VJournal is a simple blog engine with extensive commenting feature. 
It's commenting style more or less mimics LiveJournal style blog sites,
but it is designed to work on small sites with few blogs (or even just
one)

It doesn't use any relational database, only Perl builtin dbmopen
function (for information which shouldn't be directly shown to the
users).

Blog (with discission) is kept in the static HTML files which are
updated each time comment is posted. VJournal tries to limit number of
files to be updated on each comment.

It updates just

1. Main file of the  post where comment is stored.
2. Calendar files for month and 
3. day when post was posted, to update comment count
4. Main page of the blog, if post is recent enough to be displayed.

VJournal doesn't keep user authentication information. It relies on
OpenID to authenticate all users, including blog owner.

However, VJournal includes its own OpenID provider to autenticate blog
owner. Blog owner is assumed to be a user, whose OpenID url is same as
top-level URL of the blog, unless other OpenID url is specified as
-owner option in the config.


File structure of the blog:

Top level directory have index.html page which contain several recent
posts, and set of directories with older posts by year and month:

In the top level directory there is .vjournalrc file which is just
serialized perl hash for configuration options.

This hash should contain at least following options:

-owner - openid URL for blog owner
-statedir - directory outside of public tree serverd by webserver, which
contains following files:
	users.db - cached information about openid users
	session.db - information about user sessions
	owner.pwd - hash of owner password.
Optionally, there also can be file openid.lst, which contain infomation
about known openid sites. This information should be in the form 
site name:url_template.

There probably should be format specifier %s in the url_template, which would be
substituted by user name. But some sites don't need this. For instance,
google uses same url for all users.

Theoretically it is possible for several blogs to share same user cache
and session cache (LiveJournal does so for millions of blogs), so
owner.pwd can contain more than one record.

This file can be created using apache htpasswd utility (or simular)
using blog URI (without protocol prefix) as user names.

It is recommended to deny HTTP	access to .vjournalrc, using web server
access control, although filesystem permissions should allow this file
be read by CGI process.

Inside publically accessable HTML tree should be subdirectories for the
templates and user avatar pictures. By default these are subdirectories
templats and avatars in the toplevel directory of blog.
It can be redefined using options -templatedir and -avatardir in the
config file.

If several blogs share user cache, they should share avatar directory
too.

Initialization:

1. Install perl scripts and modules (once per all users)
2. Create top level directory for blog, and put .vjournalrc there.
Make directory writable to web server.
3. Create empty avatar directory and make it writable to the webserver
4. Create template directory and populate with templates.
5. Create state directory and make it writable to webserver.
6. Go to the /cgi-bin/post/urltop/, where urltop is top-level directory
of your blog (may be it is just / if your entire site dedicated for
blog). It is important to type into this url such path_info so webserver
would put path of the directory with you .vjournalrc into
PATH_TRANSLATED. You would be able to set owner password and upload your 
avatar.

Once you set up owner password, access to the post script would be
resticted for blog owner.