[Regexp] Re: Some queries

[Wes Biggs]

Hardeep Singh, Noida wrote:
> Hi Wes,
>
> I have some queries about the package:
>
> 1. The document says that the package is a traditional NFA. However, it
> appears to be POSIX NFA, although it might not conform to all POSIX
> standards. I believe this for the following reasons:
>         a) A Traditional NFA stops as soon as it has found ONE match. It
> does not try to look for other possible matches of              longer
> length. For example the regexp 't(e|es).' (without quotes) should match
> 'tes', not 'test' if given the          text 'test'. This is because this
> match is found first. Unless an NFA regexp continues with the motions, it
> will            not find the match 'test' which is longer and starts at the
> same position, which is the way a POSIX NFA works.
>         b) Ditto for 'a(one)?(onetest)?' when applied to 'aonetest'. Should
> match 'aone' and not 'aonetest'

Yes.. it's really not completely one or the other.  I've tried to 
implement POSIX features where possible -- sometimes at the cost of 
performance.  I have thought about making this configurable, so you 
could trade off speed vs. POSIX correctness.

> 2. There appears to be a bug, although it may be a design decision. When
> case insensitive matching is applied with pattern '(\w+)\1' to text 'Okok',
> it does not match.

True.  I suppose the thing to do would be match the user expectation, 
which is probably that it will match.  Is that how it works in Perl?

> 3. Could you provide a link to my writeup on regular expressions so that
> people interested in regexps can find it easily. The link is:
>
> http://www.seeingwithc.com/topic7html.html

Sure.  Next time I update the site I will put in the link.

Wes