Thomas G. Lockhart wrote:
> It turns out that the backend reports "reaping dead processes" without
> showing any successful transactions other than (perhaps) a successful
> plain connection. So, it appears that the two or three queries embedded
> into SQLConnect()/CC_connect() are not actually succeeding, or even
> getting to the backend properly.
>
I am not sure how fresh your source is, but I do recall Byron making some
tweaks in the initial connection queries recently. We'll wait and see.
> btw, I've only just started looking at the fe/be code, but istm that it
> would be really nice if libpq-fe could fully support the WIN32 interface
> or if psqlodbc could use libpq-fe when on a Unix box. Some guy in Russia
> has a driver which uses libpq-fe and the code looks like it has the same
> lineage as Byron's code (before Byron started working on it). Where does
> this code come from? Byron's code gives credit to Christian Czezatke and
> Dan McGuirk but the "iodbc project" driver doesn't.
>
Don't know about the Russian dude, but the we completely rewrote the the
front-end section in the driver. We did not (and still do not) like the
way libpq consumes the entire query result into front-end memory before
returning control to the application. This make large queries slow and
sometimes impossible. However, with the patches that Bruce and Vadim are
working on for multiple ORs (many thanks) this may become a moot issue.