On Mon, Jun 15, 2009 at 8:22 AM, James B. Byrne<byrnejb@harte-lyne.ca> wrote:
> On: 15 Jun 2009 12:08:22 GMT, Jasen Betts <jasen@xnet.co.nz>
>>
>> On Sun, June 14, 2009 15:45, Scott Marlowe wrote:
>>>
>>> Is there a reason you're not using the native windows postgresql
>>> packages?
>>>
>>
>> Because for some reason, processes running in the cygwin environment
>> could not create databases in the postgresql instance running in
>> windows.
>
> that's odd, were you specifying host=localhost, if not it may be
> that the cygwin processes were trying to use local sockets.
> (oner difference of the native windows postgres to linux is no local
> sockets)
>
> The cygwin processes had no trouble connecting to and operating on
> existing databases. They could even drop a database. They just
> could not recreate one after dropping it. As the database in
> question is the one used for integration testing and as it gets
> dropped and recreated on the fly at various points in the test suite
> this was a major annoyance.
>
> It could very well be that this behaviour is caused by some other
> issue but I have a nagging suspicion that NTFS security issues are
> at the root of it. As I have no other need to access pgsql on my
> laptop, I thought it best to bring everything under one roof, so to
> speak, and begin checking things out from that initial point.
So, for sure something like:
alter user dbuser with createdb;
didn't fix the problem?