Duplicate UUID in oleaut32:tmarshal test

[Thomas Faber]

I just tried running midl on dlls/oleaut32/tests/tmarshal.idl and it
complained:
tmarshal.idl(83) : error MIDL2270 : duplicate UUID. Same as : ItestDual [ Interface 'ITestSecondDisp'  ]

This doesn't seem like it's intentional. But just giving
ITestSecondDisp a new uuid causes the following test failures on
Windows:
tmarshal.c:1861: Test failed: external_connections = 2
tmarshal.c:1864: Test failed: external_connections = 2
tmarshal.c:192: Test failed: fLastReleaseCloses = 1

I could try to figure this out, but it's probably better if someone
familiar with the subject could have a look instead.

(also, assuming this is a bug in the test, we might want to add that
check to widl, it seems pretty useful ;])

Thanks,
Thomas

[Jacek Caban]

On 06/30/13 23:29, Thomas Faber wrote:
> I just tried running midl on dlls/oleaut32/tests/tmarshal.idl and it
> complained:
> tmarshal.idl(83) : error MIDL2270 : duplicate UUID. Same as : ItestDual [ Interface 'ITestSecondDisp'  ]
>
> This doesn't seem like it's intentional. But just giving
> ITestSecondDisp a new uuid causes the following test failures on
> Windows:
> tmarshal.c:1861: Test failed: external_connections = 2
> tmarshal.c:1864: Test failed: external_connections = 2
> tmarshal.c:192: Test failed: fLastReleaseCloses = 1
>
> I could try to figure this out, but it's probably better if someone
> familiar with the subject could have a look instead.
>
> (also, assuming this is a bug in the test, we might want to add that
> check to widl, it seems pretty useful ;])

Thanks for the report. This was indeed a stupid mistake and after fixing
it, the behaviour is more logical. I will send a patch fixing tests.
Fixing todo_wine may wait until after code freeze. Current Wine
behaviour shouldn't cause problems.

Thanks,
Jacek