Re: Minor fix for include/bits1_5.idl

[Gerald Pfeifer]

Strike that, I must have misread the documentation.

Only thing I am wondering is do we really need the (unsigned long) here?

If anyone has a pointer to clear documentation that would be nice; what I
found so far leaves some questions open...

Gerald

On Sun, 28 Jun 2009, Gerald Pfeifer wrote:
> The generated file include/bits1_5.h is the same before and after this
> patch, yet somehow the original form strikes me as a bit odd at best...
>
> Gerald
>
> ChangeLog:
> Fix prototype for GetReplyData().
>
> diff --git a/include/bits1_5.idl b/include/bits1_5.idl
> index 274a7de..29520e2 100644
> --- a/include/bits1_5.idl
> +++ b/include/bits1_5.idl
> @@ -39,7 +39,7 @@ interface IBackgroundCopyJob2 : IBackgroundCopyJob
>      } BG_JOB_REPLY_PROGRESS;
>  HRESULT GetReplyProgress([in, out] BG_JOB_REPLY_PROGRESS *progress);
> - HRESULT GetReplyData([out, size_is( , (unsigned long) *pLength)] byte **pBuffer,
> - HRESULT GetReplyData([out, size_is((unsigned long) *pLength)] byte **pBuffer,
>                      [in, out, unique] UINT64 *pLength);
> HRESULT SetReplyFileName([unique] LPCWSTR filename);
> HRESULT GetReplyFileName([out] LPWSTR *pFilename);

Minor fix for include/bits1_5.idl

[Rob Shearman]

2009/6/28 Gerald Pfeifer :
> Strike that, I must have misread the documentation.
>
> Only thing I am wondering is do we really need the (unsigned long) here?

Yes, as conformance/variance is restricted by NDR to 32-bit
quantities. The IDL compiler should warn/error without the cast as it
would be silently truncating the value used for computing the memory
allocated for storing the array.
> If anyone has a pointer to clear documentation that would be nice; what I
> found so far leaves some questions open...

The DCE/RPC specification would be a good start, but there is probably
some MSDN documentation on the limitations of size_is expressions in
IDL.