Back in June 2003 when I was looking into this, I also found
Tiger,
TimeDB,
Espresso, and
JTemporal,
all of which look like emulation layers of sorts. TimeDB is clearly
closed source, I'm not sure about Tiger. Espresso and JTemporal are
open source. Tiger and TimeDB both appear to be inactive.
Around that same time (June 2003), I also asked Rick Snodgrass about
this and got back some useful info from him:
I think the list of resources in your message is quite complete. I
don't know of any other efforts, except for our effort in the TAU
project (see
cs.arizona.edu/tau)
to add transaction-time support to the internals of BerkeleyDB and
MySQL. We're going for a production-quality system. Our alpha system
works well, but is not of sufficiently high performance, so we're hard
at work on making things run even faster. We plan on releasing the
system as open source by the end of the year.
[...]
There are no plans right now to support valid time. The support
generally occurs in different modules. For MySQL/BerkeleyDB, the bulk
of the tt support is in BDB. For valid time, the bulk would be in
MySQL. If our tBDB/tMySQL takes off, then we'll consider adding vt
support.
[...]
Yes, that's how we do it. Not that much has changed in MySQL, though
we are now considering ways to take a predicate in transaction time in
the WHERE clause and get it to BDB.
Note that they're reusing their transaction-time enabled BerkeleyDB as
the MySQL storage manager, which is probably the basic reason they're
using MySQL at all.
On the confused state of the Temporal SQL standards mentioned above,
he said:
That is still the state. Basically the US and UK delegations couldn't
agree, so the ISO committee gave up. I still have hopes that a DBMS
vendor will decide that temporal support is sufficiently important to
support natively; the draft standards provide a good design to
implement.
So, in case anyone else is ever interested in this temporal database
stuff, that's everything I learned!
