view tests/test-eol.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 EOL extension

  $ cat >> $HGRCPATH <<EOF
  > [diff]
  > git = True
  > EOF

Set up helpers

  $ cat > switch-eol.py <<'EOF'
  > from __future__ import absolute_import
  > import os
  > import sys
  > try:
  >     import msvcrt
  >     msvcrt.setmode(sys.stdin.fileno(), os.O_BINARY)
  >     msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
  > except ImportError:
  >     pass
  > eolmap = {b'\n': '\\n', b'\r\n': '\\r\\n'}
  > (old, new) = sys.argv[1] == 'LF' and (b'\n', b'\r\n') or (b'\r\n', b'\n')
  > print("%% switching encoding from '%s' to '%s'"
  >       % (eolmap[old], eolmap[new]))
  > for path in sys.argv[2:]:
  >     data = open(path, 'rb').read()
  >     data = data.replace(old, new)
  >     open(path, 'wb').write(data)
  > EOF

  $ seteol () {
  >     if [ $1 = "LF" ]; then
  >         EOL='\n'
  >     else
  >         EOL='\r\n'
  >     fi
  > }

  $ makerepo () {
  >     seteol $1
  >     echo "% setup $1 repository"
  >     hg init repo
  >     cd repo
  >     cat > .hgeol <<EOF
  > [repository]
  > native = $1
  > [patterns]
  > mixed.txt = BIN
  > **.txt = native
  > EOF
  >     printf "first${EOL}second${EOL}third${EOL}" > a.txt
  >     hg commit --addremove -m 'checkin'
  >     echo
  >     cd ..
  > }

  $ dotest () {
  >     seteol $1
  >     echo "% hg clone repo repo-$1"
  >     hg clone --noupdate repo repo-$1
  >     cd repo-$1
  >     cat > .hg/hgrc <<EOF
  > [extensions]
  > eol =
  > [eol]
  > native = $1
  > EOF
  >     hg update
  >     echo '% a.txt'
  >     cat a.txt
  >     echo '% hg cat a.txt'
  >     hg cat a.txt
  >     printf "fourth${EOL}" >> a.txt
  >     echo '% a.txt'
  >     cat a.txt
  >     hg diff
  >     "$PYTHON" ../switch-eol.py $1 a.txt
  >     echo '% hg diff only reports a single changed line:'
  >     hg diff
  >     echo "% reverting back to $1 format"
  >     hg revert a.txt
  >     cat a.txt
  >     printf "first\r\nsecond\n" > mixed.txt
  >     hg add mixed.txt
  >     echo "% hg commit of inconsistent .txt file marked as binary (should work)"
  >     hg commit -m 'binary file'
  >     echo "% hg commit of inconsistent .txt file marked as native (should fail)"
  >     printf "first\nsecond\r\nthird\nfourth\r\n" > a.txt
  >     hg commit -m 'inconsistent file'
  >     echo "% hg commit --config eol.only-consistent=False (should work)"
  >     hg commit --config eol.only-consistent=False -m 'inconsistent file'
  >     echo "% hg commit of binary .txt file marked as native (binary files always okay)"
  >     printf "first${EOL}\0${EOL}third${EOL}" > a.txt
  >     hg commit -m 'binary file'
  >     cd ..
  >     rm -r repo-$1
  > }

  $ makemixedrepo () {
  >     echo
  >     echo "# setup $1 repository"
  >     hg init mixed
  >     cd mixed
  >     printf "foo\r\nbar\r\nbaz\r\n" > win.txt
  >     printf "foo\nbar\nbaz\n" > unix.txt
  >     #printf "foo\r\nbar\nbaz\r\n" > mixed.txt
  >     hg commit --addremove -m 'created mixed files'
  >     echo "# setting repository-native EOLs to $1"
  >     cat > .hgeol <<EOF
  > [repository]
  > native = $1
  > [patterns]
  > **.txt = native
  > EOF
  >     hg commit --addremove -m 'added .hgeol'
  >     cd ..
  > }

  $ testmixed () {
  >     echo
  >     echo "% hg clone mixed mixed-$1"
  >     hg clone mixed mixed-$1
  >     cd mixed-$1
  >     echo '% hg status (eol extension not yet activated)'
  >     hg status
  >     cat > .hg/hgrc <<EOF
  > [extensions]
  > eol =
  > [eol]
  > native = $1
  > EOF
  >     echo '% hg status (eol activated)'
  >     hg status
  >     echo '% hg commit'
  >     hg commit -m 'synchronized EOLs'
  >     echo '% hg status'
  >     hg status
  >     cd ..
  >     rm -r mixed-$1
  > }

Basic tests

  $ makerepo LF
  % setup LF repository
  adding .hgeol
  adding a.txt
  
  $ dotest LF
  % hg clone repo repo-LF
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % a.txt
  first
  second
  third
  % hg cat a.txt
  first
  second
  third
  % a.txt
  first
  second
  third
  fourth
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first
   second
   third
  +fourth
  % switching encoding from '\n' to '\r\n'
  % hg diff only reports a single changed line:
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first
   second
   third
  +fourth
  % reverting back to LF format
  first
  second
  third
  % hg commit of inconsistent .txt file marked as binary (should work)
  % hg commit of inconsistent .txt file marked as native (should fail)
  abort: inconsistent newline style in a.txt
  
  % hg commit --config eol.only-consistent=False (should work)
  % hg commit of binary .txt file marked as native (binary files always okay)
  $ dotest CRLF
  % hg clone repo repo-CRLF
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % hg cat a.txt
  first
  second
  third
  % a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  fourth\r (esc)
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first
   second
   third
  +fourth
  % switching encoding from '\r\n' to '\n'
  % hg diff only reports a single changed line:
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first
   second
   third
  +fourth
  % reverting back to CRLF format
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % hg commit of inconsistent .txt file marked as binary (should work)
  % hg commit of inconsistent .txt file marked as native (should fail)
  abort: inconsistent newline style in a.txt
  
  % hg commit --config eol.only-consistent=False (should work)
  % hg commit of binary .txt file marked as native (binary files always okay)
  $ rm -r repo
  $ makerepo CRLF
  % setup CRLF repository
  adding .hgeol
  adding a.txt
  
  $ dotest LF
  % hg clone repo repo-LF
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % a.txt
  first
  second
  third
  % hg cat a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % a.txt
  first
  second
  third
  fourth
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first\r (esc)
   second\r (esc)
   third\r (esc)
  +fourth\r (esc)
  % switching encoding from '\n' to '\r\n'
  % hg diff only reports a single changed line:
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first\r (esc)
   second\r (esc)
   third\r (esc)
  +fourth\r (esc)
  % reverting back to LF format
  first
  second
  third
  % hg commit of inconsistent .txt file marked as binary (should work)
  % hg commit of inconsistent .txt file marked as native (should fail)
  abort: inconsistent newline style in a.txt
  
  % hg commit --config eol.only-consistent=False (should work)
  % hg commit of binary .txt file marked as native (binary files always okay)
  $ dotest CRLF
  % hg clone repo repo-CRLF
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % hg cat a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % a.txt
  first\r (esc)
  second\r (esc)
  third\r (esc)
  fourth\r (esc)
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first\r (esc)
   second\r (esc)
   third\r (esc)
  +fourth\r (esc)
  % switching encoding from '\r\n' to '\n'
  % hg diff only reports a single changed line:
  diff --git a/a.txt b/a.txt
  --- a/a.txt
  +++ b/a.txt
  @@ -1,3 +1,4 @@
   first\r (esc)
   second\r (esc)
   third\r (esc)
  +fourth\r (esc)
  % reverting back to CRLF format
  first\r (esc)
  second\r (esc)
  third\r (esc)
  % hg commit of inconsistent .txt file marked as binary (should work)
  % hg commit of inconsistent .txt file marked as native (should fail)
  abort: inconsistent newline style in a.txt
  
  % hg commit --config eol.only-consistent=False (should work)
  % hg commit of binary .txt file marked as native (binary files always okay)
  $ rm -r repo

Mixed tests

  $ makemixedrepo LF
  
  # setup LF repository
  adding unix.txt
  adding win.txt
  # setting repository-native EOLs to LF
  adding .hgeol
  $ testmixed LF
  
  % hg clone mixed mixed-LF
  updating to branch default
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % hg status (eol extension not yet activated)
  % hg status (eol activated)
  M win.txt
  % hg commit
  % hg status
  $ testmixed CRLF
  
  % hg clone mixed mixed-CRLF
  updating to branch default
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % hg status (eol extension not yet activated)
  % hg status (eol activated)
  M win.txt
  % hg commit
  % hg status
  $ rm -r mixed
  $ makemixedrepo CRLF
  
  # setup CRLF repository
  adding unix.txt
  adding win.txt
  # setting repository-native EOLs to CRLF
  adding .hgeol
  $ testmixed LF
  
  % hg clone mixed mixed-LF
  updating to branch default
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % hg status (eol extension not yet activated)
  % hg status (eol activated)
  M unix.txt
  % hg commit
  % hg status
  $ testmixed CRLF
  
  % hg clone mixed mixed-CRLF
  updating to branch default
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved
  % hg status (eol extension not yet activated)
  % hg status (eol activated)
  M unix.txt
  % hg commit
  % hg status
  $ rm -r mixed

  $ echo '[extensions]' >> $HGRCPATH
  $ echo 'eol =' >> $HGRCPATH

#if unix-permissions

Test issue2569 -- eol extension takes write lock on reading:

  $ hg init repo
  $ cd repo
  $ touch .hgeol
  $ hg status
  ? .hgeol
  $ chmod -R -w .hg
  $ sleep 1
  $ touch .hgeol
  $ hg status --traceback
  ? .hgeol
  $ chmod -R u+w .hg
  $ cd ..

#endif

Test cleverencode: and cleverdecode: aliases for win32text extension

  $ cat <<EOF >> $HGRCPATH
  > [encode]
  > **.txt = cleverencode:
  > [decode]
  > **.txt = cleverdecode:
  > EOF

  $ hg init win32compat
  $ cd win32compat
  $ printf "foo\r\nbar\r\nbaz\r\n" > win.txt
  $ printf "foo\nbar\nbaz\n" > unix.txt
  $ hg add
  adding unix.txt
  adding win.txt
  $ hg commit -m checkin

Check that both files have LF line-endings in the repository:

  $ hg cat win.txt
  foo
  bar
  baz
  $ hg cat unix.txt
  foo
  bar
  baz

Test handling of a broken .hgeol file:

  $ touch .hgeol
  $ hg add .hgeol
  $ hg commit -m 'clean version'
  $ echo "bad" > .hgeol
  $ hg status
  warning: ignoring .hgeol file due to parse error at .hgeol:1: bad
  M .hgeol
  $ hg revert .hgeol
  warning: ignoring .hgeol file due to parse error at .hgeol:1: bad
  $ hg status
  ? .hgeol.orig

Test eol.only-consistent can be specified in .hgeol

  $ cd $TESTTMP
  $ hg init only-consistent
  $ cd only-consistent
  $ printf "first\nsecond\r\n" > a.txt
  $ hg add a.txt
  $ cat > .hgeol << EOF
  > [eol]
  > only-consistent = True
  > EOF
  $ hg commit -m 'inconsistent'
  abort: inconsistent newline style in a.txt
  
  [255]
  $ cat > .hgeol << EOF
  > [eol]
  > only-consistent = False
  > EOF
  $ hg commit -m 'consistent'

  $ hg init subrepo
  $ hg -R subrepo pull -qu .
  $ echo "subrepo = subrepo" > .hgsub
  $ hg ci -Am "add subrepo"
  adding .hgeol
  adding .hgsub
  $ hg archive -S ../archive
  $ find ../archive/* | sort
  ../archive/a.txt
  ../archive/subrepo
  ../archive/subrepo/a.txt
  $ cat ../archive/a.txt ../archive/subrepo/a.txt
  first\r (esc)
  second\r (esc)
  first\r (esc)
  second\r (esc)

Test trailing newline

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

setup repository

  $ cd $TESTTMP
  $ hg init trailing
  $ cd trailing
  $ cat > .hgeol <<EOF
  > [patterns]
  > **.txt = native
  > [eol]
  > fix-trailing-newline = False
  > EOF

add text without trailing newline

  $ printf "first\nsecond" > a.txt
  $ hg commit --addremove -m 'checking in'
  adding .hgeol
  adding a.txt
  $ rm a.txt
  $ hg update -C -q
  $ cat a.txt
  first
  second (no-eol)

  $ cat > .hgeol <<EOF
  > [patterns]
  > **.txt = native
  > [eol]
  > fix-trailing-newline = True
  > EOF
  $ printf "third\nfourth" > a.txt
  $ hg commit -m 'checking in with newline fix'
  $ rm a.txt
  $ hg update -C -q
  $ cat a.txt
  third
  fourth

append a line without trailing newline

  $ printf "fifth" >> a.txt
  $ hg commit -m 'adding another line line'
  $ rm a.txt
  $ hg update -C -q
  $ cat a.txt
  third
  fourth
  fifth

amend of changesets with renamed/deleted files expose new code paths

  $ hg mv a.txt b.txt
  $ hg ci --amend -q
  $ hg diff -c.
  diff --git a/a.txt b/b.txt
  rename from a.txt
  rename to b.txt
  --- a/a.txt
  +++ b/b.txt
  @@ -1,2 +1,3 @@
   third
   fourth
  +fifth

  $ cd ..