3.0 beta 2

Grid containing page frozen for 8 seconds until scrollbar painted


I am experiencing very long unavailability times with some grids. The page page with grid an everything is loaded within 1200 milli seconds but then the page is frozen until the scrollbar of the grid is finally painted.

I was looking at the performance thread at /datagrid.897/grid-performance.html and added a similar performace measurement. The status bar of the page with all number is visible after a bit more than a second. But the page is frozen for another ca 8 seconds until as said the scroll bar of the grid is painted. From the measurement I can see that the actual writing of the grid takes less than 400 ms.

The grid has around 40 rows and 10 columns. I have special coloring for some rows, multi line headers, alternating colors of standard rows. The grid has no horizontal scrollbar. The grid height is limited to a 1024 x 768 resolution, about 50% of the rows are visible.

During the page paint I noticed a peek in CPU usage that drops immediately once the scrollbar is available.

The scrollbar is needed because one of the requirements is a fixed column header that is visible all the time.

I really would like to know what I can do to improve the speed of page interaction availability. Right now our clients refuse to accept the pages with such a long "load" time.



Andreas Guther
June 15,

Maybe I should add that I am using IE6. And that I do not see the problem with less fancy Grids. For example I have a two column 70 rows grid with a simple header and only alternating colors that is available within a second, i.e. it takes about another second to have the scrollbar available.
Andreas Guther
June 15,
Andreas,

Could you send me this page (or make it accessible over internet) so I can look where the problem is? It might be not related to scrollbars but something else which blocks script execution – scrollbar adjustment normally happens after 0.5s timeout but if there is another process blocking the script it will only happen after blocking condition is released.

However it might be related to the script rendering. I had one or two reports previously which look like on certain conditions the grid rendering takes much longer, like if it goes to some sort of internal loop but I never could reproduce this. If you can send me the page I’ll try to see what could be wrong.
Alex (ActiveWidgets)
June 15,
Alex:

While I was putting all the needed files together that are required for the page I copied file by file to a new fold. This gradually adding needed include files (a few CSS file and a few JavaScript include files) showed me that the problem occurs once a specific JavaScript (one of mine) is added.

The proble is definetely home grown and I now have to look into which routine in that file causes the big delay.

Thank you so much for your offer for looking at the code and I am glad I do not have to ask you for your valuable time now.
Andreas Guther
June 15,

I finally discovered the problem’s source:

I was using the csshover.htc file to solve an IE bug:

http://www.xs4all.nl/~peterned/csshover.html.

Unfortunately the csshover.htc slows everything down on huger pages like the one I am currently working on and especially if there are quite a few CSS files and definitions involved.

This explains also why the performance issue did only occur on IE.

In case someone comes accross similar performance issues with larger pages and lots of styles: check out if you use the csshover.htc.

Andreas Guther
June 19,

This topic is archived.

See also:


Back to support forum

Search

Version 2

Looking for AW 2.6 ?
The old site is here.