Damien Tournoud <damz <at> damz.org> writes: > > On Sun, Feb 27, 2011 at 12:50 PM, Jordi Boggiano <j.boggiano <at> seld.be> wrote: > The algorithm seems flawed to me, at least in its reference implementation. > There does not seem to be any guarantee that the returned position is really > the *first* occurrence of the needle in the haystack.
This is a bug. This algorithm is still something very new. > The implementation could be fixed (by adding an explicit string matching > when building the hash table, and by storing *all* the occurrences of a > given W in SS), but that will increase the overall cost (both computing and > memory) of the algorithm. Bug was fixed by reversing hash fill order (without increase of overall cost). @ All About the licence. I am not a layer, I need to figure out the licensing. I intend to give PHP project a special license, I just need to clarify several things. -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
