Re: [Issue-67] [Action-385] Work on regex for validating regex subset proposal

On 8.4.2013 17:01, Felix Sasaki wrote:
> Thanks a lot, Pablo. I think your regex does allow things that would be
> forbidden with the ABNF I had proposed (e.g. starting without "["), so
> maybe it is better not to have the ABNF. Otherwise users might be
> confused. How about closing issue-67 by putting your regex into the
> schema and change the allowed characters like this:

I'm strongly against putting this regex into schema. Even with the
effort which has been put into this expression we can't be sure whether
it's correct, whether it accepts all valid inputs and refuses all
invalid inputs.

If we are going to have own syntax for RE we should properly and
formally define it. One common way of doing this is to create grammar
for our RE subset expressed in flavour of BNF. Such grammar should be
part of the spec.

Once we have grammar we can decide whether it can even be rewritten into
regular expression. Anyway such rewrite is then mechanical and will
produce similarly unreadable expression as we have now. But we will have
grammar which is easy to digest.

> - drop reference to XML Schema regex, as suggested in the original
> comment from Yves?

We should reference some specification of regex otherwise we will have
to define semantics of our regex subset.


  Jirka Kosek      e-mail:
       Professional XML consulting and training services
  DocBook customization, custom XSLT/XSL-FO document processing
 OASIS DocBook TC member, W3C Invited Expert, ISO JTC1/SC34 rep.
    Bringing you XML Prague conference

Received on Monday, 8 April 2013 16:05:48 UTC