On Aug 19, 2007, at 10:13, Greg Sabino Mullane wrote:
* Removes a major dependency, allowing the module to be built
anywhere.
* Faster and cleaner code, as we can create our own paths and
functions.
For example, we won't need to have all the protocol 2 branching in the
libpq code.
* (primary reason). We can make up for things missing in libpq, and
not have to wait for libpq to adopt them, such as being able to
download data in chunks and not have to put everything into memory.
I guess I just assumed that libpq was mature and robust, and that
reinventing that stuff ourselves would be a lot of work. Perhaps I'm
wrong?
Best,
David
