On Oct 7, 6:14 pm, Istvan Albert <[EMAIL PROTECTED]> wrote:
> On Oct 7, 12:24 pm, Michele Simionato <[EMAIL PROTECTED]>
> wrote:
>
> > Here we disagree: I think that a programmer should know what he
> > is using.
>
> My point was that they should not *need* to know. Too much information
> can be detrimental.
Actually, in principle I *do* agree with you. For instance,
when I type 1+1, I have no idea of what exactly Python is
doing, but I don't need to know, because adding two integers
is a problem which is well understood and has been solved by
others years ago, I don't need to care about it. I wait with
impatience the time when Web programming will become a solved
problem with a standard built-in solution that works.
Unfortunately, that time is not near in the future and for the
moment I am forced to know how all the pieces work if I want to
build a reliable application. I don't like it, since I am not
particularly interested in the low levels and in the plumbings
myself, but this is the way it is.
> ... truth to
> be told I was just curious of how they would go about integrating a
> database and still keep it "pluggable".
>
> Because it would either need to know about the way the server handles
> the database connections (which makes it application specific) or it
> has to create new database connections in which case it would be
> duplicating the functionality leading to all kinds of headaches.
>
> So there, even a simple requirement exposes the fallacy of building
> anything complex out of "pluggable" components.
I suppose Beaker rely on SQLAlchemy since it comes from the
same mind, but I don't see the issue: a component may well
depend from another one, partially or totally, but this
does not detract at all from the usefulness of the component
approach. If I don't need persistent on a database, for instance,
I like to have the choice between a full-fledged component
and a poor man component that satisfies my basic requirements
and nothing more. I don't want to be forced to use a
monolithic-good-for-everybody solution.
Michele Simionato
--
http://mail.python.org/mailman/listinfo/python-list
