ hg --cwd a tip -q
  3:60effef6ab48

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 copy a file
  $ svnupanddisplay a-hg-wc 1
   4 1 test d1
   4 1 test d1/d2
   4 1 test d1/d2/b
   4 3 test b
   4 4 test .
   4 4 test c
  revision: 4
  author: test
  msg: copy a file
   A /c (from /b@3)
  $ ls a a-hg-wc
  a:
  b
  c
  d1
  
  a-hg-wc:
  b
  c
  d1

  $ hg --cwd a rm b

Remove

  $ hg --cwd a ci -d '4 0' -m 'remove a file'
  $ hg --cwd a tip -q
  4:87bbe3013fb6

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 remove a file
  $ svnupanddisplay a-hg-wc 1
   5 1 test d1
   5 1 test d1/d2
   5 1 test d1/d2/b
   5 4 test c
   5 5 test .
  revision: 5
  author: test
  msg: remove a file
   D /b
  $ ls a a-hg-wc
  a:
  c
  d1
  
  a-hg-wc:
  c
  d1

Executable

#if execbit
  $ chmod +x a/c
#else
  $ echo fake >> a/c
#endif
  $ hg --cwd a ci -d '5 0' -m 'make a file executable'
#if execbit
  $ hg --cwd a tip -q
  5:ff42e473c340
#else
  $ hg --cwd a tip -q
  5:817a700c8cf1
#endif

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 make a file executable
  $ svnupanddisplay a-hg-wc 1
   6 1 test d1
   6 1 test d1/d2
   6 1 test d1/d2/b
   6 6 test .
   6 6 test c
  revision: 6
  author: test
  msg: make a file executable
   M /c
#if execbit
  $ test -x a-hg-wc/c
#endif

#if symlink

Symlinks

  $ ln -s a/missing a/link
  $ hg --cwd a commit -Am 'add symlink'
  adding link
  $ hg --cwd a mv link newlink
  $ hg --cwd a commit -m 'move symlink'
  $ hg convert -d svn a a-svnlink
  initializing svn repository 'a-svnlink'
  initializing svn working copy 'a-svnlink-wc'
  scanning source...
  sorting...
  converting...
  7 add a file
  6 modify a file
  5 rename a file
  4 copy a file
  3 remove a file
  2 make a file executable
  1 add symlink
  0 move symlink
  $ svnupanddisplay a-svnlink-wc 1
   8 1 test d1
   8 1 test d1/d2
   8 1 test d1/d2/b
   8 6 test c
   8 8 test .
   8 8 test newlink
  revision: 8
  author: test
  msg: move symlink
   D /link
   A /newlink (from /link@7)

Make sure our changes don't affect the rest of the test cases

  $ hg --cwd a up 5
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg --cwd a --config extensions.strip= strip -r 6
  saved backup bundle to $TESTTMP/a/.hg/strip-backup/bd4f7b7a7067-ed505e42-backup.hg

#endif

Convert with --full adds and removes files that didn't change

  $ touch a/f
  $ hg -R a ci -Aqmf
  $ echo "rename c d" > filemap
  $ hg convert -d svn a --filemap filemap --full
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 f
  $ svnupanddisplay a-hg-wc 1
   7 7 test .
   7 7 test d
   7 7 test f
  revision: 7
  author: test
  msg: f
   D /c
   A /d
   D /d1
   A /f

  $ rm -rf a a-hg a-hg-wc


Executable in new directory

  $ hg init a

  $ mkdir a/d1
  $ echo a > a/d1/a
#if execbit
  $ chmod +x a/d1/a
#else
  $ echo fake >> a/d1/a
#endif
  $ hg --cwd a ci -d '0 0' -A -m 'add executable file in new directory'
  adding d1/a

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn repository 'a-hg'
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 add executable file in new directory
  $ svnupanddisplay a-hg-wc 1
   1 1 test .
   1 1 test d1
   1 1 test d1/a
  revision: 1
  author: test
  msg: add executable file in new directory
   A /d1
   A /d1/a
#if execbit
  $ test -x a-hg-wc/d1/a
#endif

Copy to new directory

  $ mkdir a/d2
  $ hg --cwd a cp d1/a d2/a
  $ hg --cwd a ci -d '1 0' -A -m 'copy file to new directory'

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  0 copy file to new directory
  $ svnupanddisplay a-hg-wc 1
   2 1 test d1
   2 1 test d1/a
   2 2 test .
   2 2 test d2
   2 2 test d2/a
  revision: 2
  author: test
  msg: copy file to new directory
   A /d2
   A /d2/a (from /d1/a@1)

Branchy history

  $ hg init b
  $ echo base > b/b
  $ hg --cwd b ci -d '0 0' -Ambase
  adding b

  $ svn-safe-append.py left-1 b/b
  $ echo left-1 > b/left-1
  $ hg --cwd b ci -d '1 0' -Amleft-1
  adding left-1

  $ svn-safe-append.py left-2 b/b
  $ echo left-2 > b/left-2
  $ hg --cwd b ci -d '2 0' -Amleft-2
  adding left-2

  $ hg --cwd b up 0
  1 files updated, 0 files merged, 2 files removed, 0 files unresolved

  $ svn-safe-append.py right-1 b/b
  $ echo right-1 > b/right-1
  $ hg --cwd b ci -d '3 0' -Amright-1
  adding right-1
  created new head

  $ svn-safe-append.py right-2 b/b
  $ echo right-2 > b/right-2
  $ hg --cwd b ci -d '4 0' -Amright-2
  adding right-2

  $ hg --cwd b up -C 2
  3 files updated, 0 files merged, 2 files removed, 0 files unresolved
  $ hg --cwd b merge
  merging b
  warning: conflicts while merging b! (edit, then use 'hg resolve --mark')
  2 files updated, 0 files merged, 0 files removed, 1 files unresolved
  use 'hg resolve' to retry unresolved file merges or 'hg merge --abort' to abandon
  [1]
  $ hg --cwd b revert -r 2 b
  $ hg --cwd b resolve -m b
  (no more unresolved files)
  $ hg --cwd b ci -d '5 0' -m 'merge'

Expect 4 changes

  $ hg convert -d svn b
  assuming destination b-hg
  initializing svn repository 'b-hg'
  initializing svn working copy 'b-hg-wc'
  scanning source...
  sorting...
  converting...
  5 base
  4 left-1
  3 left-2
  2 right-1
  1 right-2
  0 merge

  $ svnupanddisplay b-hg-wc 0
   4 2 test left-1
   4 3 test b
   4 3 test left-2
   4 4 test .
   4 4 test right-1
   4 4 test right-2
  revision: 4
  author: test
  msg: merge
   A /right-1
   A /right-2
  revision: 3
  author: test
  msg: left-2
   M /b
   A /left-2
  revision: 2
  author: test
  msg: left-1
   M /b
   A /left-1
  revision: 1
  author: test
  msg: base
   A /b

Tags are not supported, but must not break conversion

  $ rm -rf a a-hg a-hg-wc
  $ hg init a
  $ echo a > a/a
  $ hg --cwd a ci -d '0 0' -A -m 'Add file a'
  adding a
  $ hg --cwd a tag -d '1 0' -m 'Tagged as v1.0' v1.0

  $ hg convert -d svn a
  assuming destination a-hg
  initializing svn repository 'a-hg'
  initializing svn working copy 'a-hg-wc'
  scanning source...
  sorting...
  converting...
  1 Add file a
  0 Tagged as v1.0
  writing Subversion tags is not yet implemented
  $ svnupanddisplay a-hg-wc 2
   2 1 test a
   2 2 test .
   2 2 test .hgtags
  revision: 2
  author: test
  msg: Tagged as v1.0
   A /.hgtags
  revision: 1
  author: test
  msg: Add file a
   A /a
  $ rm -rf a a-hg a-hg-wc