"Stephen R. van den Berg" <<srb[at]cuci.nl>> wrote:
> I'd say it is imperative to guarantee the same database session
> to the RXML programmer, /.../
It already is. As I said, there is never any thread switch inside an
rxml evaluation session. The rare cases when a request can switch
threads occur elsewhere; more precisely it might happen during async
PUTs (although I'm not quite sure it actually happens there either,
tbh).
> Wouldn't a longer living id object actually make it desirable to provide
> the same SQL sessions alongside with it?
Maybe in some cases, but it could also be a source of db connection
leaks. I wouldn't be comfortable if a client could make the server
hold a db connection for arbitrarily long by transmitting or receiving
data very slowly.
> I've cleaned up my old patches using ideas from your reuse-in-thread
> solution, and am currently debugging that. As soon as I have it working,
> I'll see what the cleanest API could be.
Ok, very good!
|