Well I've just tried it with .NET and it doesn't seem to work there either. 
I will try Saxon later.
( I tried [A-Z-[D-Z]]{3} against "ABC" which failed.)

-- 

Joe Fawcett (MVP - XML)
http://joe.fawcett.name



"CP" <msxml@s...> wrote in message 
news:egYHXHmjHHA.4772@T......
> Hi Joe,
>
> I'd never heard of it before today either - problem is BS7666 is an 
> 'official' UK Gov Schema and it conforms to the w3 specs.  As a short-term 
> workaround we'll hack the xsd to use the 'normal' Reg Exp notation like 
> you suggest - mid-long term I think we'll be abandoning MSXML for this 
> product and looking at another method (.NET/Java/Xerces/whatever .. 
> *sigh*)
>
> Thanks!
>
> Carl
>
>
> "Joe Fawcett" <joefawcett@n...> wrote in message 
> news:%23nzmLBmjHHA.4596@T......
>> I'm not sure, I 'd never heard of it myself before your post, but surely 
>> there's an easy workaround in this case, just list the valid letters?
>>
>> -- 
>>
>> Joe Fawcett (MVP - XML)
>> http://joe.fawcett.name
>>
>> "CP" <msxml@s...> wrote in message 
>> news:eBVE0GljHHA.3472@T......
>>> Hi there,
>>>
>>> Been trying to validate (and failing) a UK XML postcode against the UK 
>>> government BS7666 v2.0 schema which has the following pattern:
>>>
>>> [A-Z]{1,2}[0-9R][0-9A-Z]? [0-9][A-Z-[CIKMOV]]{2}
>>>
>>> Am I correct in assuming that MSXML6 still does not support RegExp 
>>> Character Class Subtraction?  I saw this link:
>>>
>>> http://blogs.msdn.com/stan_kitsis/archive/2005/09/20/472185.aspx
>>>
>>> that specifies that it doesn't but as this was posted in 2005 I thought 
>>> I'd just see if anyone here knows if it's going to get updated.
>>>
>>> Thanks for any info
>>>
>>> Regards
>>>
>>> Carl
>>>
>>
>>
>
>