[Webkit-unassigned] [Bug 41630] DNS prefetching spams resolver, shoots self in the foot

Wed Mar 14 06:41:23 PDT 2012

https://bugs.webkit.org/show_bug.cgi?id=41630

--- Comment #12 from Dan Winship <danw at gnome.org>  2012-03-14 06:41:23 PST ---
(In reply to comment #10)
> That'd be the libsoup patch. This adds some new API and the GNOME 3.4 API freeze is already gone. If having this patch is interesting enough for 3.4 we'd have to ask for an exception.

API Freeze only applies to the "Platform" components (glib, gtk, etc). Although it's not encouraged, we can change libsoup's API right up until hard code freeze. (Which is Monday.) And it's clear that the libsoup patch is pretty safe, so I don't mind adding it. (The WebKit patch is bigger, but that's someone else's problem :)

On the patch itself, since we've given up on doing anything more clever than just DNS, I'd make it "soup_session_prefetch_dns", with a hostname, instead of "prefetch_uri" (which then also saves you having to fake up a URI in DNSResolveQueue::resolve()).

Also, add this bug URL to the commit message.

(In reply to comment #11)
> WK Patch

drop the proxyIsEnabledForDefaultSession() stuff; it's not really reliably in either direction; just because a proxy resolver is in use doesn't mean that there's a proxy configured. And OTOH if you're using SOCKS4, you have to do DNS on the client anyway (because the protocol only uses IP addresses).

We can add smarts to SoupSession/SoupProxyURIResolver/GProxyResolver later if we want to implement this check correctly.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.