W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > July to September 2000

RE: [hwarncke@Adobe.COM: Re: [dav-dev] Depth Infinity Requests]

From: Wright, Tom <Tom.Wright@gbr.xerox.com>
Date: Fri, 7 Jul 2000 08:45:28 +0100
Message-ID: <117E8D1FCCA3D111940C00805FEF6060016A182C@gbrwgcms01.wgc.rx.xerox.com>
To: "'Brian Morin'" <bmorin@microsoft.com>, "'WebDAV WG'" <w3c-dist-auth@w3.org>
Cc: "Hall, Shaun" <Shaun.Hall@gbr.xerox.com>
While what you say is true, it puts a huge load in the network, one of the
purposes of infinite requests was to perform an operation all in one go on
the server with out excessive comms from the client to the server ala
pipelining.  

Handling one large request is far more efficient that handling lots of small
requests, particularly when the resources are not on a file system , e.g.
accessing a Database or other custom repository.

I strongly oppose the removal of PROPFIND infinite depth requests.

As an aside, as for n depth requests, how deep is n ?? It seems a bit
arbitrary.

Regards

Tom
<All opinions are my own, not necessarily that of my employer>

-----Original Message-----
From: Brian Morin [mailto:bmorin@microsoft.com]
Sent: 06 July 2000 21:44
To: WebDAV WG
Subject: RE: [hwarncke@Adobe.COM: Re: [dav-dev] Depth Infinity Requests]


Wouldn't the scenario Hartmut gave be better addressed with a DASL request
to pull up only the list of modified files?

Doing lots of depth 1 propfinds is not all that bad.  If you pipeline the
requests you can elminate a lot of the latency, assuming a tree with
non-trivial breadth.  You get hiarchical structure built in (propfind XML
entries in a depth infinity response do not appear to be ordered, leaving
the client to rebuild the hiarchy from URLs.)  Finally, you are sending a
lot of small requests to a  web server, which is usually designed to
efficently handle a lot of small requests.

-----Original Message-----
From: Hartmut Warncke [mailto:hwarncke@Adobe.COM]
Sent: Thursday, July 06, 2000 8:23 AM
To: WebDAV WG
Subject: Re: [hwarncke@Adobe.COM: Re: [dav-dev] Depth Infinity Requests]



I am of the opinion that disallowing  depth infinity PROPFINDS
would cut off one of the main WebDAV features.  In order to
explain that opinion I  would  like to give you a short example
how we use WebDAV in Adobe GoLive 5.0 (the following sample
is not a theoretical one it's out of the everday life of a GoLive user):

Imagine that you need the modification dates of  1000 resources
on the server in order to synchronize the resources on the server
with the resources on your local machine. Even if the resources
are located in a directory with a very deep hierarchy it's not a
problem with a depth infinity PROPFIND. You ask the server for
the dates and get the response for all the 1000 resources within one
Network request/ response.

On the other hand if you stick to PROPFIND Depth: 1 you end up
in a tremendous drawback (a ftp-like solution) because the client
software has to go down the hierarchy step by step and ask for the
content of each directory  level. The User has  to be very patient
in that scenario ...

So to my mind the infinity PROPFIND is a main WebDAV feature
for practical purposes.

In addition to that I would like to draw your attention to the fact
that it is very difficult to base a software implementation on a
WebDAV RFC which changes frequently because we are not able
to ship a new Release of our software every two weeks.

Hartmut

"Clemm, Geoff" wrote:

> I think we should disallow Depth:infinity on a PROPFIND, and require that
> the client pass in some maximum depth to let the server know when it
> should stop.
>
> Cheers,
> Geoff
Received on Friday, 7 July 2000 03:56:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:54 GMT