Guriddo » Topic: 3.7.1 tree grid slow to expand or collapse nodes

3.7.1 tree grid slow to expand or collapse nodes

Guriddo
3.7.1 tree grid slow to expand or collapse nodes

This topic has 14 replies, 4 voices, and was last updated 15 years, 8 months ago by waiting.

Creator

Topic
June 16, 2010 at 5:43 pm #76490

Les
Participant

I just upgraded from 3.6.5 to 3.7.1 and I see a big boo-boo.

When you click a tree node it expands or collapses much slower compared to the previous release.

I see a large difference in IE7.
Creator

Topic

Viewing 14 replies - 1 through 14 (of 14 total)

Author

Replies
June 17, 2010 at 1:22 pm #94797

tony
Keymaster

Hello,

Thanks for considering this. Also we make the tree grid in mind that this way it will be faster!?!

Since it is a very difficult to determine where the problem is, could you please send me a link to the problem,

so I can see what is happen.

Thank you

Best Regards

Tony

June 17, 2010 at 3:48 pm #94803

Les
Participant

Tony,

I see that you made some code changes in the way tree grid row expand and collapse in 3.7.1.

June 20, 2010 at 9:29 am #94830

tony
Keymaster

Hello,

I do not see such problem, or maybe my computer is not like your that you test.

Also I will try with more nodes.

Best Regards

Tony

June 20, 2010 at 6:03 pm #94884

Les
Participant

Tony,

I did some more testing and I have measurements taken on another faster computer.

I inserted console.tiime and console.timeEnd in this click handler in 3.6.5 and 3.7.1 (Firefox 3.6.3 and Win XP)

$(".treeclick",row).bind("click",function(e){

See the results on pictures (3.7.1 is 10 times slower)

jqGrid 3.6.5

http://img191.imageshack.us/img191/9291/tree365.png

jqGrid 3.7.1

http://img217.imageshack.us/img217/7056/treew.png

I'd like to work with you to resolve this issue.

June 21, 2010 at 4:08 am #94887

Les
Participant

Tony,

Performance of expandRow() and collapseRow() in 3.7.1 will improve if you remove the context, see below.

The context is not needed because the query (by ID) maps almost directly to document.getElementById.

collapseRow : function (record) {

…

//$(“#”+id,$t.grid.bDiv).css(“display”,”none”);

$(“#”+id).css(“display”,”none”);

expandRow: function (record){

…

//$(“#”+id,$t.grid.bDiv).css(“display”,””);

$(“#”+id).css(“display”,””);

June 21, 2010 at 4:48 am #94889

Les
Participant

The same change can be made in 3.7.1 in expandNode() and collapseNode() – don't provide context when the query is by id.

I made this change and now performance is the same as in 3.6.5.

June 21, 2010 at 5:00 am #94890

Les
Participant

I found post this which explains that passing a context will slow things down when query is an ID.

http://stackoverflow.com/questions/2421782/preformance-of-jquery-selector-with-context

June 25, 2010 at 10:06 am #94930

tony
Keymaster

Hello,

First of all thank you very much for the investigations and recommendations.

I'm ready to do this, but imagine that you have two grids on the same page. It is quite possible thet both have same id,

in this case the collapsing and expanding maybe will not work, becouse it will depend which id will be find first.

Will see how to improve this.

Best Regards

and thanks again

Tony

June 25, 2010 at 3:47 pm #94953

Les
Participant

Tony,

I just wanted to point out that by HTML spec element IDs must be unique in a document

http://www.w3.org/TR/html401/struct/global.html#h-7.5.2

I'm painfully aware that the jqGrid will create elements with duplicate ids (not only in rows but also in column headers) when there are two grids populated with the same data.

I hope you realize that this is a really bad design.

June 25, 2010 at 3:58 pm #94955

tony
Keymaster

Hello,

Thanks Les. Absolutley you are right, but if we fix this, nobody can continue to use the new versions – they should rewrite your code from the beginning.

For the column headers and other duplicate ids we can fix these, but the big questions is what to do with the rows ids?

Example: I have a table with primary keys 1,2,3.. and another table with the same primary keys numbering due of using the autoincremet property and I want to use these keys when I do CRUD operations.

In the same time I need to display both grids in one page – so the question is: how we can make these rows uniquie and in the same time to use the primary keys values?

Best Regards

Tony

June 25, 2010 at 4:10 pm #94956

Les
Participant

Tony,

Perhaps you will need to make some tough decisions and brake backward compatibility in order to fix the markup.

Right now we are not able to upgrade to 3.7.1, and it's not because of the duplicate ids but because of the JS patch that you added to guard against the duplicate ids.

Les

June 25, 2010 at 4:38 pm #94957

tony
Keymaster

Hello Les,

The same will hapen if we do the change.

Instead I have begin a serious research on treeGrid and we will increase the speed .

Tony

October 19, 2010 at 11:39 am #96337

Aurelien Pernoud
Participant

Hello tony,

did you by any chance have a look at optimizing treegrid (not only expandRow and collapseRow, but also loading) ?

I've tried 3.8 version with latest jQuery UI but except the expandRow and collapseRow that are indeed really slower (we used 3.6.4 before), loading is the same.

Thanks for your feedback.

Aur

October 27, 2010 at 5:24 am #96414

waiting
Participant

tony said:

For the column headers and other duplicate ids we can fix these, but the big questions is what to do with the rows ids?

Example: I have a table with primary keys 1,2,3.. and another table with the same primary keys numbering due of using the autoincremet property and I want to use these keys when I do CRUD operations.

In the same time I need to display both grids in one page – so the question is: how we can make these rows uniquie and in the same time to use the primary keys value

Hello Tony,

In my option, i should make it by this way:

use grid-container-id (eg. 'treegrid') as prefix of rows ids (eg. 'treegrid_1/treegrid_2/treegrid_3'). when CRUD then substr the prefix — this action can be done by client side before starting post, or by server side after receive the post data (must post the prefix
Author

Replies