On Wed, May 11, 2011 at 06:17:07PM -0400, Andrew Dunstan wrote:
> On 05/09/2011 11:25 PM, Noah Misch wrote:
>> SELECT xmlcomment(E'\ufffe');
>
> That's a bit harder. Do we want to extend these checks to cover
> surrogates and end of plane characters, which are the remaining
> forbidden chars? It certainly seems likely to be a somewhat slower test
> since I think we'd need to process the input strings a Unicode char at a
> time, but we do that in other places and it seems acceptable. What do
> people think?
My thinking was that we should only make this flag day for xml-type users if
we're going to fix it all the way.