http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5780
------- Additional Comments From [EMAIL PROTECTED] 2008-01-18 06:41 ------- (in reply to comment #12) In comment #2 I list exactly what OE 6 amd Thunderbird linkify. I agree that we should do the full processing including WEIRD_PORT on any URL that OE linkifies. What I'm proposing would affect only non-linkified strings. It does create a hole if we want to do more with type this in your browser: foo.com/bar/baz than just an RBL check on foo.com. If there is some rule that we really want to have see foo.com/bar/baz in that specific case, then what I'm proposing will not handle it. Is there such a rule? I've got some more questions about the existing code for someone who knowns more about the original intentions. Is that you or Theo? The code that adds a scheme to the URI pushes both versions of the string, with and without the scheme, on to the list. In your example, both "foo.com/bar/baz" and "http://foo.com/bar/bax"; go on the list of URIs. Is that really what is intended, or should it be only "http://foo.com/bar/baz";? Also, there are separate tests for URIs and email links, but the URI testing section of the code also looks for an '@' and generates a mailto URI using a somewhat different regexp than the branch of the code that is looking for mail links. That results in two mailto URI strings being pushed on the list, possibly slightly different. Is that a bug? I've got some proposed code changes but I want to resolve those issues first before I finalize it. Another question is how to test changes. This would affect overall performance. I could use help in figuring out from nightly run results what has happened once we get to the point of being ready to test a change. ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
