In message <9701131817.AA09140@jade.noc.dfn.de>, Joachim Schmitz writes:
Dear Cengiz,
you wrote:
Date: Mon, 13 Jan 1997 10:14:20 -0800 Message-Id: <199701131814.AA22919@cat.isi.edu> From: Cengiz Alaettinoglu <cengiz@ISI.EDU> To: noc@noc.dfn.de Cc: woeber@cc.univie.ac.at (Wilfried Woeber, UniVie/ACOnet), Carol.Orange@ripe.net, db-wg@ripe.net Subject: Re: Immediate - Date in Changed field
[...]
Hi DB-folks!
On a slightly different aspect, and assuming that we change things anyway...
How about running the RIPE-DB (and maybe gradually all the others :- ) on a UTC basis, according to some recent suggestions?
I think both the 4 digit year presentation and the UTC basis (I estimat e this includes hh:mm:ss as well) are a very valuable extension for time stamps in the RIPE-db. This might as well ease some of the troubles in updates within the "near real time" mirrors.
I definitely agree for the need for this. But the changed attribute is inserted by the person who registers the object, not by the database software, he has the full control over which timezone he uses and to lie.
May be what we really want is a timestamp field with the granularity you suggest and in UTC zone, but inserted to the objects automatically by the database software. I see a lot of value in this.
I completely agree with you. This can only be done if the database software sets the time stamp. I guess this also solves most of the race conditions otherwise implied.
This would most definitely be a significant improvement (but is a different point than the "changed" field) I believe this is on the the list of Open Issues to be reviewed next week. If not it will be then. Greetings, -- Carol