Cornelia Büchen-Osmond writes:

Date: Wed, 16 Aug 2000 10:18:02 +1000 From: =?iso-8859-1?Q?Cornelia_B=FCchen-Osmond?= <buchen@rsbs.anu.edu.au> To: TDWG-SDD@usobi.org Subject: Re: Creation/Modification times and Revision Numbers

Just a comment: I was recently told by an expert in this field that documents build on the fly are usually not indexed by WWW search engines and one had to have a stable document in place as well to be seen out there. Cornelia

----- Original Message -----

This is generally true, mostly because those engines have no particular way to construct a query that will induce the creation of the page. However, /other/ software, such as that which attempts to combine the data from several sources, may well wish to know which of those sources is more recent, whether cached data previously obtained from a source is stale, etc. For this reason, both a date (GMT of course) and revision number are useful. Ideally, these would be of very fine granularity, e.g. down to the field, but minimally they would be on the record. If you don't do them on the field it becomes more complicated to decide whether a datum from one database is more recent than that from another: Source 1 Record A: Item1(Jan 01, 2000) Item2(Feb 01, 2000) Source 2 Record B: Item1(Feb 15, 2000) Item2(Jan 15, 2000) Record B was modified later than Record A, but the datum in A-2 is the more current Item2. Revision numbers are not useful across sources unless the sources have a common revision number pool, which requires distributed revision control. --Bob Morris