At 01:57 PM 15-10-02 +0800, you wrote:

>The 4th paragraph in this article:
>http://www.webtechniques.com/archives/2001/01/lilly/
>says:
>
>"... With few concurrent users, InterBase is fast on simple
>reads and complex joins, but its performance drops sharply
>under the stress of multiple queries and numerous concurrent
>users. Because InterBase uses a nonshared architecture, as
>user numbers increase, it must parcel data into ever-smaller
>partitions, diminishing its performance levels. This major
>flaw will likely be addressed if InterBase attracts skilled
>open-source developers. However, the program is new to the
>open-source world and still lacks the support of a strong
>developer community."
>
>How accurate is this statement? I've always tried to search
>for a PostgreSQL vs. Firebird comparison and that is the
>first I've found.

As I recall, this came from an article written by a guy (Patrick-someone?)
who had been commissioned by PG to compare PG (or was it MySQL?) with a few
other selected RDBMSs. It's old (about 2 years) and the author never did
explain what this statement was meant to mean:

>Because InterBase uses a nonshared architecture, as
>user numbers increase, it must parcel data into ever-smaller
>partitions, diminishing its performance levels.

"Because InterBase uses a nonshared architecture, as
user numbers increase, it must parcel data into ever-smaller
partitions, diminishing its performance levels. "

Many asked him what was "a nonshared architecture" and where he got the
idea that adding users (or queries) caused some kind of repackaging of data
packets. (It doesn't. IB's data packets are always the same size..) At
the time it was foo-bar and nothing has changed...

To date, no proper comparative evaluation has been published re IB or
Firebird vs other databases. One appeared where a whole bunch of RDBMSs
had been set up in sub-optimal configurations to compare single-user
performance with MySQL in its most optimal setup. At the time, the test
used the flaky IB 6 binary that Borl released in July 2000 with an old IB 5
ODBC driver. As I recall, it did quite well (even if Saddam Hussein did
win the election!)

>So far the ff seem to be the case:
>
>Footprint: Firebird is much(?) smaller than PostgreSQL
>while providing a feature set that closely matches it.

True; and it's more the case that PG developers have been working hard
over the past two years to try and mimic IB/Fb features.

>Ease-of-setup/deployment:
>Firebird is definitely easier to deploy under Windows
>than PostgreSQL which requires futzing around with Cygwin
>first; Under Linux, I have found Firebird to be quite easy
>to set-up (although I doubt that would be the case for
>a newbie who doesn't know how to read bash scripts or
>setup rc files). Linux PostgreSQL setup could be even
>simpler as it comes as a package with many distros.

Firebird has official RPM installations that I have never known to fail. I
certainly never had to modify any scripts or set up the rc files or
symlinks. There might be more work to do on a Linux that doesn't support
RPMs though...I consider this to be more a problem for a newbie to
Linux. A newbie to Windows (coming from Linux) will have trouble logging
in to a windows-based server if he doesn't know that the Windows
Administrator user is quite a different kettle of fish to the root user on
*nix...

>Win/Linux/*nix interoperability:
>As far as I can tell, Firebird wins this hands down.

Yea, as to the manner born. Gbak -t on one platform, restore on
another. Log in and carry on.

>SQL dialect:
>Regarding SQL dialect, I find that MySQL has a friendlier
>dialect and more polished/convenient built in user functions
>than Firebird. However, the MySQL dialect is seriously crippled
>in some critical aspects (lack of SQL features like subqueries,
>triggers, views, SPs, etc...).

MySQL isn't very conformant to SQL standards, either, which makes it not
very portable for upscaling to commercial competitors. However, comparing
MySQL with IB/Fb on Linux is like comparing Paradox or Access with IB/Fb on
Windows. MySQL does what it does quite well for its operating
environment. It lost a lot of Brownie points for its fragility as a web
backend on Sourceforge. It was an unfortunate reference site from MySQL's
point of view. Sourceforge now uses DB2, in whose class IB and Fb do belong.

>As I got used to it after a couple of days (not realizing that
>you have to quote object names if they have lowercase
>identifiers seriously tripped me up in the beginning), I found
>the basic Firebird SQL dialect to be more or less OK.

You may be giving yourself more grief than you need to. If you don't need
case-sensitive identifiers in your database, you are making a lot of extra
work for yourself by imposing that restriction. Of course, if you have a
legacy database where you rely on having a table called "Fred" and another
table called "FRED" then, yes, you are stuck with quoted identifiers.

>I have no idea what the PostgreSQL dialect is like; anyone
>care to add comments wrt that?

No, can't comment.

>Admin tools:
>Firebird's Windows-based tools should be more polished than
>any Linux-based ones you can find (Quickdesk, minor bugs
>notwithstanding is very nice) for any of the three OS
>databases; For Linux based ones, it's probably a toss-up
>with PostgreSQL maybe having a bit of an edge.

Can't comment. Choice of tools is usually pretty subjective. My personal
preference is for tools that ride close to the API and to script all
metadata manipulation; others crave tools like Access' query
builder. With PG having had Linux OS all to itself for many years, I too
would expect its tools selection to be pretty reasonable.

>Features: PostgreSQL has Object-Relational, Firebird
>doesn't. Aside from O-R, I get the impression that PostgreSQL
>has a bit more extra advanced features, but the question is
>are they worth the additional bloat (and how much bloat
>do these actually add?).

If O-R is a requirement, and PG can provide the level of support the
developer requires to implement the design, then PG would be among a range
of choices of O-R databases and IB/Fb wouldn't.

>I think a Firebird vs. PostgreSQL is as important as, if
>not more so, than a Firebird vs. MySQL one. People who
>are thinking of using PostgreSQL are likely to give Firebird
>more serious consideration than MySQL.

That's true; but is there really anything to gain from starting
internecine wars between open source databases? Surely the real decider of
which database to use is how well it does the job you ask it to do. The
beauty of having PG and Fb among your options is that you can set up test
demand scenarios with fully functional versions of the software at no
cost...no need to rely on the necessarily biased claims of competing
commercial vendors.

heLen