On Aug 6, 2004, at 4:20 PM, Richard Huxton wrote:
> webb wrote:
>
>> 1. Is there some literature on this, so I don't have to keep
>> bothering the list with beginners questions?
>
> Temporal databases - replace mentions of timestamps with
> version-numbers and it matches your requirements precisely. There's
> various stuff on the web, plenty of academic research and a couple of
> books too.
This is an area I'm very interested in, in particular Nikos Lorentzos'
IXSQL (Interval Extension to SQL) work proposed during the TSQL2
discussions.
<http://portal.acm.org/citation.cfm?id=627848>
(I haven't found the full text of this paper available freely online,
though it can be purchased from the publisher for download.)
Further work along these lines was published in "Temporal Data and the
Relational Model" by Hugh Darwn, CJ Date, and Nikos Lorentzos. An
overview of an early draft of the book can be found here:
<http://www.hughdarwen.freeola.com/TheThirdManifesto.web/
TemporalData.Warwick.pdf>
Valid-time and transaction-time proposals that appear to specifically
address the "timestamping" issues (similar in spirit I gather to the
"time travel" feature in older versions of Postgres) appear to be
limited to only that specific case, while a more full approach as
proposed by Lorentzos et al would allow versioning and timestamping and
much more. The following is a link to a critique by Darwen and Date of
various TSQL2 proposals.
<http://www.hughdarwen.freeola.com/TheThirdManifesto.web/OnTSQL2.pdf>
From what I gather, the SQL TSQL2 discussions ended without any
conclusion as to extending SQL in this direction.
Michael Glaesemann
grzm myrealbox com