Bug #2547: error when interpolating a string containing a single backslash into a regexp

• Status: Closed
• Assignee: matz (Yukihiro Matsumoto)
• Target version: -
• ruby -v: ruby 1.9.1p376 (2009-12-07 revision 26040) [x86_64-linux]

[coatl (caleb clausen)]

=begin
An error is raised if an interpolation in a regexp contains a string containing a single backslash. This case used to work just great in ruby 1.8. For example:

$ ruby19 -e 'a="\"; p /#{a}y/'
-e:1:in `': too short escape sequence (ArgumentError)

Note that if the string literal is inlined directly into the interpolation, the error does not happen. I'd guess that ruby optimizes this case away at parse time, and thus sidesteps the error:

$ ruby19 -e 'p /#{"\"}y/'
/\y/

I've tested this in 1.9.1; presumably 1.9.2 has the same behavior, but I don't have a 1.9.2 readily available.
=end

[naruse (Yui NARUSE)]

• Status changed from Open to Assigned
• Assignee set to matz (Yukihiro Matsumoto)

=begin
trunk also reproduce this.

However this may be a feature:

• Ruby's regexp forbids to split character escape like 'a="\"; p /#{a}y/'
• /#{"\"}y/ is optimized as /\y/ on parser, so this is valid
  =end

[coatl (caleb clausen)]

=begin
I can see why it would be easier to forbid splitting up escape sequences in regexp. But:

1. Ruby 1.8 did handle this; it can't be all that impossible.

2. If splitting escapes is forbidden, then for consistency, the directly inlined version should be forbidden as well.
   =end

[nobu (Nobuyoshi Nakada)]

• Status changed from Assigned to Closed
• % Done changed from 0 to 100

=begin
This issue was solved with changeset r27431.
caleb, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

=end