Errors-to: owner-drums@CS.UTK.EDU
Message-ID: <AjolLTy00WBw49xuhz@andrew.cmu.edu>
Date: Mon,  5 Jun 1995 10:34:07 -0400 (EDT)
From: John Gardiner Myers <jgm+@CMU.EDU>
To: drums@CS.UTK.EDU
Subject: Re: address syntax
In-Reply-To: <1786.802222517@munnari.OZ.AU>
References: <1786.802222517@munnari.OZ.AU>
Beak: is Not

Robert Elz <kre@munnari.oz.au> writes:
> The example you gave is what was intended by "heuristic" I
> suspect, that is, shortcut parsing, not looking at the syntax
> in detail

There is nothing heuristic about such parsers.  Given any route-addr,
it will in all cases correctly determine which part is the route and
which part is the addr-spec.

It is a correct parser because a colon cannot exist in a domain, so
the first colon after a "<@" must be the terminating colon of the
route.

The fact that it does not do a detailed parse of the route is
irrelevant.  This is a top-down parsing approach, the program can
further parse the route as it needs.

> In any case, any stripping code that doesn't ensure that the final
> '@' remains in the address, which seems to be what might go wrong
> in the example you gave seems terribly fragile to me, similarly
> any stripping without determining that the address is syntactically
> correct first.

Take a modification of Harald's example:

	<@[ip:v:6]:user@final.host.com>

This is syntactically illegal, so one can parse it as one wishes.  One
way to parse this would be:

	<@[ip:v:6]:user@final.host.com>
          ^-^ ^-------^ ^-------------^
	domain    |         domain
              local-part

         ^---^^-----------------------^
         route       addr-spec

and that's what the top-down parsing approach will produce.

-- 
_.John G. Myers		Internet: jgm+@CMU.EDU
			LoseNet:  ...!seismo!ihnp4!wiscvm.wisc.edu!give!up