I'm not sure that's the case. Apparently a few observers need to be synchronously notified when starting a browser-initiated navigations. That's only true with DidStartNavigation when there is no BeforeUnload event (since DidStartNavigation happens after BeforeUnload). We should first investigate whether a synchronous notification of browser navigation start is still needed and still provide one if that's the case - though it can be renamed.

If that's true, we should definitely change the WCO comment on the callback, which clearly says it's obsolete with Plz.

Yes. We thought it was obsolete, the changed when DidStartNavigation happens, and found that some caller need synchronous notification.

I ran a CL with DidStartNavigationToPendingEntry's removal in https://crrev.com/c/939749 , though I didn't deeply investigate the failures that arose. Here's the summary.

* android_webview/browser/icon_helper.h/cc. It's used here to clear a flag when there's a reload BYPASSING_CACHE. Probably convertible to DidStartNavigation.
* chrome/browser/ui/browser_instant_controller_unittest.cc A unittest. Needs investigation.
* chrome/browser/ui/javascript_dialogs/javascript_dialog_tab_helper.cc Asks for the override to be removed when PlzNavigate is turned on. Easy.
* chrome/browser/ui/sync/one_click_signin_sync_observer_unittest.cc Asks for the override to be removed when PlzNavigate is turned on. Easy.
* content/browser/frame_host/navigation_controller_impl_unittest.cc A unittest. Needs investigation.
* content/browser/web_contents/aura/overscroll_navigation_overlay_unittest.cc A unittest. Needs investigation.

That's it. One place that can probably easily be converted, two places that request removal when PlzNavigate is turned on, and three unit tests. From my experience, I distrust unit test failures in navigation code work.

Perhaps we still need DidStartNavigationToPendingEntry, but those existing users don't (yet) seem convincing. We should look.