W3C home > Mailing lists > Public > www-tag@w3.org > January 2012

Re: ACTION-647: Prepare product page for work on Client-Side Storage

From: Noah Mendelsohn <nrm@arcanedomain.com>
Date: Thu, 26 Jan 2012 11:47:17 -0500
Message-ID: <4F218395.6040001@arcanedomain.com>
To: Robin Berjon <robin@berjon.com>
CC: ashok.malhotra@oracle.com, www-tag@w3.org


On 1/26/2012 5:20 AM, Robin Berjon wrote:
> Understood. I would tend to think that that's okay since the items in
> the IndexedDB might not be ones that are usefully exposed to anything
> beyond the code that was written to manipulate them.

The concern I have is that while the entries "might not" be for information 
exposed outside the application, it's also quite possible that the opposite 
is the case.

The situation I'm concerned about is where an Ajax-style application is 
written to manage what we would on model on the Web as documents identified 
by URIs. Examples might include resumes, perhaps e-mails (which IMO should 
be identified by URIs on the Web), maybe views of maps, etc. As long as 
such applications run only connected, Web Arch is pretty clear on what to do.

My concern is that such an application might be written to run both online 
and offline, and might put into IndexedDB or some other local store the 
same documents that when connected are identified by URI.  Good reasons for 
doing this, rather than using a traditional HTTP proxy cache would 
including allowing disconnected update and creation of the documents. Use 
cases I'd like to see considered include the creation of hyperlinks that 
resolve to the "correct" document (I.e. the resume or whatever that the 
user wants), regardless of whether the application happens to be connected 
and/or whether it happens to have already retrived a copy of the document. 
The questions I'd like to see this finding address are: 1) what is the 
correct use of URI in such situations? and 2) what is the architectural 
relationship, if any, between the use in this manner of IndexedDB-style 
local storage vs. HTTP proxy caches as described in the pertinent RFCs.

FWIW: I agree with the other case that you describe. If the information is 
purely local to the application, not directly exposed to the user in the 
form stored, then there may be no issue of creating URIs or allowing 
linking. I'm concerned about the other case. Thank you.

Noah
Received on Thursday, 26 January 2012 16:47:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:44 GMT