- From: Murray Maloney <murray@sq.com>
- Date: Fri, 30 Aug 1996 12:03:39 -0400
- To: Christopher Seiwald <seiwald@perforce.com>
- Cc: murray@sq.com, seiwald@perforce.com, dgd@cs.bu.edu, ejw@ics.uci.edu, w3c-dist-auth@w3.org, www-vers-wg@ics.uci.edu
More questions below... At 08:46 AM 30-08-96 -0700, Christopher Seiwald wrote: >Much has floated around about this that I mean to answer, but for >now Murray has posted the most straightforward inquiry. > >| From: Murray Maloney <murray@sq.com> >| Subject: Re: Seiwald Q & A -- "GET for EDIT" cookies >| >| When a user "checks out" a document for editing, >| the revision contrl system "should" record who it >| is that is checking out the document so that when >| the same user attempts to "check in" the document >| there is a mechanism to say "Hey, remember me? I am >| checking in the document that I previously checked out >| for editing. Here it is." The RCS can verify that it is >| the same user that is recorded and proceed, or reject >| the action if it is not the same user. >| > >Very close, except I'm not asserting that the revision control system >_should_ record anything when a user starts to edit a document. Instead, >I say that there are many systems that _do_ record something, and that >HTTP _should_ cart around a token ("or cookie" ) of this recorded >information. Right. I quoted "should" to indicate that it was a goo idea, not a requirement. The cookie, in this case, is like a certificate that lets the user know that the transaction was completed and that the document has officially been checked out. > >For something like RCS or CVS, the cookie might be only a name and rev >of the document. For Clearcase or Perforce, the cookie might be an >inscrutable pointer to info in its database. For less version-stringent >systems, there may be no cookie at all. I would think that the cookie should only contain the most basic of information, such as a pair of values to validate the check out (an ID) and a record of the cookie that was issued. The RCS could, at its discretion, maintain a database of the pointer into the database etc. > >| Christopher is asserting that a "cookie" is the best, >| if not the only, way to manage the session. > >Perhaps there is another way? Certainly the document itself could carry information along with it, in a META tag or on any tag that we can legitimately propose. This would allow arbitrary authoring tools to have access to whatever information the RCS provides -- without having to inspect any cookies that might be associated with the document. I am including Notepad in my definition of arbitrary authoring tools. For debugging purposes, I think that it will be useful to carry info in the document. > >Christopher > > > Murray Maloney "Life is a daring adventure, Technical Director or it is nothing" SoftQuad Inc -- Helen Keller
Received on Friday, 30 August 1996 12:07:13 UTC