So I have been using repeaters in a couple of prototypes in Axure. Repeaters allow creating a single element that then is linked to a data table and will repeat content based on how many rows are in the data table. This can make it much easier to automatically simulate a list/grid of search results with real content. You can also then add controls to filter and sort and paginate the list. There is also the ability to dynamically load images from an external source. All of this awesome functionality was making it fun to build out some prototypes that I had intended to do some remote/unmoderated usability testing.
Unfortunately, complex repeaters (6+ columns, external images, links/refernce pages, etc) with more than around 12 rows can substantially slow down the loading of a prototype in a browser. It gets even worse from the axshare server. Also, having more than 1 repeater widget on a page can also slow down a prototype to make it basically unusable. So I fell back to using dynamic panels with states to represent more limited simulation of filtering.
There are some ways to optimize repeaters including using the pagination functionality to load a smaller set of content. However, even after this optimization I still found my prototype pages with repeaters taking upwards of 8-10 seconds to load locally. Viewing pages from the axshare server which can be slow on its own, was often pushing 20+ seconds to load a page. For all the power they offer it was a big dissappointment to not be able to use repeaters in the prototypes for the usability testing. I am still looking further in to how to optimize things. I think it would be helpful if we could make a repeater load asynchronously and perhaps have a place holder image or group of elements.
AxureEx can be found here:
And the downloadable file is here:
Credit to Sam Hepworth
Spending some of my time off exploring microdata and html5 more in depth. If the semantic web is going to finally gain wide adoption microdata is going to be the catalyst. Unfortunately as easy as it is, it still involves a some effort to add microdata to your code. Still it is a million times easier than XML or other semantic formats.
Unfortunately, I think performance will still be an issue in actual implementations. Microdata is still fairly verbose and also referencing the scope/schema from a third party server may be a problem for some high traffic servers. Adding microdata to your mark-up can easily double or triple the size and there is always the latency issue of referencing a third party server.
These were the sources I was reviewing;
More and more web designers are seeking to dispel the notion of the browser fold. Yet there is still the reality that users can not see all of a page’s content if it is taller than a browser window and they will potentially leave the page before discovering the additional content. There is often an email sent from the CEO or other senior manager stating that they want some graphic or link moved above the fold. This typically ruins the layout or requires major reworking to get things to fit.
I guess I should back up a bit and explain that the fold is an invisible line where the content of the page is not visible below the bottom of your browser window (and/or to the right side if there is also more content than can fit into one screen horizontally). It is borrowed from the print world where certain print formats such as newspapers and brochures are folded. The content that was most important appears on the front page before the fold to be sure that the consumer would notice it.
Stretchy layouts are not good, yet every Ã¼ber techie who has a 20 inch or larger monitor wants web sites to stretch when they open their browser to the width of the monitor. Well maybe not every Ã¼ber techie wants stretchy layouts, but enough that comment on design related message boards that it really irks me. Why is it bad, because paragraphs that span exceedingly long widths are difficult to read. This is of course assuming they are long enough to still take several lines. A single line of text is not hard to read. And, of course, if the site is only images or video, then it is not as much an issue if at all.
I finally completed the last of the core courses for the HCI program. Now I only need to take two more electives and do the capstone. This last class was definitely one of the better courses I have taken. While it was similar to the HCI 460 course it is different in that it focused more on the the initial user observations whereas 460 was about usability testing methods and reporting. The class started out with us doing our own user observation. This involved finding a task to study a user performing. Then finding a user and arranging a time to observe them.
Working on a recent layout, I was finally confronted with the need to have an IE CSS hack. I hate hacks, because they often break with browser updates and aren’t easily remembered. So I avoid them as much as possible. However, I finally succumbed to having to use a hack… except the method I use is not a hack so much as relying on IE functionality. My designs typically can be made to work in Opera, Mozilla/Firefox and Safari browsers very easily, and then occasionally take some head pounding to get to work for IE (first windows then mac). While some might suggest designing for IE first, that is not desirable as I use a mac to develop on, although I do use a PC at work. Also, when resolving a design for IE first, it is tempting to not check the others, as IE is still the browser of the masses. Working with Safari, Mozilla first gets the less numerous browsers out of the way. The additional nightmare of IE is that you have to deal with the subversions, 5.0 does not work the same as 5.5, which is not the same as 6.x (don’t mention 4.x browsers).