Shibboleth Developers

["Scott Cantor" <>]

This seems to be working fairly well and has some interesting side-benefits
some people have asked for previously.

I've added HostRegex and PathRegex elements that work alongside Host and
Path as follows:

1. Host and path matching are still distinct (so both types of host matching
elements can contain any kind of path rules).

2. Host matching is favored over HostRegex matching.

3. HostRegex matching is done against the entire scheme, host, and port, so
it's kind of like the servlet dispatching map in the IdP, you have to
include it all.

4. Path matching works as before, and is favored at a given level of the
tree over PathRegex matching, as I suggested earlier.

5. PathRegex matching is always performed against whatever remains of the
request URI after all successful Path matches occur, using any elements
found inside the last Path element matched.

6. Regexs default to case insensitive, but there's an ignoreCase option.

7. PathRegex elements are terminals, they can't have children.

8. The query string is NOT included in Path matching (as now) but IS
included in PathRegex matching.

9. If you consumed the whole path before regex matching starts, the query
string would be applied against any PathRegex rules, assuming those rules
started with "^\?" to consume the query separator.

This last feature is quite nice and seems to work well in my testing, and
would be where the case switch might be helpful.

Comments or suggestions welcome. I will most likely be able to backport this
to 1.3.1 as well, haven't tried yet.

-- Scott