-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David Forslund wrote:
| I haven't used python very much, but I don't see the problem with Java
that
| people are talking about.  Switching to different J2EE servers has nothing
| to do with Java itself, but with different environments that these systems
| provide (perhaps a lack of sufficient standardization in J2EE?).
|
Let me try another stab at clarification:

If you write just one application and run it on just one application
server things are very peachy.  As Dave say's an 'environment' builds up
around your work as you use bits and pieces from other projects.  This
is sort of like a linux distribution.  The external components were
built with their authors 'environment' in mind.  You end up having a
chain of dependencies.  Moving your 'project' to another J2EE
environment easily depends upon you being self-contained within the J2EE
specification and not having external dependencies.  Anyone who has used
linux for a few years must be familiar with this story about keeping all
your dependencies in order.  Most distributions have an elaborate
mechanism for checking dependencies and automatically attempting to fix
them.

Here is an example:

Recently, I decided to upgrade to Evolution 2.  I started looking for
the components for my relatively recent Fedora Core 2 distro.  Well, in
a very little time I discovered that Evolution 2 needed a more recent
Gnome then I had.  I looked into what that would take and the simplest
path was to wait and upgrade to Fedora Core 3 which has all these parts
integrated in.  Such a massive upgrade of components within my working
distribution as it was running was something I did not wish to undertake.

So, what happens in a 'server farm' that services several developer
groups and commercial applications?  Well, each of them has made
assumptions about the 'environment' that surrounds their execution
environment.  Some of these assumptions are incompatible with each
other.  Others are just different, i.e. they could be made compatible,
but that requires some work on out part, which we find is not warranted
for these applications.

Why is that work not warranted?  Because every time you go back to the
vendor for support, you have to waste effort explaining and proving that
your deviation from their 'environment' is not the cause of the problem!

And so I come back to my initial assertion: It is the distribution of
all components, integrated by a vendor, needed to support an application
that someone has written, which is an essential part of paid for support.

The concept of support for LAMP and LAMJ is based around this concept, I
believe.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFBkhvjY+HG7UEwVGERAlidAJ4/4u9Gdj1dV09cQhV1EadDvKqgUACeM3II
/r8mw9p3++7pV5lv/Suy/Oc=
=Ijct
-----END PGP SIGNATURE-----



Reply via email to