Q 4 Mr. Ford: migration to more sophisticated web app

[Gian Franco]

Dear Mr. Neal,
Welcome in this forum first of all.
The topics covered by your book are right on the
spot with regards to the issues that are playing
right now in the project I'm working on, so I'm
sure that amongst the interested contestants some
good queries will pop-up.
Let me start with my first query:
I read the description of your book, and was wondering
whether it gives an indication of how one can migrate
from an implementation choice to another (more sophisticated)
one.
This issue arised recently with a web app that everyone
in my project would like to see 'enhanced' and up-to date.
One of the main reasons is to make the maintenance more
manageable. However the plethora of possible solutions
makes it difficult to choose.
Is it in your opinion possible to distill this information by
reading the cases in your book and compare each one of them
with a real life scenario?
Thanks in advance,
Gian Franco Casula
[ February 10, 2004: Message edited by: Gian Franco Casula ]

[Neal Ford]

Gian -

I read the description of your book, and was wondering
whether it gives an indication of how one can migrate
from an implementation choice to another (more sophisticated)
one.

If you design the architecture correctly up front, it is fairly easy to move from a simple framework (ala Struts) to a more complex one (like Tapestry). The comparisons I make between the frameworks assume the same basic elements that are critical to the application (namely, boundaries, entities, and business rules). The applications written with the frameworks use the same elements in almost every case. I talk a lot about proper design and architecture, and about the separation of responsibilities. I also show how to migrate a Model 2 application up to EJB's with very few changes (in fact, with changes only to boundary classes).

Is it in your opinion possible to distill this information by
reading the cases in your book and compare each one of them
with a real life scenario?

I certainly hope so -- if not, I consider the book to be a failure. The applications I write in the book are meant to represent real applications, not toys. I don't dodge issues (such as database connectivity, client and server-side validations) because it's messy.