I've tried to reproduce it in WebKit using a unit test (Source/WebKit2/UIProcess/API/gtk/tests/TestWebKitWebContext.cpp) without success so far.

In my test I was creating 2 WebKitWebViews and use webkit_web_view_load_uri to load "foo:blank" in both views. Then I connected to the "load-changed" signal and both views finish to load properly.

I've tried to reproduce it in the ephy-web-view-test.c without success either, so maybe it's a specific bug in sessions code, not sure though.

The code example that works for me is the following:

static guint web_view_ready_counter = 0;

static void
load_changed_same_about_uri_cb (WebKitWebView *web_view,
#ifdef HAVE_WEBKIT2
                 WebKitLoadEvent status,
#else
                 GParamSpec *pspec,
#endif
                 GMainLoop *loop)
{
#ifndef HAVE_WEBKIT2
  WebKitLoadStatus status = webkit_web_view_get_load_status (web_view);
#endif

  if (status == WEBKIT_LOAD_FINISHED) {
    web_view_ready_counter--;
    g_signal_handlers_disconnect_by_func (web_view, load_changed_cb, loop);

    if (web_view_ready_counter == 0)
      g_main_loop_quit (loop);
  }

}

static void
wait_until_load_is_finished (WebKitWebView *web_view, GMainLoop *loop)
{
#ifdef HAVE_WEBKIT2
  g_signal_connect (web_view, "load-changed", G_CALLBACK (load_changed_same_about_uri_cb), loop);
#else
  g_signal_connect (web_view, "notify::load-status", G_CALLBACK (load_changed_same_about_uri_cb), loop);
#endif
}

static void
test_ephy_web_view_load_two_views_same_about_uri (void)
{
  EphyWebView *view1, *view2;
  GMainLoop *loop;
  const char *uri;

  view1 = EPHY_WEB_VIEW (ephy_web_view_new ());
  view2 = EPHY_WEB_VIEW (ephy_web_view_new ());
  loop = g_main_loop_new (NULL, FALSE);
  uri = "about:epiphany";

  web_view_ready_counter = 2;

  wait_until_load_is_finished (WEBKIT_WEB_VIEW (view1), loop);
  wait_until_load_is_finished (WEBKIT_WEB_VIEW (view2), loop);

  ephy_web_view_load_url (view1, uri);
  ephy_web_view_load_url (view2, uri);

  if (web_view_ready_counter != 0)
    g_main_loop_run (loop);

  g_main_loop_unref (loop);
  g_object_unref (g_object_ref_sink (view1));
  g_object_unref (g_object_ref_sink (view2));
}

This issue could be related to bug #695733 (I'm not sure yet). So, once WK2 bug is fixed (https://bugs.webkit.org/show_bug.cgi?id=112418) we should test this again.

With the current patch for the WebKit bug (http://webkit.org/b/112418) the issue is still present.

Anyway, it doesn't seem an Epiphany issue as I've been able to reproduce it with MiniBrowser and a unit test in WebKit, so I've reported a different bug in WebKit with the information: http://webkit.org/b/112557

Created attachment 239225 [details] [review]
Patch

Even when this is a WebKit issue, once the WK bug (http://webkit.org/b/112557) is solved we should apply this patch to re-enable some session tests. That's the reason why I've reopened it.

Comment on attachment 239225 [details] [review]
Patch

Removing from unreviewed patches list, since this is blocked on a WebKit bug

I can't reproduce this anymore, even with restore-session-delaying-loads disabled.

-- GitLab Migration Automatic Message --

This bug has been migrated to GNOME's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.gnome.org/GNOME/epiphany/issues/190.