At 11:04 AM +0100 3/6/00, J�rgen Hoffmann wrote:
>If I understand this right the search words are keys in word.db. A query
>on word.db yields a document id which in turn matches a key in
>docs.index. This yields a (coded)URL which matches a key in docdb and we
>get the document we want!
>Why do we need the docs.index?
This is how it works in versions 3.1.x and earlier. In version 3.2.x,
we don't use the docs.index. The word DB returns the DocID, which is
the key for the doc DB. We only need the docs.index for indexing,
when we must turn a URL (from a link) into a DocID.
>A second question deals with the choice of db type. If I understand this
>right all three dbs are of type B+Tree. A db with unique keys, like the
Again, this is how it works in version 3.1.x. In version 3.2, we've
switched all of the databases *except* the word DB over to DB_HASH.
The word database must be a B-Tree so we can do prefix searching and
other word retrieval techniques.
Thanks for the good questions!
-Geoff
------------------------------------
To unsubscribe from the htdig3-dev mailing list, send a message to
[EMAIL PROTECTED]
You will receive a message to confirm this.
