Instead use explicitly emptySessionID, which we can try to remove the use progressively.
Created attachment 376159 [details] Patch
<rdar://problem/54256381>
Comment on attachment 376159 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=376159&action=review > Source/WebCore/PAL/pal/SessionID.h:-35 > - SessionID() Maybe explicitly have = delete. > Source/WebCore/platform/network/CookieRequestHeaderFieldProxy.h:67 > + return { }; nullopt > Source/WebKit/Shared/WebPageCreationParameters.h:100 > + PAL::SessionID sessionID { PAL::SessionID::emptySessionID() }; I don't think this should be necessary. Just go to the places where one of these is created and put the value in the constructor. > Source/WebKit/UIProcess/API/APIPageConfiguration.cpp:52 > + : m_sessionID(PAL::SessionID::emptySessionID()) ditto.
(In reply to Alex Christensen from comment #3) > Comment on attachment 376159 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=376159&action=review > > > Source/WebCore/PAL/pal/SessionID.h:-35 > > - SessionID() > > Maybe explicitly have = delete. OK > > Source/WebCore/platform/network/CookieRequestHeaderFieldProxy.h:67 > > + return { }; > > nullopt > > > Source/WebKit/Shared/WebPageCreationParameters.h:100 > > + PAL::SessionID sessionID { PAL::SessionID::emptySessionID() }; > > I don't think this should be necessary. Just go to the places where one of > these is created and put the value in the constructor. This allows making the patch smaller. I plan to make a specific patch for this one. > > Source/WebKit/UIProcess/API/APIPageConfiguration.cpp:52 > > + : m_sessionID(PAL::SessionID::emptySessionID()) > > ditto. https://bugs.webkit.org/show_bug.cgi?id=200670 removes APIPageConfiguration::m_sessionID.
Created attachment 376252 [details] Patch for landing
This patch largely undoes my cleanup in bug #184868, which I think is unfortunate and weird given: 1. you reviewed it 2. You don't mention undoing it or any reasons in the changelog. Nothing to stop you from landing. Just wanted to point it out.
(In reply to Daniel Bates from comment #6) > This patch largely undoes my cleanup in bug #184868, which I think is > unfortunate and weird given: > > 1. you reviewed it > 2. You don't mention undoing it or any reasons in the changelog. With this patch, using decoder >> for sessionIDs is now needed. I updated the decode method to use >> for consistency. Agreed this is reverting part of bug 184868 patch but this should not change any behaviour. I believe there is a trend to use >> more and more these days. It is true we could decide on guidelines for decoders, whether to use >> or decode() by default for instance. I personally tend to use >> for new decode methods.
Comment on attachment 376252 [details] Patch for landing Clearing flags on attachment: 376252 Committed r248668: <https://trac.webkit.org/changeset/248668>
All reviewed patches have been landed. Closing bug.