OWB 2.0.17 startup performance regression

[Vicente Rossello]

Hi,

I've seen a startup performance regression since OWB 2.0.17 and latest
snapshot. Our boot times have increased from 10 to about 14 seconds (only
OWB side). I can see that it always try to load the same package-info's in:

while (true)
{
    try // not always existing but enables to go further when
getPackage is not available (graal)
    {
        pckge = classLoader.loadClass(previousPackage +
                (previousPackage.isEmpty() ? "" :".") +
"package-info").getPackage();
        break;
    }
    catch (Exception e)
    {
        if (previousPackage.isEmpty())
        {
            pckge = null;
            break;
        }
        packageVetoCache.put(previousPackage, false);
        idx = previousPackage.lastIndexOf('.');
        if (idx > 0)
        {
            previousPackage = previousPackage.substring(0, idx);
        }
        else
        {
            previousPackage = "";
        }
    }
}


I think that, in this loop, it should take into account the
packageVetoCache (whether it's true or false). Is it correct? Do you
want a PR with this correction?


Best regards,

Vicente.