Message-Id: <m0p5PZZ-000306C@rip.psg.com>
From: randy@psg.com (Randy Bush)
Subject: Re: Minor clarification on url: prefix
To: uri@bunyip.com (URI mailing list)
Date: Thu, 2 Dec 1993 17:44:41 -0800 (PST)
In-Reply-To: <CHFCxA.A5J@pandora.sf.ca.us> from "Mitra" at Dec 2, 93 08:29:33 pm
> Does anyone think that
> URL: ftp://ftp.path.net/pub
> does NOT conform to the decision of Houston, and that white space
> after the "URL:" is illegal.
Yes,. I do. As a complier hacker, I find all this equivocating language
around 'prefixes', 'wrappers', and the like very strange. I am utterly
befuddled by what appears as attempts to save a byte or pass responsibility
for parsability to some external syntactic framework.
A URL should be a simple syntactic entity described by a simple grammar.
That grammar should not describe some odd text that is 'outside of' or
'around' the URL (i.e. a 'wrapper', 'prefix', ...). It's either the URL or
it is not.
Maybe, if we can demonstrate a strong need, there could be alternative forms
of a URL depending on context, i.e. when seen in human readable text. But
why? To save a couple of angle brackets? Then just don't use them and let
a URL be something like
"URL:" <blah blah> "%."
or whatever.
Just as the start of a URL should be easily detectable, so should the end
be. Humans and code should be able to parse the darn things without a lot
of guesswork.
Now that we understand that the URL, from starting token to ending token is
a single entity with a single grammar (pardon for assumption of assent), the
decision against embedded blanks obviously precludes your proposed change.
> Of course - if noone is willing to let go of their own preferences in
> favour of the groups ...
Let me see if I understand you correctly. "I am proposing a change. If you
do not go along with this change, you are being anti-social and are
destroying the whole WG's purposes." I will be polite and go no further.
randy