""Artur Anjos"" <arsoft@...> wrote in message
news:003001c1b097$1736c460$0202a8c0@......

>
> Hey, Claudio! It was a very nice and precise description of the pass.

Let's

> work it out into the future. That's why I think that sometime, somewhere,

we

> must split out IB/FB. Maybe the little help of Borland could bring the
> people that are using IB now to move to Firebird, and give the project the
> money it needs.

Artur, I was not trying to demote the own effort where I participate, but
one has to be realist with a rational optimism and probably most
small/medium IT shops doesn't care if the product is open source or maybe
they enjoy it being open source, because they don't have to pay (they can
pay if they want to cooperate) and in an extreme case, they could introduce
a change for their developments.

However, companies of some moderate sizes have strong policies about what
can and should be purchased, or a contract with MS or Oracle or IBM or
Borland that they won't throw away or managers that won't risk their
lucrative position by betting on a solution that could disappear as a
product (and they will be fired and lose reputability), etc. A friend mine
said: the big dog is fine in the brotherhood of other big dogs (provided
they are friends) to defend mutually from other dogs. In more civilized
words, if you are a big company, usually you look for partnership with
another big company, not with the small shop running in a garage. Maybe MS
is the exception, that partners with small companies then Gates buys them
after a few years, but that's another business model. Borland only did that
with Ashton-Tate, Visigenic and recently with a Brazilian company.

Go to Ford and tell them to replace their systems with FB. Let's assume Ford
bites and looks for the options. They learn that IB and FB are based on the
same base architecture but who's behind IB and behind FB? Where's the
relatively big company supporting the product? Does the financial report
look good?

So, one of the issues is where do you go to peddle an open source database.
A database is such a critical beast that I'm horrorified when I remember
that Mike Nordel found things like this in the code, for the C people:

if (!bitfield & mask) ...
=> bitfield is an integer used to hold a bunch of flags, each one
represented by a bit that's set at one or zero and mask is obviously a
predefined set of bits that serves to check only for the activation of the
desired flags in bitfield.

Who wrote that? A newbie? A careless professional engineer? Certainly this
comes really from
if (bitfield & mask == 0) ...
but some person says, okay
"something == 0" is like writing "!something" and the code has changed
meaning, because it should have been instead
if (!(bitfield & mask)) ...
At this time, I prefer the explicit comparison with zero for readability.

If those things happen at Borland, what can the IT manager (used to
commercial SW where a company takes full responsability for it although it's
BS because you read "no liabilities" and "only 3 months warranty") expect
from a bunch of developers that don't work for a company that can be legally
held responsible in case of a fundamental failure in the engine? IBPhoenix
is not FB and FB devs aren't IBPhoenix employees, with the exception of the
IBPhoenix people that are FB contributors. How do you convince that honcho
that the current devs won't abandon the open source project? Look at SF,
there are a lof of projects, so why doesn't SF explode? Simple: because
there are lot of one-person projects, abandoned projects, zero percent
activity projects, etc. If company X the size of Sybase promises and new
version of its flagship db and abandons the engine and leaves customers in
the dust, those customers can sue and squeeze money from X. How much money
can Ford squeeze from IBPhoenix if FB devs don't deliver FB5 in time? Can
IBPhoenix be sued for what FB devs do? Certainly, I expect any moderately
big company to be extremely cautious in the steps they do, for obvious
reasons: the bigger you are, the worse the disaster if you fail.

What I'm trying to say is that the perspective of a small or medium shop is
not the same perspective than a big shop. The big shop is using Db2, MsSql,
Oracle or Teradata. They rely respectively on IBM, MS, Oracle Corp and NCR.
The big shop won't replace those engines unless they were going bankruptcy
and trying to save up to the last penny in operational costs. (And what
about their internal systems based on those engines?) Furthermore, neither
IB not FB can take on those huge loads. Do you afford to wait for a sweep on
a 1.2 TB database? What if you have to backup and restore immediately to
check your backup is correct so you need additional 1.2 TB plus temporal
space for the big sorting to regenerate the indices, just because restoring
over the original db is highly dangerous? Would you wait 5 hours for your
trends report that crunches 760 GB of raw data to produce a decision cube,
because IB/FB lacks the OLAP module present in Sybase, MsSql, DB2, Oracle
and Teradata?

Basically I'm saying the same than in my previous post: don't go to places
where you know ahead that the FB engine will be a loser. Target honestly the
cases where the FB engine can shine or do at least a job as good as the
engine being replaced. But don't bite PostGres or MySQL, because they are
peers in some way and their success help us indirectly. Of course, I expect
someone to work on the tests those people used to show FB as the worst
performer, but un-biasing the benchmarks a bit is not the same than
declaring war on those open source fellows. Furthermore, does anyone really
expect a mandatory transactional engine to surpass a transaction-less file
manager like Paradox and MySQL and in single user mode?

When the Advantage Database people did a benchmark to show how fast is their
offering, their stated conclusion with numeric results was that (of course)
they are the winners in performance, second is IB and third is MsSql. Try to
guess how they did the benchmark. It's strange, because in MsSql (unlike FB)
you aren't obliged to start a transaction to do work and in single user
mode, MsSql disables several checks so it runs really fast. So, how much
credible is the aforementioned benchmark? Do you rely on benchmarks or in
real life cases?

Yahoo is using MySql. Spamcop is using MySql. Other sites use PostGres.
Which known site uses FB? I mean which site that has importance outside the
FB/Delphi/BCB community. Maybe Bill Gates uses FB in his desktop to not
fiddle installing the personal version of MsSql?
:-)

C.
--
Claudio Valderrama C. - http://www.cvalde.com - http://www.firebirdSql.org
Independent developer
Owner of the Interbase® WebRing