[Ex-plain] F&N

Sebastian Hammer quinn at indexdata.dk
Mon Feb 25 21:38:56 CET 2002


Guys,

I am getting a sense here that the idea of combining F&N with ExPlain is 
surrounded by ambivalence. Let's just say for the time being that we treat 
them as separate problems. If it later turns out they overlap, we will 
merge them then..

F&N, then. There is a suggestion to use a simple space-separated list of 
Z-URLs instead of the XML format. I guess that is more compact. On the 
other hand, I don't like it that it really prevents us from ever extending 
the format in any meaningful way (other than assigning a new negotiation 
bit). The benefit of having a bit of XML around it, to my mind it, is that 
you *can* stuff a bit of extra info in there, and clients can ignore it... 
I guess if we postulate a validating parser at the client end, that notion 
kind of blows up a little, though.

How about the carrier mechanism. Yes, it might be worrysome to stuff 
everything in there, but it is a lot easier to shoehorn into most people's 
applications than something (like Explain) that requires a whole separate 
query processor.

Let's also agree right now that it is *bad form* to return this stuff in an 
init unless you are specifically asked to do so with an option bit.

Finally, to deal with the large target lists... would it make sense to have 
a way to simply point to an URL (could be HTTP) and say, "fetch my list 
from there".

This could be generalised (if we went with an XML representation) to a 
general pointer or link to other lists... that allows two different forks 
or level of linking... a server can return a list of other servers and/or 
it can refer you to other lists of servers. Still pretty simple.

--Sebastian

At 17:14 25-02-2002 +0000, Robert Sanderson wrote:


> > > Why bother with <database>?  Why not just return the simple text in
> > > a space separated list? That'll save us having to just discard the
> > > tags anyway.
>
> > Why indeed?  You got _me_ convinced!  Let's go with this format:
> >       copac.ac.uk 210 copac
>
> > No, that's not really true.  But it _is_ a _different_ problem,
> > despite Sebastian's well-intentioned attempts to shoehorn them all
> > into a single problem :-)  Let's solve F&N in the simplest way
> > possible (which I think it what we have above), and then do Explain++
> > properly as a separate project.
>
>I don't think that it's a single problem, but the less new stuff we try to
>introduce, the easier it will be to badger people into implementing.
>
>We could have a space separated list of host port db (authentication?) and
>an XML schema, or we could make sure that the Schema is minimal for when
>you want to do this.  Which I think it is, to the point of not having to
>repeat the host and port.
>
>
> > You should read it -- it fits on one page :-)
> > http://lists.w3.org/Archives/Public/www-zig/2000Jun/0002.html
> > Although Bob Waldstein's comment is also (as always) worth reading:
> > http://lists.w3.org/Archives/Public/www-zig/2000Jun/0003.html
> > He argues that Init otherInfo is the wrong delivery verhicle.
>
>After reading them I tend to agree with him to be honest, though less if
>there's a option bit to be set saying you want it.  (ie opt-in, not
>opt-out)
>
>Especially as Bob's is the only other server (copac, bagel/muffin,
>cheshire) that has any Explain implementation.
>
>Rob
>
>
>--
>       ,'/:.          Rob Sanderson (azaroth at liverpool.ac.uk)
>     ,'-/::::.        http://www.o-r-g.org/~azaroth/
>   ,'--/::(@)::.      Special Collections and Archives, extension 3142
>,'---/::::::::::.    Twin Cathedrals:  telnet: liverpool.o-r-g.org 7777
>____/:::::::::::::.              WWW:  http://liverpool.o-r-g.org:8000/
>I L L U M I N A T I
>
>
>_______________________________________________
>Ex-plain mailing list
>Ex-plain at indexdata.dk
>http://www.indexdata.dk/mailman/listinfo/ex-plain

--
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101




More information about the Ex-plain mailing list