From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?windows-1252?Q?Javier_Mu=F1oz?= Subject: Re: RGW S3 Website hosting, non-clean code for early review Date: Wed, 02 Dec 2015 15:02:12 +0100 Message-ID: <565EF9E4.2020501@igalia.com> References: <20150623093325.GH10754@orbis-terrarum.net> <1739382408.18844114.1435091419238.JavaMail.zimbra@redhat.com> <854662566.18884035.1435100864177.JavaMail.zimbra@redhat.com> <1563403567.1093721.1438280774967.JavaMail.zimbra@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 8bit Return-path: Received: from smtp3.mundo-r.com ([212.51.32.191]:33234 "EHLO smtp4.mundo-r.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1758371AbbLBOMq (ORCPT ); Wed, 2 Dec 2015 09:12:46 -0500 In-Reply-To: Sender: ceph-devel-owner@vger.kernel.org List-ID: To: "Robin H. Johnson" , Yehuda Sadeh-Weinraub Cc: ceph-devel@vger.kernel.org, Jonathan LaCour Hi, I would appreciate to know the current status of the implementation if possible. Any progress? Any 'deadline' to go upstream? :) Thanks in advance! Javier On 07/31/2015 10:21 AM, Robin H. Johnson wrote: > I'm busy travelling and wrapping other things up, but will resume major > work on this in August, probably week 2/3, depending on other > interrupts/priorities. > > It's almost complete, the only missing features are: > - error page fetcher gets confused on conditional requests (I think a > different approach is needed, esp re range & if-modified requests, > they still seem to leak into the custom req_info on the second request) > - cascaded errors on conditional requests need more handling (eg > 404 bounces to the error page that doesn't exist as well) > Some further work that isn't blocking is also: > - review handling of x-amz-website-redirect-location header when it > conflicts with RoutingRules > - Check what input validations Amazon does, because already they don't > match the docs > - eg docs say any http code for redirect, but it really only allows > 301-305, 307, 308 > - Exact validation on the URI/paths is also not immediately clear, I > had some unexpected results with URL-escaped inputs. > - Review cases in which AMZ sets the errors in the header as well as the > page, didn't seem consistent, esp in the double-error cases. > - more cleanup/refactor of test cases >