W3C home > Mailing lists > Public > www-tag@w3.org > November 2011

Re: Web browsers should preserve the file system Last-Modified time of downloaded files

From: Noah Mendelsohn <nrm@arcanedomain.com>
Date: Wed, 09 Nov 2011 17:35:41 -0500
Message-ID: <4EBB003D.40005@arcanedomain.com>
To: Larry Masinter <masinter@adobe.com>
CC: Scott Penrose <scottp@dd.com.au>, Karl Dubost <karld@opera.com>, Brolin Empey <brolin@brolin.be>, "www-tag@w3.org List" <www-tag@w3.org>
Well, my reasons for not wanting to deal with this in the TAG are a bit 
different. Specifically:

* I think that some of the use cases are not affected by time skew. Very 
often, files bounce through the Internet, and people just want to see the 
modified time they started with. Think things like .dll's, photos, etc. I 
don't really care whether the server clock is in sync with anyone. If I 
upload a photo, and you download it, I generally want you to get all the 
timestamps it had when I first put it up. The maintenance of creation vs. 
modification times is a little unclear to me in these scenarios, but with 
the possible exception of warning if a modified time winds up "in the 
future", I can see at least some reasons for just preserving it as the file 
moves through the network. So, I think this request has some merit.

* With apologies for the fact that Tim suggested bringing this to the TAG, 
I'm unconvinced it's a TAG-level issue. It seems to me that it's a question 
of how user agents deal with one of the details of implementing local file 
save features, and that seems like something to be settled in whatever 
working group is responsible for the user agent feature (I'm guessing HTML 
WG in practice these days). If there is no such WG, and no mandated 
behavior in RFC 2616 or related header specifications (I haven't checked), 
then I would think user agents are free to compete on the quality of their 
file download implementations.

Noah

On 11/9/2011 4:03 PM, Larry Masinter wrote:
> I think the time skew is the determining factor for me...
>
> When two systems are on the internet communicating, they *might* be in sync but often are not. Maybe their clocks are only seconds apart. Maybe the latency between the two systems is small, but I think it is actually unusual for two systems to be in perfect sync. Whether or not it is unusual, it *does* happen.
>
> Then, the "last-modified" time on the server is server-relative.  The client writes a file in its local file system, you might even get a file with "last modified" in the future.
>
>
>
> -----Original Message-----
> From: Scott Penrose [mailto:scottp@dd.com.au]
> Sent: Tuesday, November 08, 2011 7:00 PM
> To: Karl Dubost
> Cc: Brolin Empey; www-tag@w3.org List
> Subject: Re: Web browsers should preserve the file system Last-Modified time of downloaded files
>
>>
>> Le 4 nov. 2011 à 19:18, Brolin Empey a écrit :
>>> The Web browser should preserve the file system last modified time by default because this time cannot easily be recovered after it is discarded.
>
> I have been thinking about this over the last week and which way it should happen.
> It is interesting to note that many older backup systems use last modified as a method to work out if files need incremental backup.
>
> The problem is that once you download a file, they are not linked. The file on your system does not match the remote system. So what does modified mean?
>
> Lets put 3 systems into play.
>
> * Server1 - file.png, last modified 2010-01-01
> * Server 2 - file.png downloaded, last modified 2010-01-01
> * Client 1 - Download file from Server 2, use if mod since
>
> Scenario 1 - Server 2 modifies the PNG (maybe a resize). Last modified should change to that day (2011-10-01), Client can download.
>
> Scenario 2 - Server 1 had updated the logo in 2011-06-01; now Server 2 downloads from Server 1, preserving date modified as 2011-06-01; Client now checks if file has changed since last of 2011-10-01 - it has changed, but now client won't get it. On the other hand, if Server 2 had set last modified to today 2011-11-09 - it would not be an issue.
>
> Scenario 3 - A client downloads a file to Download folder, has a look at recent files. User would expect to see newest files downloaded with todays date.
>
> I don't think HTTP Clients should preserve modified date time. But there are exceptions - e.g. WebDAV / SVN over HTTP.
>
> Scott
>
>
>
>
Received on Wednesday, 9 November 2011 22:36:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:40 GMT