That JMeter testing sounds invalid. You're simulating the repeated
download of static resources that in reality happens once ever per
user.
SmartGWT is designed for applications that are used more than once per
user, and/or that users spend at least a few minutes with. In that
scenario, features like Adaptive Filtering greatly reduce the number
of requests sent *after* page load, while the user is using the
application:
http://www.smartclient.com/smartgwt/showcase/#grid_adaptive_filter_featured_category
For the enterprise applications SmartGWT is designed for, it's the
most scalable solution available, because it reduces expensive
database search operations in exchange for a once-ever download of
static, cacheable files.
Finally, on interoperability, the original poster never said what
concrete components he had trouble with, but as others have posted, if
you need to add selected GWT widgets to your SmartGWT interface, that
works fine.
100% interoperability is a bit of a myth. There's a bunch of widgets
available for GWT with varying degrees of quality. If you grab random
third-party or incubator GWT widgets and start mixing them into
complex layouts, you're going to run into problems as well.
On Nov 14, 11:46 am, Sunit Katkar <
sunitkat...@gmail.com> wrote:
> I have found that pure GWT works best, at least for our application. Since
> our app is used by thousands of users at a given time, we wanted to keep the
> code that gets downloaded and interpreted on the browser to a minimum. Also
> to avoid having to look up different vendors for fixes, etc. we chose to use
> just Google GWT. Yes, we had to write a few CSS styles on our own, but given
> the type of application and the scalability requirements, we found that pure
> GWT works best. The money spent on a graphic artists services for creating
> good looking light weight CSS and icons has paid back already.
>
> For the admin side of our app we had the first version with SmartGWT, but
> have now migrated that to pure GWT. SmartGWT is a good toolkit but to
> address scalability and load conditions, our tests (using GWTRPCCommLayer
> and JMeter, combination et-al) found that pure GWT was definitely performing
> better. So we did away with SmartGWT.
>
> - Sunit Katkarhttp://
sunitkatkar.blogspot.com/
> > <
google-web-toolkit%2Bunsu...@googlegroups.com<
google-web-toolkit%252Buns...@googlegroups.com>