In article <mailman.3067.1322361338.27778.python-l...@python.org>, Dave Angel <d...@davea.name> wrote:
> If you're using Python, you already have a "fast hash" library, in the > dictionary class. And yes, if a problem doesn't need the full > generality of a database, you may be able to implement it with > dictionaries, and it may even be practical to store those dictionaries > to disk for later retrieval. However, there are quite a few reasons > this may not be good enough. To start with just two: if there are > multiple users of the database, and they have to be synched. Or if you > have to be safe from a program or a system crashing. This is a good point. In general, databases differ from in-memory data structures in that they provide: 1) Persistence 2) Data integrity 3) Shared access Different kinds of databases provide different amounts and flavors of these (especially #2). I think it's fair to say, however, that those three in some form are essential for something that calls itself a database. Where thing get fun is when you start looking at the various things out there that call themselves databases and need to evaluate which ones give you the combination of these that best match your needs. -- http://mail.python.org/mailman/listinfo/python-list
