Here's an interesting question.  When writing T4, did you know that T5 was
going to be such a drastic rewrite?  Probably not, or you would have just
architected T4 the right way to avoid the rewrite.  Correct?  Well, then,
who is to say that two years down the road whenever you get down to working
on T6 that won't you decide the T5 architecture just won't work?  You
probably thought at the time of writing T4 that the architecture was the
right way to go for the future and now it's "untenable w.r.t. to adding new
features without breaking backwards compatibility."

Actually, T4 did come out better than I expected it to, which is very
nice. But the entire time I was working on it, I was irritated by the
many things I could not fix.

I've been talking with people as far back as 2003 about the need for a
pure POJO model with a transforming class loader.

The different between the T3 -> T4 transition, and the T4 -> T5
transition is that instead of hacking around the limitations of the
API (the cause of upgrade issues from one release to the next) I'm
building such a minimal, adaptive API that upgrade issues won't exist
in the future.

I want to bring back the metaphor from my blog:  In early 2000 I wound
the spring on the Tapestry code base really tight.  I found pretty
much the right abstractions and patterns.  I've made mistakes,
overcome limitations, added powerful features.  But all that time, the
spring has been winding down. With T5 I want to wind the spring
really, really tight again so that it'll last even longer.


I have (not that I necessarily think you were addressing me, but just in
case) started to help make Tapestry more popular (Hibernate, Acegi, and
AspectJ integration for starters) and I've contributed to Tapestry itself
(autowiring), but a lot of my work will have to be completely rewritten for
T5!

Also, as a consultant, I have to recommend to my clients what technologies
to use in their best interest.  If Tapestry continues down the path that
it's going, I could not endorse Tapestry as a viable technology solution for
a large, on-going project.  In other words, I wouldn't stick my neck out to
suggest Tapestry given its track record.



--
Howard M. Lewis Ship
TWD Consulting, Inc.
Independent J2EE / Open-Source Java Consultant
Creator and PMC Chair, Apache Tapestry
Creator, Apache HiveMind

Professional Tapestry training, mentoring, support
and project work.  http://howardlewisship.com

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to