[Webkit-unassigned] [Bug 176755] [WPE][GTK] Run tests with G_DEBUG=fatal-criticals

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=176755

Michael Catanzaro <mcatanzaro at igalia.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mcatanzaro at igalia.com
            Summary|[WPE][GTK] Run tests with   |[WPE][GTK] Run tests with
                   |G_DEBUG=fatal-warnings      |G_DEBUG=fatal-criticals

--- Comment #1 from Michael Catanzaro <mcatanzaro at igalia.com> ---
Actually I think we can't be this strict for API tests, looks like they're testing stuff that actually needs to emit warnings. E.g.:

    // There can't be more than one context with automation enabled
    GRefPtr<WebKitWebContext> otherContext = adoptGRef(webkit_web_context_new());
    test->removeLogFatalFlag(G_LOG_LEVEL_WARNING);
    webkit_web_context_set_automation_allowed(otherContext.get(), TRUE);
    test->addLogFatalFlag(G_LOG_LEVEL_WARNING);
    g_assert(!webkit_web_context_is_automation_allowed(otherContext.get()));

But API tests should still always crash on criticals. No exceptions to that. Let's use fatal-criticals for layout tests to be consistent. I'm not aware of any warnings being printed in layout tests, and the important part of this change is to get backtraces to criticals.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170912/d1fa5d9a/attachment.html>