view tests/test-pager-legacy.t @ 46607:e9901d01d135

revlog: add a mechanism to verify expected file position before appending

If someone uses `hg debuglocks`, or some non-hg process writes to the .hg
directory without respecting the locks, or if the repo's on a networked
filesystem, it's possible for the revlog code to write out corrupted data.

The form of this corruption can vary depending on what data was written and how
that happened. We are in the "networked filesystem" case (though I've had users
also do this to themselves with the "`hg debuglocks`" scenario), and most often
see this with the changelog. What ends up happening is we produce two items
(let's call them rev1 and rev2) in the .i file that have the same linkrev,
baserev, and offset into the .d file, while the data in the .d file is appended
properly. rev2's compressed_size is accurate for rev2, but when we go to
decompress the data in the .d file, we use the offset that's recorded in the
index file, which is the same as rev1, and attempt to decompress
rev2.compressed_size bytes of rev1's data. This usually does not succeed. :)

When using inline data, this also fails, though I haven't investigated why too
closely. This shows up as a "patch decode" error. I believe what's happening
there is that we're basically ignoring the offset field, getting the data
properly, but since baserev != rev, it thinks this is a delta based on rev
(instead of a full text) and can't actually apply it as such.

For now, I'm going to make this an optional component and default it to entirely
off. I may increase the default severity of this in the future, once I've
enabled it for my users and we gain more experience with it. Luckily, most of my
users have a versioned filesystem and can roll back to before the corruption has
been written, it's just a hassle to do so and not everyone knows how (so it's a
support burden). Users on other filesystems will not have that luxury, and this
can cause them to have a corrupted repository that they are unlikely to know how
to resolve, and they'll see this as a data-loss event. Refusing to create the
corruption is a much better user experience.

This mechanism is not perfect. There may be false-negatives (racy writes that
are not detected). There should not be any false-positives (non-racy writes that
are detected as such). This is not a mechanism that makes putting a repo on a
networked filesystem "safe" or "supported", just *less* likely to cause
corruption.

Differential Revision: https://phab.mercurial-scm.org/D9952

  $ cat >> fakepager.py <<EOF
  > import sys
  > for line in sys.stdin:
  >     sys.stdout.write('paged! %r\n' % line)
  > EOF

Enable ui.formatted because pager won't fire without it, and set up
pager and tell it to use our fake pager that lets us see when the
pager was running.
  $ cat >> $HGRCPATH <<EOF
  > [ui]
  > formatted = yes
  > color = no
  > [extensions]
  > pager=
  > [pager]
  > pager = "$PYTHON" $TESTTMP/fakepager.py
  > EOF

  $ hg init repo
  $ cd repo
  $ echo a >> a
  $ hg add a
  $ hg ci -m 'add a'
  $ for x in `"$PYTHON" $TESTDIR/seq.py 1 10`; do
  >   echo a $x >> a
  >   hg ci -m "modify a $x"
  > done

By default diff and log are paged, but summary is not:

  $ hg diff -c 2 --pager=yes
  paged! 'diff -r f4be7687d414 -r bce265549556 a\n'
  paged! '--- a/a\tThu Jan 01 00:00:00 1970 +0000\n'
  paged! '+++ b/a\tThu Jan 01 00:00:00 1970 +0000\n'
  paged! '@@ -1,2 +1,3 @@\n'
  paged! ' a\n'
  paged! ' a 1\n'
  paged! '+a 2\n'

  $ hg log --limit 2
  paged! 'changeset:   10:46106edeeb38\n'
  paged! 'tag:         tip\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 10\n'
  paged! '\n'
  paged! 'changeset:   9:6dd8ea7dd621\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 9\n'
  paged! '\n'

  $ hg summary
  parent: 10:46106edeeb38 tip
   modify a 10
  branch: default
  commit: (clean)
  update: (current)
  phases: 11 draft

We can enable the pager on summary:

  $ hg --config pager.attend-summary=yes summary
  paged! 'parent: 10:46106edeeb38 tip\n'
  paged! ' modify a 10\n'
  paged! 'branch: default\n'
  paged! 'commit: (clean)\n'
  paged! 'update: (current)\n'
  paged! 'phases: 11 draft\n'

  $ hg --config pager.attend-diff=no diff -c 2
  diff -r f4be7687d414 -r bce265549556 a
  --- a/a	Thu Jan 01 00:00:00 1970 +0000
  +++ b/a	Thu Jan 01 00:00:00 1970 +0000
  @@ -1,2 +1,3 @@
   a
   a 1
  +a 2

If we completely change the attend list that's respected:
  $ hg --config pager.attend=summary diff -c 2
  diff -r f4be7687d414 -r bce265549556 a
  --- a/a	Thu Jan 01 00:00:00 1970 +0000
  +++ b/a	Thu Jan 01 00:00:00 1970 +0000
  @@ -1,2 +1,3 @@
   a
   a 1
  +a 2

If 'log' is in attend, then 'history' should also be paged:
  $ hg history --limit 2 --config pager.attend=log
  paged! 'changeset:   10:46106edeeb38\n'
  paged! 'tag:         tip\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 10\n'
  paged! '\n'
  paged! 'changeset:   9:6dd8ea7dd621\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 9\n'
  paged! '\n'

Possible bug: history is explicitly ignored in pager config, but
because log is in the attend list it still gets pager treatment.

  $ hg history --limit 2 --config pager.attend=log \
  >   --config pager.ignore=history
  paged! 'changeset:   10:46106edeeb38\n'
  paged! 'tag:         tip\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 10\n'
  paged! '\n'
  paged! 'changeset:   9:6dd8ea7dd621\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 9\n'
  paged! '\n'

Possible bug: history is explicitly marked as attend-history=no, but
it doesn't fail to get paged because log is still in the attend list.

  $ hg history --limit 2 --config pager.attend-history=no
  paged! 'changeset:   10:46106edeeb38\n'
  paged! 'tag:         tip\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 10\n'
  paged! '\n'
  paged! 'changeset:   9:6dd8ea7dd621\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 9\n'
  paged! '\n'

Possible bug: disabling pager for log but enabling it for history
doesn't result in history being paged.

  $ hg history --limit 2 --config pager.attend-log=no \
  > --config pager.attend-history=yes
  changeset:   10:46106edeeb38
  tag:         tip
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     modify a 10
  
  changeset:   9:6dd8ea7dd621
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     modify a 9
  
Pager should not start if stdout is not a tty.

  $ hg log -l1 -q --config ui.formatted=False
  10:46106edeeb38

Pager with color enabled allows colors to come through by default,
even though stdout is no longer a tty.
  $ cat >> $HGRCPATH <<EOF
  > [ui]
  > color = always
  > [color]
  > mode = ansi
  > EOF
  $ hg log --limit 3
  paged! '\x1b[0;33mchangeset:   10:46106edeeb38\x1b[0m\n'
  paged! 'tag:         tip\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 10\n'
  paged! '\n'
  paged! '\x1b[0;33mchangeset:   9:6dd8ea7dd621\x1b[0m\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 9\n'
  paged! '\n'
  paged! '\x1b[0;33mchangeset:   8:cff05a6312fe\x1b[0m\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'summary:     modify a 8\n'
  paged! '\n'

Pager works with shell aliases.

  $ cat >> $HGRCPATH <<EOF
  > [alias]
  > echoa = !echo a
  > EOF

  $ hg echoa
  a
  $ hg --config pager.attend-echoa=yes echoa
  paged! 'a\n'

Pager attributes should be copied to mq repo. Otherwise pager would be started
twice and color mode would be lost.

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > mq =
  > EOF
  $ hg init --mq
  $ hg qnew foo.patch
  $ hg qpop
  popping foo.patch
  patch queue now empty
  $ hg ci --mq -m 'commit patches'
  $ hg log --mq --debug
  starting pager for command 'extension-via-attend-log'
  paged! '\x1b[0;33mchangeset:   0:6cc2ded15503e368aaf76b6cc3d12f320c9e3b87\x1b[0m\n'
  paged! 'tag:         tip\n'
  paged! 'phase:       draft\n'
  paged! 'parent:      -1:0000000000000000000000000000000000000000\n'
  paged! 'parent:      -1:0000000000000000000000000000000000000000\n'
  paged! 'manifest:    0:4980de1ae1b612014d5bcfa9507da84ce8891daa\n'
  paged! 'user:        test\n'
  paged! 'date:        Thu Jan 01 00:00:00 1970 +0000\n'
  paged! 'files+:      .hgignore foo.patch series\n'
  paged! 'extra:       branch=default\n'
  paged! 'description:\n'
  paged! 'commit patches\n'
  paged! '\n'
  paged! '\n'

Pager works with hg aliases including environment variables.

  $ cat >> $HGRCPATH <<'EOF'
  > [alias]
  > printa = log -T "$A\n" -r 0
  > EOF

  $ A=1 hg --config pager.attend-printa=yes printa
  paged! '1\n'
  $ A=2 hg --config pager.attend-printa=yes printa
  paged! '2\n'

Something that's explicitly attended is still not paginated if the
pager is globally set to off using a flag:
  $ A=2 hg --config pager.attend-printa=yes printa --pager=no
  2

Pager should not override the exit code of other commands

  $ cat >> $TESTTMP/fortytwo.py <<'EOF'
  > from mercurial import commands, registrar
  > cmdtable = {}
  > command = registrar.command(cmdtable)
  > @command(b'fortytwo', [], b'fortytwo', norepo=True)
  > def fortytwo(ui, *opts):
  >     ui.write(b'42\n')
  >     return 42
  > EOF

  $ cat >> $HGRCPATH <<'EOF'
  > [extensions]
  > fortytwo = $TESTTMP/fortytwo.py
  > EOF

  $ hg fortytwo --pager=on
  paged! '42\n'
  [42]