Toward an unambiguous grammar [Was: FTP syntax ]

Tim Berners-Lee (timbl@ptpc00.cern.ch)
Fri, 25 Mar 94 16:02:28 +0100

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Tim Berners-Lee: "Toward an unambiguous grammar [Was: FTP syntax ]"
Previous message: Tim Berners-Lee: "Toward an unambiguous grammar [Was: FTP syntax ]"

Date: Fri, 25 Mar 94 16:02:28 +0100

From: Tim Berners-Lee <timbl@ptpc00.cern.ch>

Message-Id: <9403251502.AA15327@ptpc00.cern.ch>

To: -s@ptpc00.cern.ch, Outstanding@ptpc00.cern.ch, URL:, points@ptpc00.cern.ch,

URL Specification: Oustanding Issues
OUTSTANDING URL ISSUES

This is a a running list of issues about URLs and about which which
discussion is still running, or recently closed. These are taken
mainly from summaries on the list by various people. The order is
random. As they are settled I'll leave them in as history. See also
the change history[1] .

As editor I am happy with the current spec except I would like the
WG to check whether the large amounts of Gopher+ stuff can't be
changed to a reference to some entity in the Gopher+ spec as we
have done for all other protocols. (see below[2])

FTP

TRANSFER MODE

How should the transfer mode be encoded into the URL, if at all?
The current feeling is that it should be optionally encoded as a
suffix by the use of some delimiter.

Editor suggests[3] the use of "!". Requires adding this character
to reserved set. This was succeded by:

Dan Connolly suggests[4] using ";mode=binary". (Editor points out
need to reserve the ; and = and remebers this must be done for WAIS
anyway. ) Also, Editor suggests using FTP terminolgy, ";type=i".

CHANGE ORDER OF ACCOUNT AND PASSWORD

Alan suggested[5] changing the order around. The editor rejects
this :-), subject to outcry, on the basis that:

This is not the time for incompatibe changes

The change is arbitrary

The order "user@host" has a tradition long established, and to
propose "host@user" would drive people crazy.

Gopher+ and search strings

SEARCH STRING TERMINATOR

Should the normal URI convention of using "?" as a sparator for
search strings be changed to HT (horizonatl tab) in the case of
Gopher because that is how it is represented in the protocol? This
came up from the extention of gopher to gopher+, but is in fact
orthogonan question. Should the URN spec just refer to the Gopher+
spec? See:

new[6] gopher+ version

old[7] gopher version.

Mark 's reasons[8] for switching to %09:

Tim's reasons[9] for keeping "?":

Alex Dupuy's reasons for keeping "?"[10]

Tim's suggestion[11] of making a transparent reference to the
Gopher+ spec

Current state of spec: as Mark wanted.

SLASHES AS NON-HIERARCHICAL DELIMITER

Should it be strictly enforced that "/" never appear unencoded
unless of hierarchical significance? Mark argues no. Others
argue yes. Editor points out that so long as within Gopherworld
noone uses any partial forms, there is no problem in practice
either way, the difference will not surface in any code, so its not
a big deal.

PrePrefix (resolved)

Should a prefix "URL:" or equivalent be considered part of the URL?
Should it be part of a wrapper used when writing a URL within plain
text? Consesnus declared by the chairman[12] to be that "URL:"
should be a mandatory part of the specification fothe URL for use
in all cases. Later comments by the chairman seem to indicate that
it may be optional.

Strongly disliked by the editor as much as requiring the string
"Telephone:" to be a mandatory part of a telephone number.

Consenus: It is in[13], considered part of the URL.

Plain text wrapper

In this area, questions are,

Should one define a wrapper to enable a URL to be spotted within
plain text?

Should it be part of the URL? ( no - alan[14] )

Should it be an appenix item, or mandatory? (appendix -ed)

What characters should be used? Consesnus[15] is "<", ">".
Arguments for[16] and against[17] .

Should it include or replaced the prefix mentioned above? (no
-ed)

The chairman suggests[18] the answers are YES, NO, appendix, "<"
and ">", no.

WRAPPERS IN OTHER CONTEXTS

Larry Masinter suggested that "the wrapper be allowed to be any ONE
OF < > or " " or ' ' or { }, to allow the wrapper to be used
effectively in multiple contexts. This allows maximum flexibility,
some compatibility with SGML (in HTML, the wrapper would be the ""
or '' in <A HREF="url:...."> )."

Others (I forget who, sorry) had felt that in other contexts, the
delimiters are defined.

NNTP

Should a format be allowed to refer to a news article on a
specific local NNTP server by group and article number? (
Mitra's argument[19] , TimBL's argument) (consensus: yes, with
warning)

If so, should the same "news" prefix be used? (consesnus:no)

Should "news" be changed to something else such as "nntp" or
"usenet" in all cases? (Consensus: no)

Tim BL[20]

Next message: Tim Berners-Lee: "Toward an unambiguous grammar [Was: FTP syntax ]"
Previous message: Tim Berners-Lee: "Toward an unambiguous grammar [Was: FTP syntax ]"