Re: RE: XML_Tree2 vs XML_Tree 2.0.0

Mika Tuupola wrote:

On Sun, 30 May 2004, Stefan Neufeind wrote:

You intend to BC-break a stable package twice a year? Well, then it

Did you happen to read the bug description? Where did I talk about bc breaking a stable package twice an year? Besides I am not maintaining XML_Tree so I have no intentions on that behalf. Bug #89 is IMO a good example where fixing a bug (XML_Tree did not follow XML standard, and fixing the bug might cause problems to those scripts relying on the broken feature) would force changing a package from Foo_Barvx to Foo_Barvx+1. Which to me seems plain stupid.

This does not require a BC break - instead, have a flag that is by default set to do the old, incorrect behavior, and document heavily that you should set the flag. function getAttribute($name, $bc = true) {

    if (!$bc) {
        $name = strtolower($name);
    }
    if (isset($this->attributes[$name])) {
        return $this->attributes[$name];
    }
    return null;

} BC breaks are not required to fix 99.9% of bugs. BC breaks are in themselves a major bug, much more serious than most bugs considered to be major - you can have an entire application stop working, and have no idea why, because somebody upgraded a shared copy of package X. A little ingenuity removes the need for BC breaks even with major problems like bug 98 Greg