Uses externally set environment variables to do get hold of
configuration data now, rather than hard coding the details.

HUB_CONNECTION_URI: The URI for server connections; e.g.
"drbunix:/path/to/socket/file"

HUB_PATH_PREFIX: Web server URI prefix to get to root
of Hub application, e.g. "/rails/hub".

HUB_RANDOM_FILE: Path to file containing random data.

Corrected stylesheet image path. Related to Ticket #96.

Produce a message indicating no search results if an unsuccessful
search is made, rather than just returning a blank search page as
if no search had been conducted. Fixes Ticket #89.

Increased login timeout to 60 minutes.

Protect contents of CVSROOT to fix Ticket #91.

Tweaked configuration to allow full year summary from the Web interface.

Empty forums were having their "empty" page returned with a status
code of 404, producing problems in O2. It might be an attempt to
stop cache-related problems but there are more elegant ways to do
that built into HTTP! Will have to keep an eye on this for a while
but, as is so often said - "seems to work fine" now.

Fix Ticket #93; date format was confusing. ISO-style "YYYY-MM-DD" now
shown instead. I *think* the short date format is only used in the
filtered ticket summary, since the date and time seem to be shown in
full elsewhere.

Fixed Ticket #95. Upper case 'y' in 'Yes' for account lists.

Now insists on being a Hub admin before allowing further access
to its own admin interface. Update to the breadcrumbs template
to alter views in line with recent RCVSweb changes - for details
see Changeset #144.

Updated to place public components in 'public' folder. The script
has been extended with various new configuration options and a new
configuration file added. Designed to be integrated with RCVSweb
from Changeset #144.

Various updates, mostly concerned with integration with
cvslog2web for faster and more accurate log processing. The
synthesised revision summaries from CVSHistory remain though
the user is shown cvslog2web output by default. General layout
improvements included. Requires cvslog2web from Changeset #145
(which will be committed shortly; the checked out copies do
not sit in locations that make an atomic checkin easy and at
this stage atomicity is not that important).

Adding directory to separate out files that the public might view
from files that we'd rather they didn't (at least, not routinely).

Initial import of cvslog2web, a(nother) CVS log analysis tool.

Added new attribute "escaped", which indicates that '%xx' hex sequence
escaping has already been carried out on links in the source feed so
should not be done internally. The I2 feeds have this, so were failing
when escaped characters appeared. Fixes Ticket #88.