- From: Jonas Sicking <jonas@sicking.cc>
- Date: Thu, 1 Dec 2011 16:17:34 -0800
- To: Charles Pritchard <chuck@jumis.com>
- Cc: Yehuda Katz <wycats@gmail.com>, "Tab Atkins Jr." <jackalmage@gmail.com>, Nicolas Mollet <nico.mollet@gmail.com>, public-webapps@w3.org
On Thu, Dec 1, 2011 at 3:58 PM, Charles Pritchard <chuck@jumis.com> wrote: > On 12/1/11 3:48 PM, Jonas Sicking wrote: >> >> On Thu, Dec 1, 2011 at 1:51 PM, Charles Pritchard<chuck@jumis.com> wrote: >>> >>> There are serious security implications for enabling CORS, even with >>> session-less requests. >>> It's going to be a very long opt-in process for file sharing services. >> >> This is a very strong statement backed up by absolutely no information >> or data at all. Not very convincing. >> >> Please clarify what you are referring to. > > Direct and anonymous read access is a very new thing. > > At it's most basic: UAs have always required a server, somewhere, to proxy > anonymous requests. With direct access, items like IP-based security and > auditing are not as reliable. It'd be very easy to do screen scraping on > sites that don't particular want scraping to be done. Please clarify what you mean by "like IP-based security". It seems to me that it's *only* IP-based security that is affected. > While it's easy now, it has to be done from the server-side. When hosts open > up their servers, they're allowing it to be done client-side. How does it makes a difference that it can be done client-side rather than server-side? Beyond IP-based security that is. In other words, I can see that if you're deploying IP-based filters to attempt to make certain parties unable to scrape your website you'll have a harder time doing this if you're also sending a "Access-Control-Allow-Origin: *". However you'd already signed up for a basically impossible task since it's very easy for a screen-scraper to proxy their requests through various parties around the globe in order to avoid IP-based filters. The one threat that I can see is if your website does IP-based authorization. I.e. automatically giving certain people access to private areas of your website if their request come from certain IP-numbers. But even then, I *think* sending "Access-Control-Allow-Origin: *" only allows for additional *reading* attacks. Requests that have side-effects can still be placed by faking which IP number they originate at. I'm not fully sure about this though. Someone that knows TCP/IP infrastructure better than me should confirm. Are these IP-based filters that you are referring to when you are saying "serious security implications"? Are they really common? I've never heard of sites doing that but I'm sure they exist. And to be even more on-topic, does Photobucket or Dropbox use IP-based filters to prevent read access? / Jonas
Received on Friday, 2 December 2011 00:18:38 UTC