Hi Lily,
In IRML, when a rule match is "case-insensitive", you mean the regular
expression matching is case insensitive, right? What about the name of the
header (property)? How does http handle that? For example, are both Cookie
and cookie the same?
Yes, the "case-sensitive" attribute refers to the rule pattern, not to
the property name. I know that at least some HTTP servers do case
insensitive pattern matches when they parse HTTP headers. So I'd say yes
"cookie" and "Cookie" are the same, at least it is not safe to assume
that HTTP servers differentiate between them.
2) We have "request-line" defined as standard irml header. We might consider
adding "url" to get the specific url out of request line.
So request-line is "GET http://www.intel.com HTTP/1.1" while url
is"http://www.intel.com";. It is more suitable for url-matching rules.
Originally I had thought that a combination of the IRML-defined property
"request-path" (the relative path component of the URI) and the HTTP
"Host" header field would be sufficient for pattern matches. However,
just recently a colleague of mine pointed out to me that the "Host"
field is only mandatory in HTTP/1.1 and may not be present in HTTP/1.0
requests.
Therefore I think we should definitely add an IRML-defined property
"request-host" to the next version of IRML. I also like your suggestion
of something like "request-URL" to facilitate URL pattern matches
(although it would be redundant). Comments?
-Andre