[Council] JEP-20 Comments

Mike Lin mikelin at MIT.EDU
Wed Jul 31 23:40:11 CDT 2002


this is an arbitrary example, but say you're negotiating an HTTP file
transfer where either you or i listen, and where we either use HTTP 1.0
or HTTP 1.1. (i know HTTP version is kind of silly, but bear with me for
the moment). so we could permute:

<option>http-1.0-get</option>
<option>http-1.0-post</option>
<option>http-1.1-get</option>
<option>http-1.1-post</option>

now say that you would also like to decide whether to use TLS for the
transfer connection,

<option>http-1.0-get</option>
<option>http-1.0-get-tls</option>
<option>http-1.0-post</option>
<option>http-1.0-post-tls</option>
<option>http-1.1-get</option>
<option>http-1.1-get-tls</option>
<option>http-1.1-post</option>
<option>http-1.1-post-tls</option>

add another on/off option and you get 16 elements...blah blah
combinatorics blah blah.

so all i really demand in order to be "-0" on this is a way to have
nested options, so that the syntax is at least somewhat extensible. i
have other philosophical issues with it, but whatever.

-- mike --

On Wed, 2002-07-31 at 14:12, Peter Millard wrote:
> Mike Lin wrote:
> > I think the larger objection I have is that the features are specified
> > as a rudimentary flat string. Joe points out that for version numbers,
> > you could simply add it to the string; but as soon as you have
> > additional attributes to specify, are you then supposed to permute
> > over all combinations?
> > IMHO, this is a context in which the use of a more structured system
> > for describing features is warranted.
> 
> I'm not sure I see the need for this added level of complexity.. Can you
> give me a concrete example? Different "versions" of a single feature should
> be treated as different options for that feature. If I only support "ver 2"
> of something, thats the only one I list. And if you only support ver 1, then
> we can't do whatever it is we were trying to do. Seems cut and dry enough
> for me.
> 
> pgm.
> 
> 
> _______________________________________________
> Council mailing list
> Council at jabber.org
> http://mailman.jabber.org/listinfo/council





More information about the Council mailing list