=== Changes between 0.80 and 1.00 ===

BUG FIXED: Apache hangs!

Versions prior to 1.0 would ignore input coming in on STDIN while processing
the log files.  It turns out that this is a really really bad idea.  If the
webserver only attempted to log a couple of hits during that time, this was
fine, but if too many hits came along, the webserver would block!  Not good,
especially on the poor old SPARCstation IPC I run dreamsmith.org on, which
current takes about 10 minutes to completely process the existing log files.
Now, LogJack calls select while processing the log files, and if it detects
input waiting on STDIN, processes the incoming log line before returning to
processing the files.

BUG FIXED: Doesn't work with access logs when ...

The old version assumed the current access log was as specified in
config.pl, the next newest was that plus ".0", and that subsequent ones were
".1.gz" and so on.  It turns out not everyone rotates their logfiles this
way.  LogJack now works no matter how many compressed or uncompressed access
logs you keep (as long as they all start with the same path provided in
config.pl).  It is, however, now dependent on zgrep (which is nice enough to
intelligently determine which files are compressed and which aren't, unlike
zcat which is no longer used).  An unfortunate side effect is that
"log.ignore" is no longer an optional file.

BUG FIXED: Generates too wide tables.

Within the past couple of months I've started to show up in all the major
search engines.  Because of this, I now frequently get referrer links that
are rather long.  This caused the generated tables to be too wide to display
on any existing monitors.  I've added code to insert small spaces into these
long URLs in order to allow the web browser to wordwrap them.

FEATURE ADDED: URLs are now hyperlinks.

Previously, the report tables just contained text.  Now, the file,
filebydate, uri, and referrer reports contain links.  This is really nice
with referrer since it enables you to easily browse the page that referres
to your site.  Also, it attempts to identify hyperlinks in the agents.table,
since some agents (particularly bots) include a hyperlink there.

FEATURE ADDED: The "remote" report can invoke a CGI script.

Like most people, I simply record IP addresses for the remote hosts, but
bringing up a command line and typing "host <ip address>" every time I want
to see what domain name these correspond to is annoying.  You can now
specify a CGI script to be linked to inside the remote.table that will do
hostname lookups for you.  I'm also including a quick-and-dirty hostname
lookup script this this release.

FEATURE ADDED: New optional agents.ignore file.

I was previously ignoring certain bots by IP address, but this is so much
better.  Simply add an entry like "^Googlebot/" to your agents.ignore file
and you no longer count hits from the Googlebot in your stats, no matter
what IP address it's coming from.  (That was always possible using
log.ignore, which can theoretically filter anything, but this way is a lot
easier.)

=== EOF ===