- From: Xidorn Quan via GitHub <sysbot+gh@w3.org>
- Date: Thu, 07 Jul 2016 07:58:42 +0000
- To: public-media-capture-logs@w3.org
> the sandbox model is so restrictive by default (and with a growing & unbounded set of restrictions) that it is unlikely to be used Sandbox via iframe has a use count of 0.6% of sampled pages from Chrome's stats. Why do you think it is unlikely to be used? (There is no data for how many pages have a cross-origin iframe, though) The sandbox model which has new potentially dangerous feature disabled by default is something we hope people to use for their cross-origin iframes in general, right? And by making this feature disabled by default for even non-sandbox, aren't you just making the normal model be aligned with the sandbox model that has a growing and unbounded set of restrictions which you are disapproving? > using `getUserMedia` from an iframe from a different origin ought to be the exception rather than the norm This actually applies to many of other sandbox flags, including pointerlock, top navigation, modals, and probably even popups. With time goes by, people may find new ways to escape from restrictions. By making restrictions a flag in sandbox, spec and impls are able to handle them altogether, rather than fixing each individual attributes. Also without sandbox, pages can escape from this restriction via a top navigation or popup. > Now if current usage doesn't allow that change, it is reasonable to reconsider that approach; but it would be useful to have more details on what metrics tell us (if we have any) TBH, based on Chrome's stats, I don't think the usage would really be a blocker. (GetUserMediaInsecureOriginIframe <= 0.0001%, GetUserMediaSecureOriginIframe <= 0.0001%, so combined at most 0.0002%) -- GitHub Notification of comment by upsuper Please view or discuss this issue at https://github.com/w3c/mediacapture-main/issues/268#issuecomment-231009332 using your GitHub account
Received on Thursday, 7 July 2016 07:58:52 UTC