From: Simon Strandgaard <neoneye@...>
Date: 2004-09-29T06:00:28+09:00
Subject: Re: Wish list for 2.0

On Tuesday 28 September 2004 11:26, Gavin Sinclair wrote:
> On Tuesday, September 28, 2004, 5:45:08 PM, Yukihiro wrote:
> > |Perhaps String#delete could take a regex as well?
> > |
> > |  str.delete /\s/
> >
> > Interesting idea.  What if pattern match overwrap, e.g.
> >
> >   str = "aabab"
> >   str.delete(/a./)
> >
> > ?  Should it work as str.gsub(pat, "") ?
>
> That makes sense.  Might as well implement it like that so it can be
> documented as such.


Maybe model the behavior after #scan, so that it deletes the matches that scan 
would have found.  This way we can get very flexible deletion in case 
captures are being used.

I propose #delete to erase those fragments that #scan matches.

irb(main):011:0> 'a12bc34de56fg78hi'.scan(/(\d).{2,}?(\d)/)
=> [["1", "3"], ["4", "5"], ["6", "7"]]
irb(main):012:0> 'a12bc34de56fg78hi'.scan(/\d.{2,}?\d/)
=> ["12bc3", "4de5", "6fg7"]
irb(main):013:0>


Lets change above 2 scan statements into delete

irb(main):011:0> 'a12bc34de56fg78hi'.delete(/(\d).{2,}?(\d)/)
=> "a2bcdefg8hi"
irb(main):012:0> 'a12bc34de56fg78hi'.delete(/\d.{2,}?\d/)
=> "a8hi"
irb(main):013:0>


Ok.. this is not the most obvious example.. but I hope you get the point.
Otherwise please ask :-)

--
Simon Strandgaard