- From: Yves Lafon <ylafon@w3.org>
- Date: Tue, 13 Jul 1999 13:57:35 +0200 (MET DST)
- To: Dan Hansen <DLHansen@Quark.Com>
- cc: "'www-jigsaw@w3.org'" <www-jigsaw@w3.org>
On Mon, 12 Jul 1999, Dan Hansen wrote:
> Augusto,
>
> I've included a question at the bottom - but since some won't make it down
> there
> I'll include it up there as well.
>
> Can this server start up cold turkey - meaning without all the store and
> indexer files?
> If not, is this a design decision? Or something that just happened? If I
> wanted to
> base work on top of this architecture, what approach would the designers
> suggest?
Of course it can, the basic class of the root resource is given by a
property, but no frame will be attached to that resource (unless it
auto-attach a frame like HttpPassDirectory, the property is:
org.w3c.jigsaw.root.class
To be able to index things, you will have to create also the indexers.
> ....
>
> I've also been looking into using some of the Jigsaw classes in an app.
> Even more...
> I'd hoped to replace the request/reply classes in the org.w3c.jigsaw http
> package with my
> own versions that live in my package but use the org.w3c.tools packages,
> etc.
org.w3c.jigsaw.http.* are related to server request/replies, you may want
to take a look at org.w3c.www.protocol.http.* for client-side
request/replies (and org.w3c.www.http also)
> I find that it is easy to get lost in all the ResourceReference and
> container stuff that
> is getting tossed around internally. It is not clear how all the indexing
> is done or what
> files are really needed when the server starts up.
Basically, you don't have to mess with all that, and you should let the
indexers do the trick (aka create the resources and such), but resources
are easily extensible, and the filters have a very small api.
Hope this helps,
/\ - Yves Lafon - World Wide Web Consortium -
/\ / \ Architecture Domain - Jigsaw Activity Leader
/ \ \/\
/ \ / \ http://www.w3.org/People/Lafon - ylafon@w3.org
Received on Tuesday, 13 July 1999 07:57:45 UTC