Mercurial > public > mercurial-scm > hg
view mercurial/help/hgweb.txt @ 26117:4dc5b51f38fe
revlog: change generaldelta delta parent heuristic
The old generaldelta heuristic was "if p1 (or p2) was closer than the last full text,
use it, otherwise use prev". This was problematic when a repo contained multiple
branches that were very different. If commits to branch A were pushed, and the
last full text was branch B, it would generate a fulltext. Then if branch B was
pushed, it would generate another fulltext. The problem is that the last
fulltext (and delta'ing against `prev` in general) has no correlation with the
contents of the incoming revision, and therefore will always have degenerate
cases.
According to the blame, that algorithm was chosen to minimize the chain length.
Since there is already code that protects against that (the delta-vs-fulltext
code), and since it has been improved since the original generaldelta algorithm
went in (2011), I believe the chain length criteria will still be preserved.
The new algorithm always diffs against p1 (or p2 if it's closer), unless the
resulting delta will fail the delta-vs-fulltext check, in which case we delta
against prev.
Some before and after stats on manifest.d size.
internal large repo
old heuristic - 2.0 GB
new heuristic - 1.2 GB
mozilla-central
old heuristic - 242 MB
new heuristic - 261 MB
The regression in mozilla central is due to the new heuristic choosing p2r as
the delta when it's closer to the tip. Switching the algorithm to always prefer
p1r brings the size back down (242 MB). This is result of the way in which
mozilla does merges and pushes, and the result could easily swing the other
direction in other repos (depending on if they merge X into Y or Y into X), but
will never be as degenerate as before.
I future patch will address the regression by introducing an optional, even more
aggressive delta heuristic which will knock the mozilla manifest size down
dramatically.
author | Durham Goode <durham@fb.com> |
---|---|
date | Sun, 30 Aug 2015 13:58:11 -0700 |
parents | a3f2ea1d4943 |
children | 1aee2ab0f902 |
line wrap: on
line source
Mercurial's internal web server, hgweb, can serve either a single repository, or a tree of repositories. In the second case, repository paths and global options can be defined using a dedicated configuration file common to :hg:`serve`, ``hgweb.wsgi``, ``hgweb.cgi`` and ``hgweb.fcgi``. This file uses the same syntax as other Mercurial configuration files but recognizes only the following sections: - web - paths - collections The ``web`` options are thoroughly described in :hg:`help config`. The ``paths`` section maps URL paths to paths of repositories in the filesystem. hgweb will not expose the filesystem directly - only Mercurial repositories can be published and only according to the configuration. The left hand side is the path in the URL. Note that hgweb reserves subpaths like ``rev`` or ``file``, try using different names for nested repositories to avoid confusing effects. The right hand side is the path in the filesystem. If the specified path ends with ``*`` or ``**`` the filesystem will be searched recursively for repositories below that point. With ``*`` it will not recurse into the repositories it finds (except for ``.hg/patches``). With ``**`` it will also search inside repository working directories and possibly find subrepositories. In this example:: [paths] /projects/a = /srv/tmprepos/a /projects/b = c:/repos/b / = /srv/repos/* /user/bob = /home/bob/repos/** - The first two entries make two repositories in different directories appear under the same directory in the web interface - The third entry will publish every Mercurial repository found in ``/srv/repos/``, for instance the repository ``/srv/repos/quux/`` will appear as ``http://server/quux/`` - The fourth entry will publish both ``http://server/user/bob/quux/`` and ``http://server/user/bob/quux/testsubrepo/`` The ``collections`` section is deprecated and has been superseded by ``paths``. URLs and Common Arguments ========================= URLs under each repository have the form ``/{command}[/{arguments}]`` where ``{command}`` represents the name of a command or handler and ``{arguments}`` represents any number of additional URL parameters to that command. The web server has a default style associated with it. Styles map to a collection of named templates. Each template is used to render a specific piece of data, such as a changeset or diff. The style for the current request can be overwritten two ways. First, if ``{command}`` contains a hyphen (``-``), the text before the hyphen defines the style. For example, ``/atom-log`` will render the ``log`` command handler with the ``atom`` style. The second way to set the style is with the ``style`` query string argument. For example, ``/log?style=atom``. The hyphenated URL parameter is preferred. Not all templates are available for all styles. Attempting to use a style that doesn't have all templates defined may result in an error rendering the page. Many commands take a ``{revision}`` URL parameter. This defines the changeset to operate on. This is commonly specified as the short, 12 digit hexidecimal abbreviation for the full 40 character unique revision identifier. However, any value described by :hg:`help revisions` typically works. Commands and URLs ================= The following web commands and their URLs are available: .. webcommandsmarker