[Webkit-unassigned] [Bug 126384] [SOUP] WebSockets must use system proxy settings

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Aug 10 04:14:10 PDT 2017


https://bugs.webkit.org/show_bug.cgi?id=126384

--- Comment #15 from Carlos Garcia Campos <cgarcia at igalia.com> ---
(In reply to Dan Winship from comment #14)
> (In reply to Carlos Alberto Lopez Perez from comment #13)
> > (In reply to Michael Catanzaro from comment #12)
> > > I've also considered if maybe we should be using libsoup's implementation of
> > > WebSockets rather than WebKit's implementation. I always assumed that
> > > WebKit's implementation was likely to have fewer bugs. Silly me.
> 
> (Of course, this bug predates libsoup's WebSockets implementation.)

Yes, that's why we don't use the libsoup one, not because it's more or less buggy, we always prefer to use libsoup (well, except in the case of the disk cache, because the webkit implementation is much better and shared with all other ports).

> > Do you know something about the quality of the libsoup WebSocket
> > implementation and/or about any project that uses it? 
> > Keep in mind _any_ implementation is going to have bugs.
> 
> The actual WebSockets implementation isn't all that complicated. The code
> for letting SoupSession make a (possibly-proxied) connection to a server,
> and then stealing that GIOStream from SoupSession to use as a websockets
> connection is kind of tricky, but of course, that's exactly the code that
> the patch here makes use of, so you'd be using it either way. (Well, the
> patch here is written against Dirkjan's original proposal that isn't quite
> how things eventually got implemented, but it would be similar.)
> 
> Anyway, a handful of people are using libsoup's websockets code now, (and it
> was originally based on cockpit's code that had been in use for a few years
> before that). And I assume WebKit has some WebSockets tests anyway that
> would validate that it was good enough for WebKit purposes?

WebSockets were added to libsoup in 2.50, IIRC, and we depend on 2.40, so we still need to fix our implementation. Then we can use libsoup, but we will have to use ifdefs.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170810/b585b742/attachment.html>


More information about the webkit-unassigned mailing list