Hi, I haven't thought hard about it, but that sounds reasonable. Whenever possible, I try to avoid absolute (VS relative) URLs, to avoid this kind of problems. So removing the scheme when possible seems like something along the same lines of reasoning. Can you please open an issue (a copy/paste of your e-mail works)? And a CL that references the issue, if that tickles your fancy.
Thanks, Mathieu On 24 April 2017 at 11:08, Attila Tajti <[email protected]> wrote: > I've set up camlistore+publisher running behind a reverse proxy that handles > HTTPS therefore camlistore and publisher uses HTTP. > > Issue 818 [1] and CL 6646 [2] mentioned this issue and I could make > camlistore+publisher work together with the following snipplet in > server-config.json: > > { > "baseURL": "https://external.address";, > "listen": ":3179", > ... > "publish": { > "/pics/": { > "listen": ":3155", > "backendURL": "http://localhost:3155/";, > "apiHost": "http://localhost:3179/";, > "camliRoot": "publisher-pics", > "cacheRoot": > "/volume1/camlistore/home/var/camlistore/publisher-pics-cache", > "goTemplate": "gallery.html" > } > } > } > > Publisher works fine when using http://diskstation:3179/pics/foo (internal > address), but fails in the browser because of mixed content when being > accessed through the reverse proxy over https://external.address/pics/foo. > > The problem is that > 1. publisher serves the HTML with the JS-global variable "scheme" set to the > protocol it uses to access the parent camlistore process, and > 2. the publisher gopherjs/js client in the browser uses this variable to > fetch page members (instead of the scheme the publisher page was accessed > with) > > Fixing either of the above problems would fix my issue, but wouldn't it be > reasonable to remove scheme from the page header, and just use > window.location.protocol instead? Doing that seemed to fix the issue on my > end. The browser UI of publisher should not know what protocol the server > uses internally to talk to its parent, and browser links will only work with > the correct protocol after all. > > [1] https://github.com/camlistore/camlistore/issues/818 > [2] https://camlistore-review.googlesource.com/c/6646/ > > -- > You received this message because you are subscribed to the Google Groups > "Camlistore" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Camlistore" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
