Bug 43351

Summary: Platform network layer should be able to differentiate WebFrameLoaderClient from WebCore platform clients
Product: WebKit Reporter: Balazs Kelemen <kbalazs>
Component: PlatformAssignee: Nobody <webkit-unassigned>
Status: RESOLVED DUPLICATE    
Severity: Normal CC: andersca, ap, aroben, darin, dglazkov, gustavo, mjs, sam, webkit.review.bot, xan.lopez
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: All   
Attachments:
Description Flags
proposed patch
none
proposed patch none

Balazs Kelemen
Reported 2010-08-02 08:31:24 PDT
In most of the ports the network backend is somewhat bounded to the platform specific API-s. For example Qt has a QWebPage::setNetworkAccessManager(QNetworkAccessManager*) API. The object that has set here is used in ResourceHandle. In the web process implementation of WebKit2, we have no platform specific FrameLoaderClient neither platform specific wrappers around WebPage, WebFrame, etc. so the backend should just do it's job without taking care of those API-s. However, it should be informed about that. My idea is to introduce a new virtual function in FrameLoaderClient (e.q. isPlatformClient).
Attachments
proposed patch (9.06 KB, patch)
2010-08-02 08:50 PDT, Balazs Kelemen 		no flags
DetailsFormatted DiffDiff
proposed patch (21.12 KB, patch)
2010-08-03 04:30 PDT, Balazs Kelemen 		no flags
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Balazs Kelemen
Comment 1 2010-08-02 08:50:08 PDT
Created attachment 63223 [details] proposed patch
WebKit Review Bot
Comment 2 2010-08-02 10:05:46 PDT
Attachment 63223 [details] did not build on chromium: Build output: http://queues.webkit.org/results/3639315
WebKit Review Bot
Comment 3 2010-08-02 12:13:01 PDT
Attachment 63223 [details] did not build on gtk: Build output: http://queues.webkit.org/results/3616557
WebKit Review Bot
Comment 4 2010-08-02 16:54:15 PDT
Attachment 63223 [details] did not build on win: Build output: http://queues.webkit.org/results/3621441
Balazs Kelemen
Comment 5 2010-08-03 04:30:08 PDT
Created attachment 63322 [details] proposed patch Taking care of the world aside from qt.
Balazs Kelemen
Comment 6 2010-08-07 13:49:35 PDT
Please, let me know your feelings about the idea of this patch. I know that this is somewhat of a layering violation, but I think that as long as the network backends are allowed to touch the FrameLoaderClient this is acceptable.
Darin Adler
Comment 7 2010-08-08 21:34:20 PDT
(In reply to comment #6) > Please, let me know your feelings about the idea of this patch. I know that this is somewhat of a layering violation, but I think that as long as the network backends are allowed to touch the FrameLoaderClient this is acceptable. Network back ends should not touch the FrameLoaderClient. Bug 42292 is our attempt to fix the layering of the networking code. Adding virtual functions to FrameLoaderClient is going the wrong direction on this. Is there some way we could fast track the work in bug 42292 instead of doing this?
Balazs Kelemen
Comment 8 2010-08-17 07:52:08 PDT
*** This bug has been marked as a duplicate of bug 42292 ***
Note You need to log in before you can comment on or make changes to this bug.