Re: [PATCH 2/2] ws2_32/tests: test a WSARecv error condition

[Paul Vriens]

On 03/22/2010 07:56 PM, Mike Kaplinskiy wrote:
> -
> - bret = GetOverlappedResult((HANDLE)dest,&ov,&bytesReturned, FALSE);
> - todo_wine ok(!bret&&  GetLastError() == ERROR_NETNAME_DELETED, "Did not get disconnect event: %d, error %d\n", bret, GetLastError());

Hi Mike,

This one fails on Win98 and WinMe with what appears just a different 
last error:

http://test.winehq.org/data/tests/ws2_32:sock.html

Could you have a look?

Maybe it also makes sense to add a SetLastError(0xdeadbeef) before the 
actual call as we are testing the last error afterwards. That way we are 
sure that the last error is indeed set to the expected.

[PATCH 2/2] ws2_32/tests: test a WSARecv error condition

[Mike Kaplinskiy]

On Tue, Mar 23, 2010 at 3:56 PM, Paul Vriens  wrote:
> On 03/22/2010 07:56 PM, Mike Kaplinskiy wrote:
> > -
> > - bret = GetOverlappedResult((HANDLE)dest,&ov,&bytesReturned, FALSE);
> > - todo_wine ok(!bret&&  GetLastError() == ERROR_NETNAME_DELETED, "Did
> >
> > not get disconnect event: %d, error %d\n", bret, GetLastError());
>
> Hi Mike,
>
> This one fails on Win98 and WinMe with what appears just a different last
> error:
>
> http://test.winehq.org/data/tests/ws2_32:sock.html
>
> Could you have a look?
>
> Maybe it also makes sense to add a SetLastError(0xdeadbeef) before the
> actual call as we are testing the last error afterwards. That way we are
> sure that the last error is indeed set to the expected.
>
> --
> Cheers,
>
> Paul.

Hey Paul,

I sent a fix for the failures, turns out win9x overlapped I/O is
broken in that case (it wakes up but says not done...). Setting the
last error would be more correct, but we check GetOverlappedResult
failure codes all over the place this way and it's probably better to
stay consistent with the rest of them.

Thanks for the heads up,
Mike.