Automating analysis of behaioural differences between Wine and Windows?

[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

[George Gonzalez]

When I asked about something similar, whether it was okay to stopwatch 
how long it takes the Windows default windowproc to process a certain 
message, in order to determine whether that message got handled at all, 
I got some dubious looks, so I didn't pursue that path.  It's probably 
an ambiguous point, are we violating the "reverse engineering" clauses 
if we notice how long a Windows API takes?   Is 4'33 of silence, 
anywhere, a violation of John Cage's copyright?   On such abstruse legal 
decisions pettifogging legal beagles gold-plate their yachts.


( In the 1940's, Northwest Airlines, on Dec 31'st each year, they would 
fly every plane to a small airport in North Dakota, where their auditor 
would photograph their presence, to document that all their planes were 
"domiciled" in North Dakota, thereby they only had to pay ND taxes on 
them.  That case went all the way to the Supreme Court.   One judge's 
opinion was that since they had ruled in 1881 and 1933 that railroad 
cars had to be taxed in their state of longest residence, planes should 
be likewise taxed.   ) FYI: Excerpts: 
[https://supreme.justia.com/cases/federal/us/322/292/case.html]

but, bottom line, I have no idea.

On 01/26/2018 04:47 PM, Daniel Santos wrote:
> 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

[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