There are many and varied search engines (SEs) on the Internet but it is still hard to locate and concentrate only on materials relevant to a specific task. Digital libraries (DLs) could better provide such services on the Web. There is a real need to formulate a methodology for the understanding of both these types of Web data repositories. We classify both SEs and DLs using similar criteria on both a functional scale and a generational time-line. In particular, we introduce the idea of the third generation Harvested DL and its resulting DL harvesting model. By comparing them and analyzing their characteristics, we discover that they actually share much in common. Finally, we note the expected incorporation of intelligent techniques and knowledge management in fourth generation SEs and DLs and the expected convergence of their interfaces and structures in the fifth generation - the Mega-Portal.