Worried

[rich02818]

A couple of comments re: performance. First, Ext JS 4.1 fixed a lot of the layout performance issues that existed in 4.x. The layouts are drastically faster compared to 4.0. In some cases they will be slower than 3.x and in other cases faster.

Please specify and quantify these statements re: the many very specific use cases that have been posted repeatedly in these forums. We need to see direct comparisons of the latest v4 vs v3.4.1 for specific scenarios to believe that considering the upgrade path makes business sense. These *must* include performance on IE.


For example:


http://www.sencha.com/forum/showthre...ormance-vs-3.4 (BTW, I asked for an update on this one 10 months ago and got no response...)


http://www.sencha.com/forum/showthre...ues-under-IE-8


http://www.sencha.com/forum/showthre...ry-simple-page


I've asked for Sencha quantification on performance for specific tests many times in these forums and have yet to see any. Would love to be surprised this time. It makes no sense for several developers to individually upgrade the test cases already provided to the latest v4 release only to find out that the performance is still slower than v3.

I can let you know that we'll start work on it once we wrap up Ext JS 4.2.0 so we can have some of the dev team go back and work on 3.4.2. That's the best answer I can give you right now.

So nobody is currently working on v3 fixes... I appreciate the honest response.

[jimmylu98]

I am too worried about the performance of 4.x.x and not sure Sencha is focusing to solve it since release of 4.0.x. It should be very simple to improve performance if Sencha try to seriously understand the performance problem.

I have spent near 6 month of my spare time to migrate our application from 3.4.x to 4.1.x, the biggest issue is performance and I even didn't start test with IE7/8 yet. That is too much time which I could use for our application improvement.

I have created a very simple grid test case and requested a Sencha premium support ticket. The Sencha page-analyzer indicates that initial load grid will need total 3 layouts with 4.1.x/4.2.x, while 3.4.x only need total 2 layouts. What I got from Sencha support is that all 3 layouts are needed and cannot be reduce to two for the performance. Then I spent a few hours to study the ext-all-debug.js code, and found the following patch (add 2 line codes), that will reduce the layouts from 3 to 2, and it also will not increase any layouts with hide/show columns. I am not happy we paid for the support, since we have to spend time to create simple test case for the extjs problem which will benefit more to Sencha, but we could not get valuable supports. See
http://www.sencha.com/forum/showthre...mpare-to-3.4.x

However, the patch along is still not enough to match 3.4.x performance. The big problem is that I don't see the patch is even applied to latest 4.2.0 RC - Don't care?

The tree data store has same problem as grid data store, you will need similar patch to improve the tree performance.

The biggest sale point of 4.x.x is MVC. Yes, MVC is great for developer (us). However, anyone proves that MVC code performance is better than or equal to non-MVC. My simple test code cannot prove that. If MVC performance is worse, why should use it.

Ext.define('My.patch.data.Store', {     override: 'Ext.data.Store',
    onProxyLoad: function(operation) { 
        var me = this,
            resultSet = operation.getResultSet(),
            records = operation.getRecords(),
            successful = operation.wasSuccessful();

        if (resultSet) {
            me.totalCount = resultSet.total;
        }

        Ext.suspendLayouts(); // added code
        if (successful) {
            me.loadRecords(records, operation); // this will trig layout
        }
        me.loading = false;
        if (me.hasListeners.load) { 
            me.fireEvent('load', me, records, successful); // this will trig layout
        }
        if (me.hasListeners.read) {
            me.fireEvent('read', me, records, successful);
        }
        Ext.resumeLayouts(true); // added code
        
        Ext.callback(operation.callback, operation.scope || me, [records, operation, successful]);
    }
}); 

[chrisminniti]

The fact that you say the update from 3.0 to 4.0 cost you customers. Why as a business owner would you roll out an imperfect solution with missing features that you so clearly did not believe in? You should have kept your 3.0 application online until you were sure that 4.0 fit your needs. You can't make bad decisions and then blame the company that provided the solution you've made a career on for it not working out.

I can understand that you are passionate, but I personally feel your flames are shooting in the wrong direction. I'm glad that Sencha continues to push innovation no matter how hard it makes backwards compatibility. We can't support old legacy browsers forever. That would kill our industry and kill Sencha's ability to lead in the market.

[jchau]

We went from 3.x to 4.x and it was hell. It took us about 6 months to get everything semi-working and we are still finding new bugs every day. We encountered major performance issues and the reply was to just wrap everything in suspendLayouts and resumeLayouts. So now we sprinkled that crap all over our code. Support here is a joke. Forum is a joke. The tones some Sencha forum moderators use are unprofessional and just rude. Please Sencha, please go back to your roots and focus on your core products and services. From an outsider perspective, it's almost as if the entire management team was replaced by marketing and sales. "Forget about delivering quality product. Let's just create a bunch of packages and bundles and cool stuff to sell."