From: mitra@path.net (Mitra)
Subject: Re: Final edits - summary of outstanding points.
Message-Id: <CF7tL1.18n@pandora.sf.ca.us>
Date: Wed, 20 Oct 1993 21:41:24 GMT
Tim Berners-Lee (timbl@www3.cern.ch) wrote:
: >3 & 6) Are fragment identifiers part of the URL - I think they were
: >removed at
: >the IETF, Tim thinks they are still in.
: The discussion is in but the URI BNF
: defines a URI as not including the fragment ID. OK?
I dont think so - having a BNF not match the text is a bad move. The current
version of the document <http://info.cern.ch/hypertext/WWW/Addressing/URL>
currently retains the existing discussion which says that fragments are allowed
in the URL. How about some comment on it, that this is historical usage
retained until UR*s handle it in some other way?
: >9) Does the news URL need rewriting - I think the current spec is a
: > URN, Tim (& Marc ?) think its fine.
: Yes, it's fimne and in use and useful and not suitable
: as a URL.
Tim, I presume you either mean "is suitable as a URL" or "is not suitable
as a URN". Marc suggested a reasonable compromise, i.e. grandfathering this
into the URL definition, but also defining a real URL for news. I suggest
this would be "news://server/group/article-id" its easily distinguishable
from the old version so it shouldnt confuse existing clients. I'd also
be open to it having a seperate scheme id nntp://server/group/article-id.
: >10) Should the length field be removed from the wais definition.
: > Everyone agrees its no longer needed. The question is whether
: > to leave it in for backwards compatibility with WWW.
: I'd be prepared to remove it from our WWW code. Objections?
If I dont here any, then I'll remove this from the outstanding points list.
:
: >16) Should the type field be retained in the Gopher0 URI, I think
: >its redundant, I think Tim believes its needed?
: Yes. It is part of the gopher link and to say that
: you can do a retrieval without knowing what you are
: getting back is in principle true but quite useless!
See other points on the list - this is definately an outstanding issue.
:
: >17) Do we allow trailing "." in FQDNs - I forget who took which side
: > here?
: I agreed to remove the bad example, so we leave the
: spec as not allowing the trailing dot.
I've left the BNF off of this list, since once we agree on the rest, then
knocking the BNF into shape to match the text is easy.
- Mitra