I was going to clean this up and found that there are two different
URL_SHORTENER rules defined:
khopesh/20_khop_experimental.cf:uri URL_SHORTENER
/^http:\/\/(?:bit\.ly|tinyurl\.com|ow\.ly|is\.gd|tumblr\.com|formspring\.me|ff\.im|youtu\.be|tl\.gd|plurk\.com|migre\.me|j\.mp|cli\.gs|goo\.gl|yfrog\.com|lnk\.ms|su\.pr|fb\.me|alturl\.com|wp\.me|ping\.fm|chatter\.com|post\.ly|twurl\.nl|tiny\.cc|4sq\.com|ustre\.am|short\.to|u\.nu|flic\.kr|budurl\.com|digg\.com|twitvid\.com|gowal\.la|om\.ly|justin\.tv|icio\.us|p\.gs|loopt\.us|tcrn\.ch|xrl\.us|wpo\.st|bkite\.com)\/[^\/]{3}\/?/
wtogami/20_url_shortener.cf:meta URL_SHORTENER
__URL_SHORTENER_NUMS || __URL_SHORTENER_ABC || __URL_SHORTENER_DEFGHI
|| __URL_SHORTENER_JKLM || __URL_SHORTENER_NOPQ || __URL_SHORTENER_RS
|| __URL_SHORTENER_T || __URL_SHORTENER_UVWXYZ ||
__URL_SHORTENER_TINYARRO
Both are explicitly labeled nopublish, but Warren's appears to have
been disabled from day one.
We apparently still have some problems with nopublish not always being
respected...
nopublish doesn't apply if a rule is pulled in because of a meta.
Also, there are several rules dependent upon URL_SHORTENER yet the
build was only complaining about one. Does the test only report the
first dependency error?
KAM, SPOOFED_URL_HOST (and the others dependent on URL_SHORTENER) are
in your sandbox and it says it's based on khopesh's URL_SHORTENER, so
I'm going to split khopesh's into __subrule and scored-rule and fix
the dependencies to unwedge the build, and you, darxus and Adam can
look at it in more detail at your leisure.
http://svn.apache.org/viewvc?view=revision&sortby=date&revision=1432712
Thanks.
Overall, what was confusing me is that on my system it passed the test.
What's different about Jenkins so I can test better next time do you think?
