Re: Raining Data's offer to upgrade U2 customers to D3 for the cost of a Software Maintenance Agreement

This is amusing because the same discussion could have taken place
about 8 years ago - misconceptions and all. Where to start?

Peter, I'm sorry but based on this thread and many others it's obvious
that you're quite confused about how D3 and many other things work.
My recommendation is to input and process a good helping of current
data every day.

FSI doesn't have GFE's but there are remote file system errors and
other anomalies.

VME shouldn't hold application data in Win32 for many reasons
including volatility. If your VME is over 100MB it's way too big. I
only use the VME to test how VME and FSI do things differently, not
for apps/data if I can help it.

(Note the paradox, VME is volatile and FSI has anomalies. After so
many years of this I've had to re-evaluate my appreciation for D3.)

The idea that the FSI is bulletproof is generous and ignores the fact
that to this day the FSI doesn't support many features of the VME.
Would you rather be bulletproof and missing required functions, or
have the functions with bugs? I dunno - Six of one...

VME in *nix can be expanded by simply appending new volumes. It's as
easy as NT VME.

Option statements have always been with us. Confusion in emulation of
MV features? Where? I hate switches too but these products are full
of them. It's not that I don't trust them, I just don't like them.
My solution: get over it.

D3 confused between *nix and NT? No, D3 isn't confused. I think
you're still hung up on that damned unix.h, aren't you? Holy cow,
reminds me of people in the "deep south" of the USA still seething
over the civil war. Can you guess my solution? Get over it.

D3 FSI was supposed to replace VME for *nix years ago but they have
too much code that relies on VME "services". The big change was
supposed to be in D3 8.0, and all product changes were being driven
toward that release. When they realized this couldn't be done they
made the decision to roll back v8 fixes and enhancements into v7.
While v7.5 has had more bumps than it should, this is the first
release to incorporate many of those v8 changes, minus the VME/FSI
stuff. Technically it's a major release but with a minor number.
They could easily have rolled all of those changes into a major 8.0
release and put the FSI changes in a 9.0 but I think it was a good
move to allow people to get all of the benefits for free just by being
on a current support contract. I know that will stir some wallets but
is there any need to argue when you got something for free?

Anyway... the D3 FSI has a lot more potential than the other MV
implementations which bring the DBMS more inline with the OS. While
being close to the OS is a good thing (everyone has DIR files and
jBASE is as much a part of the OS as the OS itself) the D3 FSI was
designed to go much further to become a network layer for distributed
data and distributed processing. Is that a good thing? I dunno if I
was ever really sold on the concept, but that ideal is ages ahead of
the common thinking that the FSI is or should be just like what
everyone else has.

Forget the obvious VME/FSI differences, from an engineering
standpoint, one of the big drives was that moving to the FSI would
open development doors that have been closed for years due to
dependencies on and in the VME. The cludgy ODBC interface would be
one of the major benefactors of such a change.

Will RD ever get rid of the VME? No soft spoken high-level exec will
know any more than the rest of us. The phrase "or die trying" comes
to mind and considering what an effort this is, it's more likely that
the company itself will change direction before this major aspect of
the product changes to a point of stability. The problem with D3
isn't technical, it's the company that owns it. A change in corporate
direction will have a more profound effect on the future of D3 than
the questionable goal of dumping the VME.

Nuf outta me. Please feel free to use slightly larger words, mommy
got me a dikshunary.

T

.