implement OleLoadPictureFile

[Jeremy Drake]

On Sun, 16 Nov 2008, Juan Lang wrote:

> Do you have a bug open?  Sorry, I've forgotten.  If not, please do open
> one.  You can describe your findings there.

No, but a quick search on bugzilla for OleLoadPictureFile turned up bug
10156.  A comment on that bug suggests changing the summary to
"OleLoadPictureFile not implemented", which would exactly sum up my
situation.  I posted a comment to that bug to that affect.  Should that
bug be renamed, I could post my findings there...

> > If I provided a patch to dlls/oleaut32/tests/olepicture.c adding tests
> > that verify the behaviors of the functions, would that be accepted?  Or is
> > writing tests for functions you've seen the disassembly for also
> > prohibited?
>
> In general, writing tests is okay, as they are an exercise in
> black-box reverse engineering, which is what's allowed.  So yeah,
> tests would be great.

I've been looking at writing some tests (I've been compiling using make
crosstest and running the exe on windows 2000), and I've found that there
must be some sort of mapping between error codes from GetLastError when
accessing the file and the HRESULT values returned from OleLoadPicutreFile
(and not of the standard HRESULT_FROM_WIN32 variety that I'm familiar
with).  For instance, CTL_E_FILENOTFOUND is returned when the file does
not exist.  I'm trying to invoke as many error cases in the test as I can
to flesh out this mapping, but I'm not sure I can find all of them in this
manner...

I think at some point in the next couple of days I'll decide I've got as
many as I can and send in a patch to the tests.