_dbus_get_oom_wait magic

[Havoc Pennington]

On Wed, 2005-08-03 at 18:06 +0200, Timo Hoenig wrote:
> 
> My previous analysis of the problem was wrong, sorry.  The actual delay
> of the test runs where due to a invocation of  _dbus_wait_for_memory ()
> in bus_connection_disconnected ().
> 
> The attached patch fixes the problem. Please review.

Good catch - "make check" runs about a zillion times faster now ;-)

I put in a similar patch (just shuffled around so the transaction_new
was in only one place)

Havoc