Re: possible parsing problem with OSI-DS 23 (Distinguished Names)

[Allan Cargille <[e-mail]>]

} OU="ba"d"  is illegal, as '"' is a special, and thus not in stringchar.
} The way to deal with this is OU="ba\"d".   Perhaps I should add some text
} on this.
} 
} Steve

Hi, maybe I'm being dense, but my reading of the BNF is that OU="ba"d" is
completely legal.  Here's the BNF from version 5:

    <string> ::= *( <stringchar> | <pair> )
	     | '"' *( <stringchar> | <special> | <pair> ) '"'
     | "#" <hex>


    <special> ::= "," | "=" | '"' | <CR> | "+" | "<" |  ">"
	     | "#" | ";"

    <pair> ::= "\" ( <special> | "\" )
    <stringchar> ::= any char except <special> or "\"

My understanding is that using the definition

    <string> ::= '"' *( <stringchar> | <special> | <pair> ) '"'

is that a <string> has a double-quote, then as many <stringchar>s,
<special>s, or <pair>s as I want (intermixed), and then another
double-quote.  In that case OU="ba"d" would be legal.  It could be
parsed as

    <string> ::= '"' <stringchar> <stringchar> <stringchar> <special>
    		     <stringchar> '"'

I also believe that under the current grammar, the value OU=""" would
also be valid (don't ask me why someone would use this!  Do you intend
for that to be valid?  I think we would want that represented as
OU="\"".

I believe that the grammar needs to be fixed in some way or another.
I think that one way to fix it would be to restrict the definition of
<string> that I'm using to

    <string> ::= '"' *( <stringchar> | <pair> ) '"'

However, that's nasty because it requires all the other special
characters to be quoted with backslashes (not nice).  I would prefer to
see us distinguish between "normal" special's and the double-quote.
    
You're right on a separate point,  I was misreading the rule for <pair>
(adding a trailing backslash as well).

Hope I'm not just being dense.  Am I right?  Or can someone explain this
to me in more detail?

Cheers,

allan