OK, having read both P&C and TWI next to each other I can see now how this can work - there is just a slight confusion in the two specs with regard to default text direction.
Basically, P&C's rules for text direction always return a direction (default: LTR) whereas the TWI rules - and test cases - assume the default is no direction (Null). So by using the P&C algorithm to generate the properties, I ended up failing a lot of the TWI tests as all my properties had a "dir" value. To make both sets of tests pass, I have to set the default as Null rather than LTR despite what P&C says - this is fine as the browser treats no-direction as LTR by default anyway. Not a big problem to work around, but it did trip me up and does seem inconsistent. On 4 Apr 2011, at 10:57, Scott Wilson wrote: > I'm having problems passing test i18nlro06 for the Widget Interface test > suite [1]. > > The test expects a value of: > "\u05DD\u05E4\u05DC\u202D\u05D7\u05DC\u202C\u05E7" for the Name attribute; > > Following the localizable strings algorithm[2], we actually return: > "\u202A\u05DD\u05E4\u05DC\u202D\u05D7\u05DC\u202C\u05E7\u202C" > > This may also apply to test i18nlro07. > > -S > > [1] http://dev.w3.org/2006/waf/widgets-api/test-suite/ > [2] http://dev.w3.org/2006/waf/widgets-api/#getting-localizable-strings-
