The format of a label in a Revision-Selector header
Geoffrey M. Clemm (gclemm@tantalum.atria.com)
Sat, 16 Oct 1999 14:41:26 -0400
Date: Sat, 16 Oct 1999 14:41:26 -0400
Message-Id: <9910161841.AA20467@tantalum>
From: "Geoffrey M. Clemm" <gclemm@tantalum.atria.com>
To: ietf-dav-versioning@w3.org
Subject: The format of a label in a Revision-Selector header
I will propose that the format of a label in a Revision-Selector
header should just be a URL segment (i.e. encode the label as a
sequence of URL pchar's).
To see how this is not crazy (:-), it is important to distinguish
two use cases for labels.
The first use case is one where the user (i.e. an actual human)
memorizes the name of a label, gives it to a browser, and the
browser passes that string directly down to the server in the
Revision-Selector header.
In this case, I believe that a URL segment will be about as much
as a human can reasonably be expected to remember. Anything more
complicated (such as the id for the character set encoding) is
just not something the user will remember or be willing to type in.
The second use case is where the client gives the user the
opportunity to point at (select) some nicely formatted display
(possibly even a graphical icon), and then the client encodes
that selection into some obscure string (including char-set,
language, whatever) that then gets passed down to the server
in the Revision-Selector header.
In this case, I still believe that this string can easily be
encoded into a URL segment by the client before sticking it in the
Revision-Selector header.
So either way, a URL segment is reasonable. QED (:-).
I further propose that a revision identifier be encoded as a URL
segment, using similar kinds of arguments.
Objections?
Cheers,
Geoff
Note: the issue of whether revision identifiers and revision labels
share the same namespace or not is an orthogonal question, which
I believe will require dueling pistols at 10 paces to resolve ...
where this duel is currently scheduled for the Washington IETF (:-).