As of May 2009 we are running an experimental
link check in the early hours of each morning.
To enable it you must create a world-readable file
in the directory you
wish to be checked.
For a sensibly indexed directory a single linkcheck.txt
file will link-check that directory and all of its sub-directories.
linkcheck.txt has the usual sidebar-like syntax in that
options consists of a word followed by a colon, everything else is
taken to be a URL to be excluded from the check.
Example minimal linkcheck.txt file
# URLs not to check
Details of any broken links will be emailed to
(assumed to be you!).
If there are no broken links no email is sent.
You must specify an email address for the check to run.
# Only check links to depth n with respect to this directory.
# (I believe this refers to the number of links a page is
# away from the original, not the directory depth.)
# For links of the form:
# check that "fragment" exists (should this be the default?)
# Only run on the morning(s) specified.
# Verbosity: by default the link-checker runs in quiet mode
# (no error, no email), if n is omitted or zero, it runs
# in standard mode, otherwise runs in verbose mode.
# Disables the link check
When you first create the file
I would advise deliberately creating a broken link in one
of your web pages just to check it is behaving as you would expect.
The link checker is a slightly modified version of the
W3C linker checker