I'm sorry to disagree. Everything you noted above is true, although I don't think the documentation is bad! Scarce maybe, but the articles and books that I've found did a good job of explaining the mechanics of a cookie and what the browsers don't do (but should). The one thing you forgot to mention is that when you set a cookie's host, you MUST access the host using the same name.
Ie: http://foo/ ne http://foo.bar.com/

I started using cookies with cookie-lib.pl which is a cookie library (doh!) inspired by cgi-lib.pl. That was a mistake! Many of the odd features that I blamed cookies and browsers for ended up being faulty implementations on the lib I was using. Once you fully understand the consequences of keeping user state (since HTTP still is a stateless protocol) through cookies, you can plan ahead, test, re-test and make sure that your applications run smooth.

Without cookies, my coding would soon become a living nightmare. cookies++ !!

UPDATED:
BTW: You can always inspect your cookies by doing something in the lines of:

# this line is from perldoc
my $c = new CGI::Cookie(-name    =>  'foo',
                        -value   =>  ['bar','baz'],
                        -expires =>  '+3M');
print "Set-Cookie: $c\n";
print "Content-Type: text/html\n\n";
print "<!-- inspect your cookie $c -->\n";
#etc...
[download]

... and that would keep you from having to telnet to port 80 to retract your cookie info. (make sure to remove the comment line on production code, it looks odd)

#!/home/bbq/bin/perl
# Trust no1!