Re: Link Verification

   Date: Tue, 7 May 1996 16:44:30 -0700
   From: (Richard Connamacher)

   At 3:47 PM 5/7/96, Joseph Arceneaux wrote:
   >I'm addressing this group as a last resort;  I've not seen this
   >addressed anyplace.
   >My problem is that I have a database of HTTP links, and I would like
   >to batch-check them to see if they remain valid or have changed in
   >some significant way.

   I believe the way to do that is to use the HEAD method in your request, so
   that the request line will read "HEAD /folder/page.html HTTP/1.1".  The
   response will be all the headers that it would normally send before that
   document in a GET response, but no document.  This is the standard method
   for verifying links or checking if a document has changed without actually
   getting that document.

   I personally am having trouble implimenting the HEAD method, even though
   the HTTP 1.1 specs, which I am trying to follow, require it.  The problem
   is that my server is designed to generate pages on the fly, and each page,
   in all likelyhood, will be different every time the client retrieves it.
   Some pages actually carry out an action, such as move your virtual persona
   into a new confrencing area, although sensitive actions would be reserved
   for POST requests.  What would be the standard response in this case?
   Right now, it only responds with 501 Not Implimented.


It seems to me that there is a substantial difference between "data"
pages, which provide information even though the HTML may be generated
on the fly, and "action" URL's which, though they may generate HTML
as a side effect, primarily cause some action.  Perhaps such action
pages such return a new error code for HEAD requests?

I think a more general sheme would be standardization of some META
tags for, e.g., last-modified, page moved to, ...

All comments welcomed,

Received on Wednesday, 8 May 1996 09:16:29 UTC