Is it just me, or is Sqlite3 goofy?

[Magnus Lycka]

While I can understand your frustration, I think it is
important to think about the tone in our postings here.
Hydrocephalus is one of the most common birth defects,
and it's not terribly unlikely that someone who reads
this has a family member or someone else in his proximity
who suffers from this condition.

mensanator at aol.com wrote:
> "Fixed"? Up until now, I didn't think it was possible for
> crackpot theories to be implemented in computer science.
> This is absolutely the craziest thing I've ever heard.

Still, many people with lots of experience in databases
use it, and prefer it for certain kinds of applications.
All systems have limitations and deviations, and those
limitations and deviations are stated more clearly for
SQLite than for most commercial products at least. The
market leader Oracle still can't store empty strings in
VARCHAR fields for instance. They are silently converted
to NULL. I'm pretty sure that has been in clear violation
to the official spec since 1986 at least.

As far as I understand, noone here is forcing you to use
SQLite, and with your long experience of MS Access I'd
expect you to be fairly used to "almost SQL"... It's
some time since I used Jet/Access now, but I had much
more problems with that than I've had with SQLite.

SQLite is built in Tcl, by someone who appreciates the
way Tcl works, with its weak typing. I don't think Tcl's
type handling is nearly as clever as Python's, but I
think it's a good thing that Python's standard lib finally
has a DB-API compliant module, and while I would have
preferred something that was closer to standard SQL, I
don't know of a better candidate than SQLite.

It's good that it's usable without a server setup, and
that it's very light weight. A Jet engine is obviously
not an option, and I would have preferred SQLite even
if Jet was open source and worked on all platforms.
(Well, if JET *was* open source, I suspect it would
have been fixed by now.) It's possible that one could
have used the embedded version of Firebird instead, but
in my experience that's not nearly as lean or easy to
deploy.

With your long experience of Access and SQL Server I'm
sure you know well that any attempt to build a working
database application requires extensive knowledge of
the backend to understand its peculiarities and
limitations.

The list of software projects where not quite competent
developers built Access applications that worked ok in
small scale tests and failed catastrophically in real
life is looong...

Of course, if you've stayed with one vendor for 15 years,
I can imagine that you've forgotten how long it took you
Having worked with half a dozen backends or so, I'm no
longer surprised that SQL can be interpreted in so many
ways... I agree that SQLite is unique in it's approach
to typing, but if you are aware of this, it's really not
a big problem.