FW: [OAI-implementers] Open Archives Initiative Protocol for Meta data Harvesting Version 2 news

Alan Kent ajk@mds.rmit.edu.au
Fri, 8 Feb 2002 10:44:25 +1100

On Thu, Feb 07, 2002 at 03:06:48PM -0800, Walter Underwood wrote:
> There should be some way to get the current time at the repository.
> Clock skew will cause nasty problems in time-based harvesting. The only
> safe solution is to always use the clock at the server, and to require
> that it is non-decreasing.

Not letting lack of knowledge from stopping me from diving in
immediately :-), I had thought that it would be good if for every
request, the server also returned a guaranteed date range that the
fully covered up to. If it was not the end of the day yet, then it
would take that into account to make sure nothing was missed
(possibly resulting in the same information being sent more than

The reason I thought this would be useful relates to resumption tokens.
If 10 requests are done with resumption tokens then something messes up,
then at client can work out a safe time to resume from without having
to go back to the very original date. If the server returns the records
in date order, then recovering during a big download would be more
efficient. If the records are not in date order, the server would
return the 'from' date every time (meaning recovery would just start
again from scratch). Making the extra <resumptionFromDate> or whatever
optional means clients could optimized the recovery if the server
provided the details.