Automating analysis of behaioural differences between Wine and Windows?

[Stefan Dösinger]

Hi Daniel,

My sense is that this would be legally still OK, even though it gets
very close into that "not sure" grey area. Conceptually how is this tool
different from e.g. http://www.rohitab.com/apimonitor or Wine +relay
logs? It seems that making API monitor work on Wine (if it doesn't
already) and comparing its output would get you the same things.

It does become a problem if you observe that winfoo.dll calls winbar.dll
and replicate this behaviour on Wine. That's essentially the same as
observing a native Windows library in Wine with Wine's logging tools.

I don't think comparing application runs on a large scale will be useful
though. It may be useful to answer well-defined questions like "This
game is passing incorrect parameters to IDirect3DVertexBuffer9::Lock.
Does it to the same on Windows, or are we giving the game bad input?".
Applications respond to differences in the environment (language,
windows version, GPU, available hardware), so a behavioural difference
is often not a problem.

Stefan

Am 2018-01-26 um 23:47 schrieb Daniel Santos:
> Some time back I had investigated a concept for a tool to aid in
> discovering behavioral differences between Wine and Windows.  I have
> most of the technological issues figured out -- it would be a bit of an
> undertaking, but can be developed iteratively so that it could begin to
> pay for its self fairly soon.  However, I need advise on rather or not
> it would create tort.
> 
> Put simply, it would be a layer built into Wine that records win32/64
> API calls and their results and another layer on Windows that does the
> same.  The two results can then be analyzed to discover behavioral
> differences.  Of course, this is a heavily over-simplified explanation
> as there are many gory details of how to make this work, perform
> acceptably, produce useful information, remove noise, etc., but could
> this produce a legal problem?  I would hope to be able to use such a
> tool on a typical piece of commercial software with a typical EULA.  To
> me, it doesn't seem differ much from what antivirus software does.
> 
> Thanks,
> Daniel
> 
>