From: Nikolai Weibull <ruby-talk@...>
Date: 2004-09-23T10:05:18+09:00
Subject: Re: Regexp Arity

* Florian Frank <flori@nixe.ping.de> [Sep 23, 2004 03:00]:
> > > > Regexp#groups ?
> > > > Regexp#groupcount?

> > > Regexp#ngroups

> > Best by far, so far,

> Which groups are counted here? There are capturing groups and
> non-capturing groups. Or is this the sum of those two numbers?

Precisely why it isn't great.

> At the moment I don't see any obvious reason to count those
> groups in a Regexp at all. I should know everything about them
> before I created the regexp. If I want to find out how many
> captures I got after matching the regexp I can easily get that
> info from MatchData#captures.size.

> On the other hand the grouping is a tree structure and if I simply
> compute one number from a regexp I loose a lot of information. They can
> be quite complicated like /(((a))(?:(b|c)|d))/. From the group
> size alone I cannot infer very much about the values that will be
> captured after a match. Simons tree construction example makes
> much more sense if one wants to examine the structure of
> a regexp.

The only viable thing to do is return exactly the number of capturing
groups given, not excluding anything.  If you wish to write a regex such
as yours above, then you are in it for the trouble.  I agree with you,
however, that designing based on the knowledge of how many capturing
groups are in a regex isn't very good design at all.
	nikolai


--
::: name: Nikolai Weibull    :: aliases: pcp / lone-star / aka :::
::: born: Chicago, IL USA    :: loc atm: Gothenburg, Sweden    :::
::: page: www.pcppopper.org  :: fun atm: gf,lps,ruby,lisp,war3 :::
main(){printf(&linux["\021%six\012\0"],(linux)["have"]+"fun"-97);}