RE: [xml-dev] [OT] bugs in JDK regex engine ?

["Michael Kay" <mike@saxonica.com>]

> I had been trying to enhance it since last few days, mainly 
> using the JDK 1.6.0 regex facilities.
> 
> But few of the things documented in the JDK 1.6.0 regex 
> documentation don't work as defined.
> 
> Has anybody who has done serious regex programming with JDK 
> 1.6.0 encountered some bugs?

Saxon translates XML Schema and XPath regexes into JDK regexes, so it's
pretty heavily dependent on the underlying regex engine. There are some
cases where the behaviour is very incompletely specified, for example the
effect of the "i" (case-blind) flag, but I've found very few cases where the
expected behaviour is clear and the actual behaviour differs. In my
experience, it's much more likely to be a user error.

However, I think it might be stretching the (highly elastic) patience of
this list to hold a discussion of JDK regex behaviour here.

In any case, I think the whole concept of checking XML well-formedness using
regular expressions is misguided, for the simple reason that (on theoretical
grounds) regular expressions aren't up to the job.

Michael Kay
http://www.saxonica.com/