Re: crypt32: CryptUnprotectData: test zero-length payload

[Paul Vriens-2]

On 10/28/2009 02:40 AM, Dan Kegel wrote:

> Passes on Vista and on Wine (with one todo_wine).
> Exposes http://bugs.winehq.org/show_bug.cgi?id=20495
>
> (Note: 'make crosstest' doesn't work stunningly well in
> crypt32 for me -- there were lots of undefined symbols --
> but fortunately the file this test is in compiles ok standalone.
> Had to use LoadLibrary rather than GetModuleHandle,
> else it failed to run standalone.)
>
>

Hi Dan,

Some testfailures on everything below W2K. Some tests were already
skipped because of this.

I tried several things but the last error is always set to
RPC_X_WRONG_STUB_VERSION. Maybe we should test for that before we do any
tests and bail out? Does this error make sense on 9x/Me/NT4?

--
Cheers,

Paul.

[Dan Kegel-2]

On Thu, Oct 29, 2009 at 1:06 AM, Paul Vriens <paul.vriens.wine@...> wrote:
> Some testfailures on everything below W2K. Some tests were already skipped
> because of this.
>
> I tried several things but the last error is always set to
> RPC_X_WRONG_STUB_VERSION. Maybe we should test for that before we do any
> tests and bail out? Does this error make sense on 9x/Me/NT4?

http://www.codeguru.com/forum/archive/index.php/t-326734.html
may have a clue, it mentions that midl has a /robust option that
produces this error if used with older windows?
http://msdn.microsoft.com/en-us/library/aa367363(VS.85).aspx
- Dan

[Dan Kegel-2]

On Thu, Oct 29, 2009 at 7:05 AM, Dan Kegel <dank@...> wrote:

> On Thu, Oct 29, 2009 at 1:06 AM, Paul Vriens <paul.vriens.wine@...> wrote:
>> Some testfailures on everything below W2K. Some tests were already skipped
>> because of this.
>>
>> I tried several things but the last error is always set to
>> RPC_X_WRONG_STUB_VERSION. Maybe we should test for that before we do any
>> tests and bail out? Does this error make sense on 9x/Me/NT4?
>
> http://www.codeguru.com/forum/archive/index.php/t-326734.html
> may have a clue, it mentions that midl has a /robust option that
> produces this error if used with older windows?
> http://msdn.microsoft.com/en-us/library/aa367363(VS.85).aspx

Oh, and the other part of the clue: I hear that on Windows,
this function's guts are implemented with COM.  So perhaps
Windows is tripping over its shoelaces internally on this?