Re: syntax howto question

[Sean Middleditch]

Ugh.  That sounds less than intuitive.  ~,^  I'll mess around and see if
I can get it working.  Thankies.

On Sun, 2002-11-03 at 09:04, Thomas Troeger wrote:
> Hello,
> 
> The problem is between rule #2 and #3. After parsing "expr '.' name",
> there is a s/r conflict. The canonical collection is somewhat like
> this:
> 
> ...
> state 7
> [EX := EX . NAM . ( ARG ) , (] *** s/r ***
> [EX := EX . NAM . ( ARG ) , .] *** s/r ***
> [EX := EX . NAM . ( ARG ) , #] *** s/r ***
> [EX := EX . NAM . , (] *** s/r ***
> [EX := EX . NAM . , .]
> [EX := EX . NAM . , #]
> ...
> 
> So the parser doesn't know if it should reduce from here, or shift
> over the next '(' token to complete with "expr '.' name '(' args ')'".
> Maybe you could insert a special token which is generated by
> the lexer if it has read "expr '.' name" and the next symbol is NOT
> '('. Then you could modify your rules like this:
> 
> expr: ...
>   | expr '(' args ')' { ... }
>   | expr '.' name '(' args ')' { ... }
>   | expr '.' name SPECIAL_TOKEN { ... }
> 
> I hope this approach does work (any comments if this is or isn't
> reasonable pls).
> 
> mfg.
> 
> --tst.
> 
> -- 
> /"\                          "They that give up essential liberty to
> \ / ASCII Ribbon Campaign     obtain a little temporary safety
>  X  Against HTML Mail         deserve neither liberty nor safety." --
> / \                                           Benjamin Franklin, 1759