W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1995

Re: Moving HTTP 1.0 to informational

From: Dave Long <dave@navisoft.com>
Date: Wed, 27 Dec 1995 12:38:41 -0800
Message-Id: <199512272038.MAA16439@geek.navisoft.com>
To: Fisher Mark <FisherM@is3.indy.tce.com>
Cc: HTTP Working Group <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>

>Well said.  I would fully expect to have my users beating down my door if 
>they always had to upload their files twice because of authentication 
>failures/requests!  We are talking file upload here, not "upload lotto".

As one can tell by my previous message, we are concerned about having
to upload files twice, but there are many cases where it isn't as bad
as it sounds:

	- on a LAN, twice is often only a second or two longer than once

	- even over dialup, because of the way the URL permission
	  scheme works, we don't have that many twices:

		if someone saves something they edited, they
		had to GET it to begin with, so if they used
		a PUT-authorized identification, the PUTs go
		through on the first try.

		if someone does a save as, the embedded assets
		tend to fall under the same path-prefix and
		have the same access control as the HTML file,
		so even if the HTML file (which is short) has
		to be retransmitted, the images (which may be
		large) tend to go through on the first try.

So, in regular use, there aren't that many retransmissions (certainly
not "always twice").  Where we get bitten by permissions are where we
fake up atomic transactions by saving an entire URL tree in one HTTP
transaction -- if one does a "save as", one may wind up with many MB
hitting a particular URLspace for the first time.

-Dave
Received on Wednesday, 27 December 1995 12:43:02 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:38 EDT