Address for an index Search
If a given hypertext node is an index, or the server has an index
associated with it, then a search may be done on that index by suffixing
the name of the index with a list of keywords, after a question mark:
<![CDATA[ address_of_index ? keywordlist
]]>The address of the index is a normal hypertext address. In the keywordlist,
multiple keywords are separated by plus signs (+) . (See BNF syntax
description .) The resulting string still does not contain any spaces.
It may be considered to be the hypertext address of a document which
is the result of making the keyword search on the index. Normally,
if the search was successful, the document returned will contain anchors
leading to other documents which match the selection criteria.
The search method, and the logical and lexical functions, weights,
etc applied to the keywords will depend on the index address. One
actual index may have several hypertext addresses, which when searched
on will behave in different ways. For example, one may allow a search
on author-given keywords only, while another may be a full text search.
These things particular to an index should be descibed in the hypertext
page for the index node itself (or in linked documents). For example,
a server may allow specific boolean search combinations may be represented
by the words "and", "or" and "not".
Example:
<![CDATA[ http://cernvm/FIND/?sgml+cms
]]>indicates the result of perfoming a search for keywords "sgml" and
"cms" on the index http://cernvm/FIND/.