Update automation and related issues (was: Python Webstart ?)

[Cameron Laird]

In article <87vfe8zlgv.fsf at pobox.com>, John J. Lee <jjl at pobox.com> wrote:
			.
			.
			.
>Not sure, as a user, that I'm a great fan of the mutable
>program-plus-data-in-a-big-lump concept...
>
>- One can usually rely on data files to stay put, and not to start
>  rewriting themselves (well, ok, at least on non-windows OSes ;-/).
>
>- Do I want to be forced to back up 10Mb of Python+Metakit every time
>  I back up my data?
>
>- Other than that, as a user, why do I care?
>
>As a developer, what does this approach offer above tools like py2exe
>/ Installer &c?
>
>I am underwhelmed by the idea of dynamic debugging.  Seems like a
>solution looking for a problem.
			.
			.
			.
John, those are healthy instincts.

It'll take a while to answer fully.  Part of the answer depends on
this:  10 megabytes is inconsequential.  That's false, of course; I
often work on old equipment, through dial-up, and so on, so I know
that bytes count.  For much of current computing, though, 10 Mb 
doesn't matter (and incidentally, my deliverables frequently fit on
a floppy).

One big benefit is what Metakit's inventor calls freedom from
installation.  The point is that an application is a single file.
If you want it, you have that single file.  If you want to discard
it, you remove a single file.  End-user configuration confusions
are so common and costly that this consideration alone makes big-
lumps a win (and keep in mind that, as I wrote above, I often fit
my big lumps on a floppy; compare that to typical VC++ or Java
practice).

So, as a user, you care because the model is simple for you.  There's
one file.  Back it up, move it, delete it--the concept doesn't com-
plexify.  It's the way computers are supposed to work.

More, later.