From: Robert Feldt <feldt@...>
Date: 2001-10-25T09:11:44+09:00
Subject: [ruby-talk:23242] Re: Parsing Ruby's interpolated strings/regexps

On Thu, 25 Oct 2001, Yukihiro Matsumoto wrote:

> Hi,
> 
Hi and thanks for your prompt reply,

> In message "[ruby-talk:23237] Parsing Ruby's interpolated strings/regexps"
>     on 01/10/25, Robert Feldt <feldt@ce.chalmers.se> writes:
> 
> |and then recursively parses the Evstr "b+1" when in evaluation of the
> |program with the regexp.
> 
> For the first time only. Evstr stores its parse tree.
> 
> |But if you have a separate parser (ie. not bound to a special
> |interpreter) would you expect the "b+1" string to
> |have been parsed in the Abstract Syntax Tree you get back?
> 
> If you have reentrant parser, yes.  But unfortunately yacc does not
> allow it.
> 
Aha, so its a workaround. I thought you decided to do it that way for some
"philosophical" reason. Thanks, this is much clearer then.

BTW, I know you often have good reasons for your decisions but out of
curiosity: What is the reason for allowing %!str! etc type strings
(ie. the interpolated once without brace/paren delimiters). IMHO,
this is rarely used and doesn't add much (more than complexity in
lexer).

/Robert