Peter Stephenson wrote:
> On Wed, 25 Mar 2009 02:20:02 -0700
> Jon Strait <jstrait@moonloop.net> wrote:
>   
>> A few adjustments since last time, with documentation.
>>
>> No reset of the special variables is done on a match failure.
>>     
>
> Er, I can't remember what Phil said (I haven't been following this in any
> detail), but the documentation now says variables aren't altered on a
> failure, so presumably that is now incorrect.  I don't think this is
> crucial as long as it's documented correctly.
>
> Could you in any case send a documentation patch against the current source
> and with lines wrapped to 80 columns?
>
> Thanks.
>
>   
No, I ended up keeping the original behavior:  On match failure, none of 
the special variables are modified (reset).

Here is the updated doc patch.

Please let me know if anything I added isn't clear enough.  Thanks.