Re: query that Bob is descendant of Alice
From: Robert Vazan <robova_at_nextra.sk>
Date: Mon, 23 Apr 2001 22:56:44 GMT
Message-ID: <3ae4915f.1973751_at_news.nextra.sk>
Date: Mon, 23 Apr 2001 22:56:44 GMT
Message-ID: <3ae4915f.1973751_at_news.nextra.sk>
"fervvac" <fervvac_at_ust.hk> wrote:
>+) In this way, A is a parent of B iff A's (start,end) covers B's
>(start,end). If we have B+-tree built for start and end attribute, I guess
>such query can be efficiently evaluated.
>
>+) If you have a DAG other than a tree, it seems still possible to apply the
>similar scheme, but there are some thing to consider what value is there in
>the counter.
Tricky. Numbering for DAGs can be done the same way. But, in DAG, numbers of children of particular node get fragmented. I can store largest fragment or all. For complex enough DAGs, intervals get very short and very numerous, which makes both alternatives useless.
/Robert Received on Tue Apr 24 2001 - 00:56:44 CEST