I completely agree with about 90% of what Todd writes. This is
definitely not a flash in the pan, and the idea of using an
intermediate language (Java, in this case) that compiles to client-
side code is a brilliant and revolutionary one.
Finally, there was nothing wrong with the original MVCs. Swing (or
any other traditional MVC) worked.
Actually, I think Swing kind of sucks, and looked good when it came
out only because MFC, X, and AWT were so much worse. Swing ain't no
Cocoa. And honestly, I still kind of miss Metrowerks Powerplant.
But my real concern about GWT is that it appears to bring us back to
the world where everything is just a mess of one-size-fits-all
widgets. Konstantin is right, of course -- there is no web text
editor than can compare to a dedicated text editor rich GUI. The
reason for that, however, is because people took a *lot* of time to
work out all the minutiae of making a good UI for editing text.
By contrast, most desktop apps stick their domain into existing
widgets (one of which is a text editor) instead of going to the
enormous trouble of build a new, highly specialized UI with custom
graphics.
DHTML+CSS is quite expressive, but much lower cost, than build a
custom desktop UI component pixel by pixel. Right now, GWT seems to
lead away from some of that flexibility, and put us back in the world
of predefined widgets.
Note that this concern does *not* depend on GWT's fundamental
architecture, which is quite promising. Rather, it's a complaint
about GWT's emphasis on widgets and widgety UIs. One need only look
at Google Maps to see that GWT does not imply ultra-modal widget
overload hell.... But will GWT really lead us to fine apps like that?
Or will it lead us to apps that look like the config dialogs for Word
(bleah)?
Regardless, it's exciting to live in a world where all these great
technologies are pushing and learning from one another. Compare that
to the stagnant software world of ten years ago!
Cheers,
Paul
On May 20, 2006, at 4:12 PM, Todd Orr wrote:
This isn't really a Tapestry vs GWT thing. This is the latest
(greatest?) push to remove the application-web disconnect. If this
means that other frameworks are rendered less effective by comparison,
then so be it. This is evolution at work.
Some posts seem to indicate that this is just some flash in the pan
technology, but there is far more at work here. The development time
may be able to be accelerated to very a large degree thanks to the
traditional java based GUI paradigm being exploited here. This
technology also has the backing of google. At the end of the day, this
is more than just an ajaxy flash in the pan. Look around you. Apps
utilizing this technology are on a very sharp incline. Not because
they are flashy, or at least not just for that reason. These ajaxified
components allow developers to make better use of available bandwidth
at the same time as building more responsive GUIs. Yes, tacos (and
others) have been enabling this, but the leap here is in the learning
curve, time to market, and testability. These are where GWT seems to
be able to shine.
Whether you like the ajax stuff or you prefer the old webapp view is
immaterial. It is happening. It will likely shape the "web 2.0" world.
How you make use of these components is up to you, but there hasn't
been anything like this available in such a clean package with such a
major player backing it ever before. If you do not want to leverage
these types of (maybe rehashed) technologies, that's fine. There are a
lot of apps out there that do and there're not all just desktop app
imitators. Check out http://techcrunch.com. There are many, many very
interesting projects that are more than just desktop app wannabes.
Most of these wouldn't be what they are without the aid of ajax and
related technologies.
GWT is compelling and doesn't sit well with devs that have finally
mastered framework X. Sure, it is encouraging a change in design
paradigms. That's the best part. I see the same convo popping up on
many forums. Will there be competitors? Maybe, yes, who cares. IMHO,
one of jee's shortcomings is the lack of focus, but that's another
debate altogether. This is here. It's only in beta and it rocks
already. It hits at an ideal time when development focus is on writing
more efficient and more responsive, and more flashy apps. Few other
frameworks are addressing this. As good as Tacos is, it's clunky by
comparison.
The "code in java" ideal is the next logical step. I remember how hard
it was for my coworkers to deal with the abstractions that Tapestry
offered over dealing with the servlet api directly. Eventually, these
same people came to appreciate this. The technique that GWT employs is
the same level of shift. We're not only going to isolate you from the
servlet, we're going to isolate you from the web. This is a logical
evolution. The Web is just another view technology. I should be able
to work with it in the same manner as swing.
Finally, there was nothing wrong with the original MVCs. Swing (or any
other traditional MVC) worked. The reason that web frameworks popped
up wasn't because GUI MVCs were not good. They were formed because GUI
MVCs were impossible (or nearly) to implement on the Web. We've
reached a point where this is no longer true.
On 5/20/06, Peter Svensson <[EMAIL PROTECTED]> wrote:
Also, the viability of GWT is maybe related not to whether it
adheres to or
refers to any specific framework but whether it kills development
time and
can be integrated.
Cheers,
PS
On 5/20/06, Alan Chaney <[EMAIL PROTECTED]> wrote:
>
> I agree with both Konstantin and Paul Contrell on this one. My
> experience as a developer has been with both desktop applications
> (especially in the consumer space) and web applications
(especially in
> PHP). I have been going through the learning curve of Tapestry
because
> it offers scope to build large scale, powerful web applications
which
> can do more than just entering forms or clicking checkboxes.
>
> I am impressed with GWT after downloading and playing with it,
but I
> think that Tapestry has a whole lot more to offer than just the
ability
> to do Java-to- Javascript UI.
>
> AJAX is an overworked buzzword. The key point is that you can
design
> pages which don't need a whole page refresh to update some of the
> displayed data. The 'cool' UI stuff can be useful, but is not
essential
> in the design of an engaging and powerful application (web or
desktop.)
> Interestingly UI designers of desktop apps are tending to
'webify' them
> to give them the semantics of a web page, so it seems ironic
that web UI
> designers are so keen to go the other way!
>
> In summary, GWT is interesting and will be useful. Tapestry is
useful
> for more than just its UI components especially when combined with
> Hivemind. I don't see GWT either being the death knell for
Tapestry OR
> Tacos. I look forward to examples of integrating Tapestry and
GWT and if
> I find that I need to do one myself I'll report on my results to
this
> list, as I hope others will too.
>
>
> Alan Chaney
>
>
> Konstantin Ignatyev wrote:
>
> > >I prefer the average webapp's UI to the average
> >
> >
> >
> >>desktop (well, Windows) or Swing UI.
> >>
> >>
> > First of all let me state the obvious: there are different
types of
> applications and they have different requirements.
> >
> >
> >I yet to see a convenient web based text editor or accounting
> application, much less an IDE. Please point me at just one
robust and
> convenient text editing component: they are not a match to
'desktop' .ones.
> >
> >
> > The whole idea of overhauling html with javascript and other
types of
> augmentation technologies does not seem to be conceptually
correct. From the
> conceptual point of view it all looks like XWindow reinvention
with the
> help of ducktape and gluegun.
> >
> >
> > IMO the whole buzz around GWT validates Swing framework as
solid basis
> for building certain types of UI. Ant therefore rather than
compile it to
> Javascript or whatever within a browser it would make much more
sense to
> let Swing components to work within browsers.
> >
> >
> > And what is interesting is that all the technologies are here
in place
> ant they need just slight adjustments:
> > - Browser Components were invented long time ago and they are
called:
> Applets. All we need is to make Java Web Start technology to
work with them
> well;
> > - JavaWebStart, JNLP actually needs to be altered a bit to
allow using a
> shared repository of components per developers choice. The
ability is
> present now but the feature is artificially limited to the same
source
> domain and does not allow multiple signatures on components;
> > - And Java RT should be made modular and become a must have
for the
> clients, which is going to be easy enough since Sun is going to
opensource
> it;
> >
> > I think that this set of technologies if far superior to
anything else
> we have in the space: Flash, Ajax, and current JWS applications.
Yes the
> technologies are 'old' and have some stigma attached but we need
to overcome
> it in order to have some meaningful progress rather than be
obsessed with
> 'new' stuff that on many occasions is just reinvent the wheel,
but makes it
> square or octahedral (I guess it improves traction.).
> >
> >
> >
> >
> >
>
>
---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
_________________________________________________________________
Piano music podcast: http://inthehands.com
Other interesting stuff: http://innig.net
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
