From: matz@... (Yukihiro Matsumoto)
Date: 2003-04-08T14:17:16+09:00
Subject: Re: Is this an intended change in string interpolation

Hi,

In message "Is this an intended change in string interpolation"
    on 03/04/08, Dave Thomas <dave@thomases.com> writes:

|In the latest CVS:
|
|irb(main):002:0> /c(a)t/ =~ "cat"
|=> 0
|irb(main):003:0> "d#$1ve"
|=> "d"
|
|In older Rubies, the #$1 sequence was interpolated without swallowing 
|the rest of the string.
|
|Also now I can create globals such as
|
|irb(main):004:0> $1ve = "cat"
|=> "cat"
|irb(main):005:0> "d#$1ve"
|=> "dcat"
|
|Cool feature, or wee bug?  :)

From CVS history, I allowed variables like "$1ve" on 2000-04-10.  The
problem is I completely forget the reason why I did allow "$1ve".  But
later, I explicitly disallow variables like "@1ve".  So, for the
consistency, I should disallow "$1ve" as well.

OK, I will change the behavior to:

/c(a)t/ =~ "cat"
=> 0
"d#$1ve"
=> "dave"

$1ve = "cat"
-:1: syntax error
$1ve = "cat"
    ^
Thank you for pointing out inconsistency.

							matz.