- From: Yoav Weiss <yoav@yoav.ws>
- Date: Fri, 31 Aug 2018 17:56:44 +0200
- To: "ietf-http-wg@w3.org Group" <ietf-http-wg@w3.org>
- Cc: Charles Vazac <cvazac@gmail.com>
Received on Friday, 31 August 2018 15:57:18 UTC
Hello, As part of SRI <https://github.com/w3c/webappsec-subresource-integrity>, browsers became sensitive to response transformations, and HTTP servers need to be aware of that, and avoid such transformations. That's true for intermediaries, CDNs, as well as for HTTP servers which may want to apply optimizations but are not be aware of the content that they are serving. Previously the SRI spec required requests for SRI resources to include a `cache-control: no-transform` headers, to indicate that to servers. However, as Mark pointed out <https://github.com/w3c/webappsec/issues/217#issuecomment-111945173>, the current language says that servers shouldn't apply transformations "to the payload", presumably the request's payload. Would there be any objections to correcting that language, and changing those semantics, such that the `no-transform` semantics of requests also apply to the eventual response? Any reason we shouldn't do that? Cheers, Yoav
Received on Friday, 31 August 2018 15:57:18 UTC