Google integrating MVP tools in GWT 2.1 M1...
27 views
Skip to first unread message
Mike
May 20, 2010, 11:00:36 PM5/20/10
to GWTP
Has anyone see this?
http://code.google.com/webtoolkit/doc/latest/ReleaseNotes.html#MVP_Framework
Curious to hear any thoughts.
Cheers
Mike
http://code.google.com/webtoolkit/doc/latest/ReleaseNotes.html#MVP_Framework
Curious to hear any thoughts.
Cheers
Mike
Philippe Beaudoin
May 21, 2010, 12:16:31 AM5/21/10
to gwt-pl...@googlegroups.com
I've seen it yes. You bet I'm curious... But it's very poorly
documented for the moment. (Activities? ActivityManager?
RequestFactories? Hmmm...) It also look like it will use a lot of
automated code generation, and I'm not a huge fan of adding layers in
my compilation process. I think Christian has experimented with Spring
Roo a little, maybe he could tell us about it? Anyway, we'll wait and
see.
That being said, support for GWTP is not compromised in any way, since
it's being actively used in a number of projects, so don't hesitate to
us it in your project. For all I know, it's still the best MVP
framework out there. ;) We will also make everything we can so that
any cool new feature works with GWTP (like this server-side speed
tracer, yummy!).
Cheers,
Philippe
documented for the moment. (Activities? ActivityManager?
RequestFactories? Hmmm...) It also look like it will use a lot of
automated code generation, and I'm not a huge fan of adding layers in
my compilation process. I think Christian has experimented with Spring
Roo a little, maybe he could tell us about it? Anyway, we'll wait and
see.
That being said, support for GWTP is not compromised in any way, since
it's being actively used in a number of projects, so don't hesitate to
us it in your project. For all I know, it's still the best MVP
framework out there. ;) We will also make everything we can so that
any cool new feature works with GWTP (like this server-side speed
tracer, yummy!).
Cheers,
Philippe
Gabriel
May 22, 2010, 9:47:49 AM5/22/10
to GWTP
I built a simple application with the latest milestones of Roo and
GWT. It's extremely fast to build an initial scaffold (CRUD for all
entities), but I'm not sure how easy it is to customize it for real
world usage. When skimming the generated sources I saw A LOT of
artifacts, which I don't feel comfortable with because it means that
although "officially" my code is not coupled with Roo, if I were to
drop it I would have to manage all these generated artifacts myself.
As I understand from the generated code, It looks like the client-side
model is mapped to the server-side persistent model, which is good on
one hand because using the same model on both sides is a bad practice
for GWT-RPC, but on the other hand such "mapping" seems less safe in
the scope of type-safety. Also it looked like client-side operations
are mapped to server-side queries using strings, which is not type-
safe at all.
We should wait for some documentation before hypothesizing, but from
what it seems to me the GWT 2.1 MVP is quite different than GWTP (and
from GWT-MVP), and is not targeted to the same audience. I would say
it is targeted to developers who want to develop Spring-GWT
applications in a Grails/RoR way.
Personally I think I will use GWTP for the overall application
framework ("pages", tabs, code-splitting) and the GWT 2.1 data widgets
for the parts that make sense (search, sort, filter, browse).
My 2 cents...
Gabriel
On May 21, 7:16 am, Philippe Beaudoin <philippe.beaud...@gmail.com>
wrote:
GWT. It's extremely fast to build an initial scaffold (CRUD for all
entities), but I'm not sure how easy it is to customize it for real
world usage. When skimming the generated sources I saw A LOT of
artifacts, which I don't feel comfortable with because it means that
although "officially" my code is not coupled with Roo, if I were to
drop it I would have to manage all these generated artifacts myself.
As I understand from the generated code, It looks like the client-side
model is mapped to the server-side persistent model, which is good on
one hand because using the same model on both sides is a bad practice
for GWT-RPC, but on the other hand such "mapping" seems less safe in
the scope of type-safety. Also it looked like client-side operations
are mapped to server-side queries using strings, which is not type-
safe at all.
We should wait for some documentation before hypothesizing, but from
what it seems to me the GWT 2.1 MVP is quite different than GWTP (and
from GWT-MVP), and is not targeted to the same audience. I would say
it is targeted to developers who want to develop Spring-GWT
applications in a Grails/RoR way.
Personally I think I will use GWTP for the overall application
framework ("pages", tabs, code-splitting) and the GWT 2.1 data widgets
for the parts that make sense (search, sort, filter, browse).
My 2 cents...
Gabriel
On May 21, 7:16 am, Philippe Beaudoin <philippe.beaud...@gmail.com>
wrote:
> I've seen it yes. You bet I'm curious... But it's very poorly
> documented for the moment. (Activities? ActivityManager?
> RequestFactories? Hmmm...) It also look like it will use a lot of
> automated code generation, and I'm not a huge fan of adding layers in
> my compilation process. I think Christian has experimented with Spring
> Roo a little, maybe he could tell us about it? Anyway, we'll wait and
> see.
>
> That being said, support for GWTP is not compromised in any way, since
> it's being actively used in a number of projects, so don't hesitate to
> us it in your project. For all I know, it's still the best MVP
> framework out there. ;) We will also make everything we can so that
> any cool new feature works with GWTP (like this server-side speed
> tracer, yummy!).
>
> Cheers,
>
> Philippe
>
> >http://code.google.com/webtoolkit/doc/latest/ReleaseNotes.html#MVP_Fr...
> documented for the moment. (Activities? ActivityManager?
> RequestFactories? Hmmm...) It also look like it will use a lot of
> automated code generation, and I'm not a huge fan of adding layers in
> my compilation process. I think Christian has experimented with Spring
> Roo a little, maybe he could tell us about it? Anyway, we'll wait and
> see.
>
> That being said, support for GWTP is not compromised in any way, since
> it's being actively used in a number of projects, so don't hesitate to
> us it in your project. For all I know, it's still the best MVP
> framework out there. ;) We will also make everything we can so that
> any cool new feature works with GWTP (like this server-side speed
> tracer, yummy!).
>
> Cheers,
>
> Philippe
>
PhilBeaudoin
May 22, 2010, 11:52:55 AM5/22/10
to GWTP
Thanks, Gabriel, for this initial look at a SpringRoo-based
development process. I think the idea that "you can stop using Roo at
anytime!" is a bit of an over-sell. It's a bit like saying you can
fire that lead architect who have designed the entire system but never
told anyone how he did it.
That being said, I think SpringRoo is a very interesting idea, as long
as you don't see it as a silver bullet. Don't give Roo a blank check,
make sure you understand and like the architecture it is generating.
If it's using too many artifacts "just because it can" (a risk with
any code-generation tool) then I think you should exercise caution.
From a quick glance, it looks like GWT 2.1 MVP classes have been
designed with such an automatic-generation tool in mind. If you don't
have such a tool, I believe it will lead to code that is harder to
understand/maintain than GWTP-based code.
However, if possible, I would like to integrate GWTP with Roo as some
kind of extension or target-architecture. So you get both the benefit
of Roo's rapid scaffolding and the readability and small code
footprint of GWTP. I've been biefly exchanging tweets with SpringRoo's
Ben Alex on that topic (see @benalexau and @PhilBeaudoin) and wish I
can get him to join this conversation at some point, so that we can
evaluate the feasability of such a project. If you are interested in
contributing to this, let me know.
Cheers,
Philippe
From a (very brief) tweet exchange with Spring Roo's Ben Alex, it
looks these GWT 2.1 MVP classes are meant to be used by Roo's
automatic
development process. I think the idea that "you can stop using Roo at
anytime!" is a bit of an over-sell. It's a bit like saying you can
fire that lead architect who have designed the entire system but never
told anyone how he did it.
That being said, I think SpringRoo is a very interesting idea, as long
as you don't see it as a silver bullet. Don't give Roo a blank check,
make sure you understand and like the architecture it is generating.
If it's using too many artifacts "just because it can" (a risk with
any code-generation tool) then I think you should exercise caution.
From a quick glance, it looks like GWT 2.1 MVP classes have been
designed with such an automatic-generation tool in mind. If you don't
have such a tool, I believe it will lead to code that is harder to
understand/maintain than GWTP-based code.
However, if possible, I would like to integrate GWTP with Roo as some
kind of extension or target-architecture. So you get both the benefit
of Roo's rapid scaffolding and the readability and small code
footprint of GWTP. I've been biefly exchanging tweets with SpringRoo's
Ben Alex on that topic (see @benalexau and @PhilBeaudoin) and wish I
can get him to join this conversation at some point, so that we can
evaluate the feasability of such a project. If you are interested in
contributing to this, let me know.
Cheers,
Philippe
From a (very brief) tweet exchange with Spring Roo's Ben Alex, it
looks these GWT 2.1 MVP classes are meant to be used by Roo's
automatic
Reply all
Reply to author
Forward
0 new messages