atom feed19 messages in com.googlegroups.google-web-toolkitGoogle does not use its own GWT frame...
FromSent OnAttachments
David BalmeJul 30, 2008 9:12 am 
Ian BamburyJul 30, 2008 3:53 pm 
Ian BamburyJul 30, 2008 3:58 pm 
TimJul 30, 2008 6:21 pm 
David BalmeJul 30, 2008 7:30 pm 
PingJul 31, 2008 4:49 am 
rustyJul 31, 2008 5:06 pm 
Magno MachadoAug 1, 2008 10:43 am 
waldenAug 1, 2008 12:24 pm 
Magno MachadoAug 1, 2008 12:35 pm 
Ian BamburyAug 1, 2008 12:36 pm 
Magno MachadoAug 1, 2008 12:48 pm 
Magno MachadoAug 1, 2008 12:51 pm 
Srini MarreddyAug 1, 2008 1:10 pm 
Ian BamburyAug 1, 2008 1:11 pm 
MaximAug 1, 2008 2:30 pm 
David BalmeAug 2, 2008 1:13 pm 
Arthur KalmensonAug 4, 2008 8:02 pm 
bru...@google.comAug 13, 2008 3:04 pm 
Subject:Google does not use its own GWT framework
From:David Balme (davi@gmail.com)
Date:Jul 30, 2008 9:12:45 am
List:com.googlegroups.google-web-toolkit

Hello GWT developers:

I have been working with GWT for some time now. I'd say since about Nov 2008. I am very very impressed with your work.

I have to admit I did have concerns when deciding to formally adopt this framework within my organization for a couple of reasons. I'm a dev team lead and sometime dev-manager so I have to make responsible technology decisions.

(caveat - keep in mind that I refer to GWT 1.4 as I cannot put java 1.5 code into our prod environment yet)

So here were (and still are) my concerns:

1) Blurring of the traditional java developer and Web designer roles: The GWT application that I am writing is like a full blown desktop application - but in the browser. To make it look halfway decent requires good .css knowledge and the abiltity to use the widgets correctly so that layout works out. It takes a java developer to put the widgets together but it takes a good webdesigner to make layout look good. To be effective and efficient in this with GWT, arguably, requires that one person be good at both skillsets. This can be an issue in dev shops that have java developers (doing things like struts and j2ee stuff) and then the web designer who marks up the jsp pages with their HTML and css and javascript magic. So adopting GWT requires acknowledging this blurring of roles issue and taking a hard look at staff skillsets. I always make technology decisions with this issue in mind.

2) Databinding value objects to web forms GWT (in 1.4) does not have decent (convenient) databinding capabilities. ie In struts and and other frameworks of it's 'ilk', reflection is typically used to map the http request data onto the form (valueobject). This is convenient, seemless and usually forgotten by many developers. In GWT this is not so conveniently done. GWT does not have reflection. I understand why it doesn't have reflection (its because of the static code analysis that the GWT needs and the fact that javascript has no reflection capabilities because of the lack of meta data in its objects). Of course we can use the form listeners and fire events to our hearts content (the swing way) - but that translates to a lot of boilerplate (tedious) code. I have seen some ideas from others where GWT 'generators' can be used to solve this issue. But it is complicated and not straight-forward. I think a binding standard should be created.

.. But now my real concern... and my real motivation for writing this note..

3) As far as I can tell Google does not use GWT!!!! This is a problem of GWT credibility! Issues (1) and (2) are easily overcome through programming mechanics, training and creating GWT conventions and standards. However, a true test of any tool is that the creator of said tool actually uses the tool in a production setting. The sample applications, though good, are not enough to create this tool's credibility. Why is the rest of Google ignoring GWT? For example: As far as I can tell (by viewing page source) http://docs.google.com does not use GWT. This is a perfect Google GWT application candidate and would prove that Google is truly serious about GWT. I understand the realities of large organizations and different dev teams working in departmental sylos and not doing the requisite cross pollenation of ideas across different development initiatives. That happens everywhere. But if you GWT developers could convince your Google-docs counter-parts to use GWT then credibility would be established. But much more than that would occur. The google doc guys would end up being your clients and the GWT team would be forced to enhance the libraries as the Google-Doc team demanded more and more from you. Naturally the GWT code base would improve. So why not do it? Or have you done it already and I have not noticed?

Thanks very much for creating such a great Web 2.0 dev platform. I've stuck my neck out and picked your technology. I'd like to see that you guys have done the same and pick your own technology to write some of your own production applications.

Cheers and thanks for reading..