PDA

View Full Version : Migrating away from GWT-Ext



costa_basil
20 Nov 2009, 10:42 AM
Hello:

I am at a crossroad with my project. So far, I have been using GWT-Ext with ExtJs 2.x. Now, I want to move forward with the project and use the latest goodies from the extjs libraries.

There are two ways we can move forward.

Scenario 1:
- Keep using GWT-Ext and make all the changes necessary in the library to get it working with Extjs 3.0. I end up owning and maintaining GWT-Ext, and when I say that I don't mean take over GWT-Ext rather maintain it internally as part of my project and implement only the features that I need in my project (for the sake of discussion let's ignore the licensing issues that the GWT-Ext folks imposed)

- A scenario 1 variation would be to implement my own wrapper library but I do not want to talk about it now. It is a remote posibility.

Scenario 2:
- Migrate to GXT


Some of the reasons that pushed me away from using GXT were:
- The features implemented in the library lagged behind the features in ExtJs but that doesn't seem to be the case anymore.

- GXT is slower than ExtJs. Unfortunately this is still the case with the latest version. I am using IE 7, which we have to support, and if you check out the sample application, http://www.extjs.com/examples/explorer.html, when you switch between tabs the tab change is not instantaneous. It takes a good second until the tab switches and on each tab there are only a few grids. This scares me greatly because my application has GWT modules with more than 100 panels - tabs, trees, forms, accordions etc.

To be fair, FF 3.5 is much faster than IE 7. FF 3.5 works as it should be and I would be happy with its performance but unfortunately using FF at work is out of the question.

- Another thing that scares me are the quirks of the library and which you can find only by working with it. It would be interesting to find, on a scale from 1 to 10, how would you rate the library when it comes to things working as you thought they would work as opposed to having to tweak things to get them where you want. Is there a lot of tweaking that would have to be done?

I guess will have to spend a few days and play with it and see how far I can push it.

However, I would like to hear your thoughts on this, especially the developers that already worked with the library ...

thank you!

sven
20 Nov 2009, 10:51 AM
On one tab of that examlpe you linked is far more than one tab. You cannot compare speed in the way you just did. Also there are a couple of linked actions when switching the tabs. It is not simple a tabswitch.

costa_basil
20 Nov 2009, 11:32 AM
>You cannot compare speed in the way you just did.

Why not? What I did actually was to open all the tabs. Now, when I am switching between tabs - again, these are tabs that have been opened already - it takes in IE7 a good second or two until the app displays the content under the other tab. In FF 3.5 the change is instantaneous.

>Also there are a couple of linked actions when switching the tabs. It is not simple a tabswitch.

Sorry, I don't know what you mean by linked actions. Are you saying that when I switch the tabs the app is doing more stuff behind the scenes, even if the tabs have already been opened? I really need to get my hands on this and implement a sample app myself.


I am just curious, since you are a GXT developer, did you create an app that pushes the limits of GXT in terms of how many panels/grids/trees/forms you can have before the browser explodes? Do you have some rough numbers? How about the memory usage?

Any other thoughts about the migration? What would you recommend?

Also, I noticed there is an extjs designer product on the works. Can I use that to generate GXT code?

Thanks

sven
20 Nov 2009, 11:37 AM
I would recomment a native java solution over a wrapper, so GXT.


I am just curious, since you are a GXT developer, did you create an app that pushes the limits of GXT in terms of how many panels/grids/trees/forms you can have before the browser explodes?

Yes i did. I build a couple very big applications without any problems.

For big applications there are many things you can do to make them fast, also if a lot of stuff is involved. The explorer does not contain this and probably should never do because it would make it ununderstabable complex.

dardison
20 Nov 2009, 1:26 PM
I will recommend you test the App using IE 8 you will find a performance similar to FF 3.5 there, but please don't do it with Vista, there the problem will be the S.O. try it with XP or W7.
We are building a huge Application, and so far we are doing ok with GXT.

Regards.
Daniel

costa_basil
23 Nov 2009, 10:48 AM
We are building a huge Application, and so far we are doing ok with GXT.

Thank you, dardison. You don't sound convincing though :) You didn't use the words great or excellent. Did you have to build a lot of scaffolding?

Any other opinions, thoughts please, please, please :s

dardison
23 Nov 2009, 11:22 AM
Sorry if you're not convinced. But that was not my intention after all.;)

I only try to answer your question honestly. For us GXT is not so important, the important thing is using GWT, so we don't need to deal with all the JS different flavors for different browsers, luck of OO, and I can continue naming disadvantages.
But if you came from GWT-Ext I asume you already know. I can assure you that in our experience the performance problems you mentioned are because of the version of the Software you are using.
Regarding GWT-Ext vs GXT we decided not to use the first in early 2008, When GXT was known as yet MyGWT, and had 30% of the features that it has today, because we think GWT-Ext tends to disappear, and GXT will grow. And time has proved we were right.

Regards.
Daniel

costa_basil
23 Nov 2009, 11:45 AM
For us GXT is not so important, the important thing is using GWT

These are my thoughts exactly.


Regarding GWT-Ext vs GXT we decided not to use the first in early 2008, When GXT was known as yet MyGWT, and had 30% of the features that it has today, because we think GWT-Ext tends to disappear, and GXT will grow. And time has proved we were right.


Yes, you were right, and GWT-Ext will disappear while GXT will grow. Unfortunately I had to stay with GWT-Ext due to the richer features that it provided from ExtJs. But now it is time to jump ship. And in order to adopt GXT, I wanted to see that its performance is comparable with the JavaScript version of ExtJs, based on the experience of the adopters such as yourself.

Due to the tight deadlines I couldn't shell the time to try it myself hands on, and unfortunately the demo nurtured my worries about the library performance.

Thanks again