Re: Matching programming language identifiers, not "words"

[Tim Van Holder]

Jim Blandy wrote:

> Here's the patch --- please test it.  Once we've got some positive
> reports on it, I think it'll be ready to go in.
>
>
> *** src/regex.c.~2~     2004-04-29 15:56:53.000000000 -0500
> --- src/regex.c 2004-04-29 17:44:24.000000000 -0500
> ***************
> *** 1095,1100 ****
> --- 1098,1110 ----
>         case wordend:
>           printf ("/wordend");
>  
> + 	case symbeg:
> +         printf ("/symbeg");
> +         break;
> + 
> + 	case symend:
> +         printf ("/symend");
> + 
>  	case syntaxspec:
>           printf ("/syntaxspec");
>           mcnt = *p++;
>  

This part seems suspect - I don't think the fall-through from wordend to 
symbeg is intentional.
You may need to combine wordend and symend so both will fall through to 
syntaxspec (which
I assume is the intended behaviour).

> *** man/search.texi.~1~ 2002-07-06 08:44:06.000000000 -0500
> --- man/search.texi     2004-04-29 17:38:41.000000000 -0500
> ***************
> *** 672,677 ****
> --- 672,689 ----
>  @item \W
>  matches any character that is not a word-constituent.
>  
> + @item \_<
> + matches the empty string, but only at the beginning of a symbol.  A
> + symbol is a sequence of one or more word or symbol constituent
> + characters.  @samp{\_<} matches at the beginning of the buffer only if
> + a symbol-constituent character follows.
>  
perhaps a new term should be introduced for word+symbol - otherwise the 
last part of this paragraph
would probably have to read "... if a word- or symbol-constituent 
character follows" to be
unambiguous.  The same goes for the \_> doc.

-- This e-mail was scanned by RAV Antivirus --