- From: John Franks <john@math.nwu.edu>
- Date: Sat, 19 Sep 1998 08:07:47 -0500 (CDT)
- To: "David W. Morris" <dwm@xpasc.com>
- Cc: Daniel Hellerstein <danielh@mailbox.econ.ag.gov>, http-wg@hplb.hpl.hp.com
On Fri, 18 Sep 1998, David W. Morris wrote: > > I think the trend in general is to move away from user entered URLs. > ... > > Also, adding new syntax to the URL has the possiblity of breaking > existing implementations because the parsing implementaton if often > flakey at best. > Byte ranges (and indeed line ranges) in URLs have been implemented in the WN server since long before the Range: header. They are extremely useful, but are never entered by the user. > > The whole range of ways and kinds of pieces users might want to > request doesn't lend itself to user entry of byte level encoding. > There is no reason the current URL scheme can't retrieve whole documents > as well as pieces with simple URLs ... outside of the scope of HTTP > and URLs, the server implementation defines the interpretation of the > URL. With a SINGLE data file and/or DB, the URLs > http://www.bookserve.com/clancy/HotBook > http://www.bookserve.com/clancy/HotBook/chapter1 > http://www.bookserve.com/clancy/HotBook/chapter1/para6 > could all produce variant levels of zoom into the document. This can > be achieved today on some servers by creative use of mapping to cgi > programs. As I said this has nothing to do with user entered URLs. The functionality which is missing is the ability to make a link to a subset of a document. It can, of course, be done with CGIs, but at great cost in complexity and efficiency. Examples: It is useful to be able to create links on server A to parts of documents on server B without server B having CGIs or a database. It is useful to create a page of customized links to parts of a single document (e.g. a mail archive) based on a query. There are more. On the other hand I agree that adding extra complexity to the URL syntax has serious drawbacks also. At present much of the URL is opaque and a server can do with it what it wants. This is a good thing and allows, for example, WN to implement byte and line ranges in the URL with its own syntax (which users never enter because they don't know). The way to have our cake and eat it too is an extension to the HTML syntax which allows one to create a link to a URL *and* specify that a Range header be sent with the request. > > On Fri, 18 Sep 1998, Daniel Hellerstein wrote: > > > > Thanks. Now I see how to do a GET with range request. > > > Now, another question: > > > Is there a way to specify the offset or range with the http: URL > > > syntax ? > > > i.e. something like http://host/dir/file@range=0,500 > > Just for your information the WN syntax is http://host/dir/file;bytes=0,500 or http://host/dir/file;lines=20,50 You can see more at http://hopf.math.nwu.edu/docs/range.html John Franks john@math.nwu.edu
Received on Saturday, 19 September 1998 06:09:56 UTC