[python-win32] How to create a com_record?

[Kevin Patterson]

Ok, more digging, it looks like the OLE Automation TypeLib should have a
GUID Record defined. I see this when using the Python Object Browser

However, when I generate the py file for the OLE Automation type library, I
don't see GUID in the RecordMap (or any reference to the GUID Record)...Is
there suppose to be something in the generated file, but maybe its not
working ? How do I actually create one of these Records ?

On Tue, Mar 11, 2008 at 10:04 PM, Kevin Patterson <patter001 at gmail.com>
wrote:

> I found the comtypes equivalent if that helps any. The generated comtypes
> file looks like this:
>
>    COMMETHOD([dispid(1610743812), helpstring(u'GetInterface() method')],
> HRESULT, 'GetInterface',
>               ( ['in'], POINTER(GUID), 'riid' ),
>               ( ['retval', 'out'], POINTER(POINTER(IUnknown)), 'ppUnknown'
> )),
>
> With the above, I'm able to use the comtypes.GUID.GUID structure to pass
> the information over through the COM. Unfortunately I still can't find the
> win32com equivalent of passing my structure through the COM.
>
> The win32com generated py file is saying that it expects a VT_RECORD, but
> from above, maybe it should really be expecting a POINTER ? or VT_POINTER ?
> I tried to modify the win32com generated file to take a VT_PTR but that
> didn't seem to help. I really don't want to redo my infastructure in
> comtypes, so if anyone knows the win32com equivalent, that would be awesome.
>
> In the meantime, I'll keep digging, and will post if I find more.
>
>
>
> On Tue, Mar 11, 2008 at 12:58 PM, Kevin Patterson <patter001 at gmail.com>
> wrote:
>
> > Old thread, but I got a little more data on the function. The function
> > actually has a REFIID as the first parameter. This apparently is a standard
> > structure for C COM code.  I'm not entirely sure why the win32com changes
> > the REFIID to be a VT_RECORD. Does the following definition of the functions
> > help at all (they are from the IDL file)
> >
> > HRESULT GetInterface( [in] REFIID riid, [out, retval, iid_is(riid)]
> > IUnknown** ppUnknown);
> > should be same as:
> > HRESULT GetInterface( [in] GUID* pGuid, [out, retval, iid_is(pGuid)]
> > IUnknown** ppUnknown);
> >
> >
> >
> > On Mon, Feb 4, 2008 at 6:44 PM, Mark Hammond <mhammond at skippinet.com.au>
> > wrote:
> >
> > > > guid = win32com.client.Record("GUID", some_ob)
> > >
> > > > I don't have the some_obj that allows me to build "GUID".
> > >
> > > It should be any object from the same typelib - almost certainly
> > > whatever
> > > object you intend passing this record to.
> > >
> > > > VB.NET (GetType) and C++ (__uuid) both already know that a GUID is a
> > > record.
> > >
> > > C++ doesn't know it is a COM record.  A COM record is something quite
> > > different to a C/C++ struct or from .NET types.  A COM record is
> > > specifically designed to be cross-language.
> > >
> > > > It seems the python CLSID is the same thing, but i don't know what
> > > to call
> > >
> > > > to get a CLSID that is a com_record type.
> > >
> > > As I've explained, we are looking for the COM record definition you
> > > are
> > > trying to use.  A C++ struct definition for a GUID is *not* what you
> > > are
> > > looking for - you are looking for the COM record definition of a GUID.
> > >  I
> > > see now that you simply copied me a C++ struct you found, which is not
> > > at
> > > all useful in this context.
> > >
> > > To clarify: in a *typelib* you will find a struct definition that you
> > > are
> > > trying to use.  Finding a struct definition of the same name somewhere
> > > else
> > > is not suitable.
> > >
> > > > Am I understanding the win32com.client.Record call correctly? It
> > > needs
> > > GUID to be defined by the
> > > > COM right?
> > >
> > > Not "defined by COM", but "defined using COM" - the author of a
> > > typelib must
> > > have defined it somewhere.
> > >
> > > > Its odd b/c the GIUD (python's CLSID) is all over the COM
> > > definitions its
> > > just type
> > > > VT_RECORD type that I'm not understanding how to convert to.
> > >
> > > Yes, I'm afraid you are confused.  All existing GUID or CLSID
> > > definitions
> > > you can find are useless - there is only 1 that is relevant, and that
> > > will
> > > be defined in a typelib.
> > >
> > > I think the confusion is such that you need to find basic docs for the
> > > object, so we can get back to square 1.
> > >
> > > Mark
> > >
> > >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/python-win32/attachments/20080313/3a571f85/attachment.htm