The TIPing point

One of the things that the new MSN Search points out is a flaw in the Tablet: A web page is unable to pass along context hints to the TIP. The result is that when you try to handwrite the new MSN search filters as shown below, you get garbage:

It’s a bit hard to see here, but the handwritten string “Tablet PC {frsh=100}” is recognized as “Tablet pc Etrsh=loo}”. The problem is that the recognizer has no idea about the new syntax in MSN search. So for us Tablet users, we have no choice but to tap through drop lists and work with the MSN sliders. Too bad. Handwriting a query would be faster.

All us Tablet users recognize this problem. As soon as the Tablet launched and we all ran to IE to try to handwrite a URL, we found it couldn’t. SP2 fixed that problem by programmatically enabling IE to tell the TIP that when in the edit field of the address bar the handwriting recognizer should be expecting a URL. The result is much improved recognition. As a programmer you can do the same for your application or even use the context tagging tool for apps you use that weren’t customized for the Tablet.

But that doesn’t help where much of the development world has been going: web pages. As the MSN search field illustrates, from a TIP handwriting standpoint, the search field is going to be limited to full words–no special commands.

As I’ve written about before, it seems to me this isn’t all that hard to solve. Of course, it’s not something any of us could do without the source to IE, but doesn’t it seem that a regular expression for each field in a web page (that tells the recognizer what strings are valid) could be passed along with the page contents? Then IE could intercept this string and pass it to the TIP when the corresponding field gains focus. I can appreciate the security implications, but aren’t they manageable? Hmm. I must be missing something. Time to download the latest Firefox code.