From: Robert Klemme <bob.news@...>
Date: 2004-09-29T21:44:19+09:00
Subject: Re: Range of regexps?


"Alexey Verkhovsky" <alex@verk.info> schrieb im Newsbeitrag
news:1096368560.937.5.camel@localhost.localdomain...
> Have a question (as usual, from a thread on ruby-forum.org,
> http://www.ruby-forum.org/bb/viewtopic.php?t=75) :
>
> Expression /start/../end/ is invalid because Regexp doesn't have <=>
> method, but this code is printed in Pickaxe:
>
> while gets
>   print if /start/../end/
> end
>
> with comment
> QTE
> As well as representing sequences, ranges may also be used as
> conditional expressions. For example, the following code fragment prints
> sets of lines from standard input, where the first line in each set
> contains the word ``start'' and the last line the word ``end.''
> UNQTE
>
> As the author of the thread puts it,
> QTE
> What in the world is this program really supposed to do? Whatever I type
> in at standard input it prints out. /start/../end/ just seems to be a
> synonym for "true". I can't make sense of this example. Is it wrong?
> UNQTE

It once was the togglig boolean operator that used implicitely $_.  You
will notice a message similar to this if you use "ruby -w" (with
warnings):

range-test.rb:15: warning: range literal in condition

The short form is equivalent to this verbose and much clearer form:

while ( line = gets )
  print line if /start/ =~ line .. /end/ =~ line
end

The ".." in conditionals ("if", "while", "unless", "until") is special as
it stores a boolean flag that is set to true when the first condition
matches and remains true until the second condition matches.

> To which I would like to add: what in the world is the syntactic
> difference between
> /start/../end/
>
> and
>
> if (/start/../end/)
>
> that the former is rejected, and the latter is accepted by the
> interpreter?

It could be the precedence of "if" and "..".

Regards

    robert