It's useful to distinguish history items for failures such as "domain not found" from history items that represent successful navigation and loading of a website.
Created attachment 25197 [details] patch
Comment on attachment 25197 [details] patch > +#include "CachedPage.h" > +#include "CString.h" Should headers #includes be sorted alphabetically or in ASCII order? I thought the latter. It seems a little dangerous that the presence of the lastVisitWasFailure key in the dictionary representation always means "true". r=me
(In reply to comment #2) > It seems a little dangerous that the presence of the lastVisitWasFailure key in > the dictionary representation always means "true". I could change the rule to be that it has to be exactly "boolean true". Or that it can be anything except "boolean false". Or that any numeric thing that's non-zero is counted as true. Do you have a preference?
(In reply to comment #2) > > +#include "CachedPage.h" > > +#include "CString.h" > > Should headers #includes be sorted alphabetically or in ASCII order? I thought > the latter. I used the Xcode sort, which I thought was the same as the command line sort. I'll redo it. The ASCII order you get from the sort tool is our standard I think.
http://trac.webkit.org/changeset/38449
This only worked for back/forward history, not global history. Oops!
Created attachment 25446 [details] patch
Comment on attachment 25446 [details] patch Can you assert that the url parameter to updateGlobalHistory is equal to the documentLoader's urlForHistory()? > + [[WebHistory optionalSharedHistory] _visitedURL:cocoaURL withTitle:title wasFailure:wasFailure]; I think you can get rid of cocoaURL and pass url directly. r=me
(In reply to comment #8) > Can you assert that the url parameter to updateGlobalHistory is equal to the > documentLoader's urlForHistory()? I'd much rather remove the url parameter altogether. In a future patch. You think that's OK? > > + [[WebHistory optionalSharedHistory] _visitedURL:cocoaURL withTitle:title wasFailure:wasFailure]; > > I think you can get rid of cocoaURL and pass url directly. I'll try it and land it that way if it compiles.
(In reply to comment #9) > (In reply to comment #8) > > Can you assert that the url parameter to updateGlobalHistory is equal to the > > documentLoader's urlForHistory()? > > I'd much rather remove the url parameter altogether. In a future patch. You > think that's OK? Yes.
http://trac.webkit.org/changeset/38728