Re: Proposed HTML ping attribute

Julian Reschke writes:

> For the record: a GET (or a HEAD) request can be sent in a way 
> (cache-control: no-cache) such that intermediaries are disallowed to 
> return a cached response. So cacheability itself IMHO is not an argument 

> in favor of POST.

Fair enough, but I think it's also true that a user agent, search spider, 
etc. would be compatible with Web Architecture if it did a GET on the URIs 
in question in cases where the user had not in fact cliked on a link.   I 
can't see a reason why software that's written with awareness of the ping 
attribute would do this, except maybe maliciously, but as far as I know 
GET is distinguished not just by being (typically) cacheable, but also by 
the fact that you can do a GET on any URI at anytime.  If someone found 
out that their link access statistics were messed up due to various agents 
doing GETs, I think Web Architecture would give them very few grounds for 
complaint.

Noah

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Wednesday, 23 January 2008 02:28:12 UTC