Re: PEAR DB abstraction layers

Lukas Smith wrote:

Hi, I have been pondering how things should proceed in terms of DB abstraction layers in PEAR. As some of you may know MDB was created to add all the nice Metabase features into PEAR under a PEAR DB like API. It was decided that the task would be so large that it would make sense to have an interim package for the entire development: so MDB was born. During the same time DB was maintained and even extended further. With the DB 1.6 series the package has stabelized greatly. Daniel has made some very valueable portability improvements etc. I refined the MDB API more and more which resulted in MDB2. Both MDB and MDB2 are faster than DB. They both provide a superset of the functionality of DB. However they both still lack in drivers and documentation. Also most add on packages are developed for DB. Especially since right now we still bundle DB, but probably also because DB is simply the more obvious name alot of people outside of PEAR think that DB is recommended over MDB[2]. Of course DB currently has a much greater installed userbase and alot of applications build around it and tutorials covering it. But the question really is what is our recommended DBAL? I think its time that MDBx is moved to the status of the recommended DBAL in PEAR. This means to me that MDB2 is actually turned into DB_v2 (DB2 has the danger of confusion due to IBM's DB2), as the original intention from the very beginning. This will not change any of the existing code out there nor will it deminish the fact that we will maintain DB further. But its an important step in the terms of making it clear where the future lies.

<snip> I'm very much +1 on this happening. A *lot* of new developers I see, are finding PEAR::DB and *not* PEAR::MDB[2]; this is simply because of the name. It is beneficial for new developers to be using MDB[2] because of the speed increase. I've yet to see somebody go back to DB after trying MDB[2] :) - Davey