mercurial/help/templates.txt

revlog: use an LRU cache for delta chain bases

Profiling using statprof revealed a hotspot during changegroup
application calculating delta chain bases on generaldelta repos.
Essentially, revlog._addrevision() was performing a lot of redundant
work tracing the delta chain as part of determining when the chain
distance was acceptable. This was most pronounced when adding
revisions to manifests, which can have delta chains thousands of
revisions long.

There was a delta chain base cache on revlogs before, but it only
captured a single revision. This was acceptable before generaldelta,
when _addrevision would build deltas from the previous revision and
thus we'd pretty much guarantee a cache hit when resolving the delta
chain base on a subsequent _addrevision call. However, it isn't
suitable for generaldelta because parent revisions aren't necessarily
the last processed revision.

This patch converts the delta chain base cache to an LRU dict cache.
The cache can hold multiple entries, so generaldelta repos have a
higher chance of getting a cache hit.

The impact of this change when processing changegroup additions is
significant. On a generaldelta conversion of the "mozilla-unified"
repo (which contains heads of the main Firefox repositories in
chronological order - this means there are lots of transitions between
heads in revlog order), this change has the following impact when
performing an `hg unbundle` of an uncompressed bundle of the repo:

before: 5:42 CPU time
after:  4:34 CPU time

Most of this time is saved when applying the changelog and manifest
revlogs:

before: 2:30 CPU time
after:  1:17 CPU time

That nearly a 50% reduction in CPU time applying changesets and
manifests!

Applying a gzipped bundle of the same repo (effectively simulating a
`hg clone` over HTTP) showed a similar speedup:

before: 5:53 CPU time
after:  4:46 CPU time

Wall time improvements were basically the same as CPU time.

I didn't measure explicitly, but it feels like most of the time
is saved when processing manifests. This makes sense, as large
manifests tend to have very long delta chains and thus benefit the
most from this cache.

So, this change effectively makes changegroup application (which is
used by `hg unbundle`, `hg clone`, `hg pull`, `hg unshelve`, and
various other commands) significantly faster when delta chains are
long (which can happen on repos with large numbers of files and thus
large manifests).

In theory, this change can result in more memory utilization. However,
we're caching a dict of ints. At most we have 200 ints + Python object
overhead per revlog. And, the cache is really only populated when
performing read-heavy operations, such as adding changegroups or
scanning an individual revlog. For memory bloat to be an issue, we'd
need to scan/read several revisions from several revlogs all while
having active references to several revlogs. I don't think there are
many operations that do this, so I don't think memory bloat from the
cache will be an issue.

Mercurial allows you to customize output of commands through
templates. You can either pass in a template or select an existing
template-style from the command line, via the --template option.

You can customize output for any "log-like" command: log,
outgoing, incoming, tip, parents, and heads.

Some built-in styles are packaged with Mercurial. These can be listed
with :hg:`log --template list`. Example usage::

    $ hg log -r1.0::1.1 --template changelog

A template is a piece of text, with markup to invoke variable
expansion::

    $ hg log -r1 --template "{node}\n"
    b56ce7b07c52de7d5fd79fb89701ea538af65746

Strings in curly braces are called keywords. The availability of
keywords depends on the exact context of the templater. These
keywords are usually available for templating a log-like command:

.. keywordsmarker

The "date" keyword does not produce human-readable output. If you
want to use a date in your output, you can use a filter to process
it. Filters are functions which return a string based on the input
variable. Be sure to use the stringify filter first when you're
applying a string-input filter to a list-like input variable.
You can also use a chain of filters to get the desired output::

   $ hg tip --template "{date|isodate}\n"
   2008-08-21 18:22 +0000

List of filters:

.. filtersmarker

Note that a filter is nothing more than a function call, i.e.
``expr|filter`` is equivalent to ``filter(expr)``.

In addition to filters, there are some basic built-in functions:

.. functionsmarker

Also, for any expression that returns a list, there is a list operator::

    expr % "{template}"

As seen in the above example, ``{template}`` is interpreted as a template.
To prevent it from being interpreted, you can use an escape character ``\{``
or a raw string prefix, ``r'...'``.

New keywords and functions can be defined in the ``templatealias`` section of
a Mercurial configuration file::

  <alias> = <definition>

Arguments of the form `a1`, `a2`, etc. are substituted from the alias into
the definition.

For example,

::

  [templatealias]
  r = rev
  rn = "{r}:{node|short}"
  leftpad(s, w) = pad(s, w, ' ', True)

defines two symbol aliases, ``r`` and ``rn``, and a function alias
``leftpad()``.

It's also possible to specify complete template strings, using the
``templates`` section. The syntax used is the general template string syntax.

For example,

::

  [templates]
  nodedate = "{node|short}: {date(date, "%Y-%m-%d")}\n"

defines a template, ``nodedate``, which can be called like::

  $ hg log -r . -Tnodedate

Some sample command line templates:

- Format lists, e.g. files::

   $ hg log -r 0 --template "files:\n{files % '  {file}\n'}"

- Join the list of files with a ", "::

   $ hg log -r 0 --template "files: {join(files, ', ')}\n"

- Separate non-empty arguments by a " "::

   $ hg log -r 0 --template "{separate(' ', node, bookmarks, tags}\n"

- Modify each line of a commit description::

   $ hg log --template "{splitlines(desc) % '**** {line}\n'}"

- Format date::

   $ hg log -r 0 --template "{date(date, '%Y')}\n"

- Display date in UTC::

   $ hg log -r 0 --template "{localdate(date, 'UTC')|date}\n"

- Output the description set to a fill-width of 30::

   $ hg log -r 0 --template "{fill(desc, 30)}"

- Use a conditional to test for the default branch::

   $ hg log -r 0 --template "{ifeq(branch, 'default', 'on the main branch',
   'on branch {branch}')}\n"

- Append a newline if not empty::

   $ hg tip --template "{if(author, '{author}\n')}"

- Label the output for use with the color extension::

   $ hg log -r 0 --template "{label('changeset.{phase}', node|short)}\n"

- Invert the firstline filter, i.e. everything but the first line::

   $ hg log -r 0 --template "{sub(r'^.*\n?\n?', '', desc)}\n"

- Display the contents of the 'extra' field, one per line::

   $ hg log -r 0 --template "{join(extras, '\n')}\n"

- Mark the active bookmark with '*'::

   $ hg log --template "{bookmarks % '{bookmark}{ifeq(bookmark, active, '*')} '}\n"

- Find the previous release candidate tag, the distance and changes since the tag::

   $ hg log -r . --template "{latesttag('re:^.*-rc$') % '{tag}, {changes}, {distance}'}\n"

- Mark the working copy parent with '@'::

   $ hg log --template "{ifcontains(rev, revset('.'), '@')}\n"

- Show details of parent revisions::

   $ hg log --template "{revset('parents(%d)', rev) % '{desc|firstline}\n'}"

- Show only commit descriptions that start with "template"::

   $ hg log --template "{startswith('template', firstline(desc))}\n"

- Print the first word of each line of a commit message::

   $ hg log --template "{word(0, desc)}\n"