view tests/test-dispatch.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 command parsing and dispatch

  $ hg init a
  $ cd a

Redundant options used to crash (issue436):
  $ hg -v log -v
  $ hg -v log -v x

  $ echo a > a
  $ hg ci -Ama
  adding a

Missing arg:

  $ hg cat
  hg cat: invalid arguments
  hg cat [OPTION]... FILE...
  
  output the current or given revision of files
  
  options ([+] can be repeated):
  
   -o --output FORMAT       print output to file with formatted name
   -r --rev REV             print the given revision
      --decode              apply any matching decode filter
   -I --include PATTERN [+] include names matching the given patterns
   -X --exclude PATTERN [+] exclude names matching the given patterns
   -T --template TEMPLATE   display with template
  
  (use 'hg cat -h' to show more help)
  [255]

Missing parameter for early option:

  $ hg log -R 2>&1 | grep 'hg log'
  hg log: option -R requires argument
  hg log [OPTION]... [FILE]
  (use 'hg log -h' to show more help)

"--" may be an option value:

  $ hg -R -- log
  abort: repository -- not found!
  [255]
  $ hg log -R --
  abort: repository -- not found!
  [255]
  $ hg log -T --
  -- (no-eol)
  $ hg log -T -- -k nomatch

Parsing of early options should stop at "--":

  $ hg cat -- --config=hooks.pre-cat=false
  --config=hooks.pre-cat=false: no such file in rev cb9a9f314b8b
  [1]
  $ hg cat -- --debugger
  --debugger: no such file in rev cb9a9f314b8b
  [1]

Unparsable form of early options:

  $ hg cat --debugg
  abort: option --debugger may not be abbreviated!
  [255]

Parsing failure of early options should be detected before executing the
command:

  $ hg log -b '--config=hooks.pre-log=false' default
  abort: option --config may not be abbreviated!
  [255]
  $ hg log -b -R. default
  abort: option -R has to be separated from other options (e.g. not -qR) and --repository may only be abbreviated as --repo!
  [255]
  $ hg log --cwd .. -b --cwd=. default
  abort: option --cwd may not be abbreviated!
  [255]

However, we can't prevent it from loading extensions and configs:

  $ cat <<EOF > bad.py
  > raise Exception('bad')
  > EOF
  $ hg log -b '--config=extensions.bad=bad.py' default
  *** failed to import extension bad from bad.py: bad
  abort: option --config may not be abbreviated!
  [255]

  $ mkdir -p badrepo/.hg
  $ echo 'invalid-syntax' > badrepo/.hg/hgrc
  $ hg log -b -Rbadrepo default
  hg: parse error at badrepo/.hg/hgrc:1: invalid-syntax
  [255]

  $ hg log -b --cwd=inexistent default
  abort: $ENOENT$: 'inexistent'
  [255]

  $ hg log -b '--config=ui.traceback=yes' 2>&1 | grep '^Traceback'
  Traceback (most recent call last):
  $ hg log -b '--config=profiling.enabled=yes' 2>&1 | grep -i sample
  Sample count: .*|No samples recorded\. (re)

Early options can't be specified in [aliases] and [defaults] because they are
applied before the command name is resolved:

  $ hg log -b '--config=alias.log=log --config=hooks.pre-log=false'
  hg log: option -b not recognized
  error in definition for alias 'log': --config may only be given on the command
  line
  [255]

  $ hg log -b '--config=defaults.log=--config=hooks.pre-log=false'
  abort: option --config may not be abbreviated!
  [255]

Shell aliases bypass any command parsing rules but for the early one:

  $ hg log -b '--config=alias.log=!echo howdy'
  howdy

Early options must come first if HGPLAIN=+strictflags is specified:
(BUG: chg cherry-picks early options to pass them as a server command)

#if no-chg
  $ HGPLAIN=+strictflags hg log -b --config='hooks.pre-log=false' default
  abort: unknown revision '--config=hooks.pre-log=false'!
  [255]
  $ HGPLAIN=+strictflags hg log -b -R. default
  abort: unknown revision '-R.'!
  [255]
  $ HGPLAIN=+strictflags hg log -b --cwd=. default
  abort: unknown revision '--cwd=.'!
  [255]
#endif
  $ HGPLAIN=+strictflags hg log -b --debugger default
  abort: unknown revision '--debugger'!
  [255]
  $ HGPLAIN=+strictflags hg log -b --config='alias.log=!echo pwned' default
  abort: unknown revision '--config=alias.log=!echo pwned'!
  [255]

  $ HGPLAIN=+strictflags hg log --config='hooks.pre-log=false' -b default
  abort: option --config may not be abbreviated!
  [255]
  $ HGPLAIN=+strictflags hg log -q --cwd=.. -b default
  abort: option --cwd may not be abbreviated!
  [255]
  $ HGPLAIN=+strictflags hg log -q -R . -b default
  abort: option -R has to be separated from other options (e.g. not -qR) and --repository may only be abbreviated as --repo!
  [255]

  $ HGPLAIN=+strictflags hg --config='hooks.pre-log=false' log -b default
  abort: pre-log hook exited with status 1
  [255]
  $ HGPLAIN=+strictflags hg --cwd .. -q -Ra log -b default
  0:cb9a9f314b8b
  $ HGPLAIN=+strictflags hg --cwd .. -q --repository a log -b default
  0:cb9a9f314b8b
  $ HGPLAIN=+strictflags hg --cwd .. -q --repo a log -b default
  0:cb9a9f314b8b

For compatibility reasons, HGPLAIN=+strictflags is not enabled by plain HGPLAIN:

  $ HGPLAIN= hg log --config='hooks.pre-log=false' -b default
  abort: pre-log hook exited with status 1
  [255]
  $ HGPLAINEXCEPT= hg log --cwd .. -q -Ra -b default
  0:cb9a9f314b8b

[defaults]

  $ hg cat a
  a
  $ cat >> $HGRCPATH <<EOF
  > [defaults]
  > cat = -r null
  > EOF
  $ hg cat a
  a: no such file in rev 000000000000
  [1]

  $ cd "$TESTTMP"

OSError "No such file or directory" / "The system cannot find the path
specified" should include filename even when it is empty

  $ hg -R a archive ''
  abort: *: '' (glob)
  [255]

#if no-outer-repo

No repo:

  $ hg cat
  abort: no repository found in '$TESTTMP' (.hg not found)!
  [255]

#endif

#if rmcwd

Current directory removed:

  $ mkdir $TESTTMP/repo1
  $ cd $TESTTMP/repo1
  $ rm -rf $TESTTMP/repo1

The output could be one of the following and something else:
 chg: abort: failed to getcwd (errno = *) (glob)
 abort: error getting current working directory: * (glob)
 sh: 0: getcwd() failed: $ENOENT$
Since the exact behavior depends on the shell, only check it returns non-zero.
  $ HGDEMANDIMPORT=disable hg version -q 2>/dev/null || false
  [1]

#endif