[osis-core] Grain regex

[Todd Tillinghast]

Patrick,

> Todd,
> 
> Todd Tillinghast wrote:
> 
> >>Todd,
> >>
> >>cp = character position
> >>
> >>str = string
> >>
> >>Lack of delimiter on cp was due to lack on one on the regex Steve
and
> >>
> >I
> >
> >>wrote. ;-) Inserted ":" on the expressions I am about to send. Not
> >>wedded to it, suggest [ and ] for consistency with str?
> >>
> >>I am trying with \L and \Nd to allow all letters and digits. (Note
> >>
> >that
> >
> >>character position can only be \Nd, if that is the question? May be
> >>missing the real question so if I am, please repeat.)
> >>
> >
> >That was the question and a clear answer!
> >
> >It would be nice if str and cp were consistent in their delimiters.
> >
> Hmm, did it differently because cp does not need a container around
> ...., well, actually that might be nice.
> 
> Votes on adding square brackets to cp? In other words, cp\[\p{Nd}\]
(in
> the crystal clear schema syntax).
> 
> >
> >Why string and not word?  Is word to Western oriented?
> >
> Not really, although word tokens being white space delimited is
Western
> oriented. More pragmatic reason that you may want to address an
> arbitrary string, which may or may not contain whole words at each
end.
> Doesn't cost any more and you get more robust addressing. Would be
more
> difficult to specify and implement white space delimited tokens only,
> and the ultimate ability is less useful. (I know, an argument for MS
to
> do it that way but I prefer the other result. ;-)
> 
> Note that unless there is a very good reason to not do the square
> brackets that lands in my inbox by tomorrow morning, I will issue 1.1
> (as a beta for buidling texts) final to appear by say 1 September(?)
by
> tomorrow at NOON.
> 
> Patrick

What is the motivation behind [ and ] for string?  Why not use : for
both cases?

Todd