I for one am more than interested in honest, informed, opinion such as
yours, David.  Thanks!  To be fair, Struts presently is in a bit of a
transition with things a bit in the air.  On the one hand, the
origional architect is hired on another competing project and is
trying to pull a heist on the name (keep smiling, Craig ///;-)),
evenyone has gone a bit over-ga-ga about using Chain/Command to solve
a problem self-created by using the Template Method pattern in writing
a composable RequestProcessor, etc.  But, I think this is a really
creative time around here and that much good is bound to come of it. 
Ted will tell you that only code matters.  But, I think some things
other than just code matter to some of the people involved.  Glad to
hear your voice.

Jack


On Sat, 19 Mar 2005 02:01:28 +0000, David Kennedy
<[EMAIL PROTECTED]> wrote:
> Sorry, this is going to be a fairly off-topic reaction post, but
> Lawrie's comments struck a chord with me.
> 
> Lawrie Gallardo wrote:
> > I'm relatively new to Struts and I have to say that I've found it to
> > have a realtively steep learning curve. And the only reason for this is
> > that there are so many different ways to do things and no definitive
> > best ways of achieving anything.
> 
> I totally agree. I have just transitioned to the web tier from a career
> slaving away at the back-end of large Java server systems for telecomms
> and finance. I was looking forward to using Struts quite a bit,
> suspecting, perhaps wrongly, that the nature of web tier work would mean
> a fast evolution of high-quality tools. Maybe I've been spoilt by a life
> dominated by strict (read: paranoid) coding practices, Ant and JUnit,
> but Struts is rather a confusing mess once past the initial "Oh, cool, a
> working action and page!" phase. And this is coming from someone who has
> just fled EJB land...
> 
> For example, I've got a couple of Really Common Requirements - logon and
> a side nav bar / tree view. I've been amazed that I've not found a
> simple, standard example for either of these. Instead, I've tripped over
> a wealth of little fiddly details that derailed my progress. Nothing
> like lots of soft XML config to really trip you up with details I
> find... "I've never even heard of that option, and when I tried to use
> it I wasted 5 hours not noticing a typo in one of the repeated strings!"
> 
> > I'd much rather there was a bit
> > of focus to the framework than the mass of competing options. "Make the
> > simple things simple and the hard things possible" and all that...
> 
> Totally agree. For example, I did some work with Struts Layout:Treeview
> for my nav bar prototype. I very much appreciate volunteer effort, and
> don't wish to knock anyone's efforts, but I have to say that for a
> standard solution advocated by a couple of books it's fairly poor...
> There were a couple of immediate show-stoppers with working with the
> Treeview which was very disappointing. (Turns out the license precludes
> commercial use anyway.) The point of this isn't to knock Layout (again,
> I appreciate effort and feel guilty that I don't pay back into the
> commnunity that makes me fine tools), but rather to illustrate how
> something fairly standard - a treeview nav bar with a load of
> actions/forwards - is surprisingly difficult for a beginner with the
> toolkit to knock up cleanly.
> 
> I guess some of this cognitive whiplash comes from the fact that several
> of the core components are Very Cool. The core idea of the actions is
> just Very Sensible, the basic idea of using ResourceBundle keys
> everywhere Just Works, etc. I particularly like Validator, although I
> can see, as this thread is discussing, room for disagreement about
> alternative implementations. Custom rules are very nice, and very easily
> added. I'm just troubled by the fact that most of the elements I find
> straightforward and cleanly finished are those which my boss just
> doesn't see! The basic project elements he wants in a couple of days -
> that tree view in a readable/maintainable form, a nice simple PAM login
> & timeout mechanism* - turn into real timesinks. (Actually, to be fair,
> the move from 1.0 -> 1.1 and similar shifts in several libaries also led
> to confusion when working from Googled, and often conflicting, advice.)
> 
> This is a rather rude rant I know, and I appreciate people are
> scratching their own itches, but it would be nice to seem some concerted
> effort to solve some of the FAQs cleanly, and to generate more of a core
> catalogue of Patterns for noobs like myself. (Kudos to people like Ted
> Husted who do maintain some useful resources.) In that vein, I'm very
> much looking forwards to the new Struts Cookbook that O'Reilly have just
> put out, was rather hoping to see it in the post this morning. Might
> well answer lots of questions for me.
> 
> I'm not sure I should hit 'Send' on this for fear of offending people,
> please just see this as constructive initial impression feedback. I see
> a lot of promise in the Struts toolkit, better be, sort of bet the
> project on it now!, and would love to see more consensus emerging about
> Best Practices.
> --
> David Kennedy
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 


-- 
"You can lead a horse to water but you cannot make it float on its back."
~Dakota Jack~

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

Reply via email to