[Bug 681] Changed - Two of the shlwapi DLL regression tests fail under wine configured as an NT-like OS

[Paul Millar]

On Fri, 17 May 2002 wine-bugs at winehq.com wrote:
> + ------- Additional Comments From juergen.schmied at debitel.net  2002-05-17 09:44 -------
> + Running the test native shlwapi shows exactly the same behaviour. The 
> + bug is in  a deeper layer (advapi32 or ntdll).

As I put in my original (wine-devel) email
   http://www.winehq.com/hypermail/wine-devel/2002/05/0364.html
I tracked this down to RegQueryValueExA.

AFAIK, under NT-like modes, this is mapped to the advapi32 dll's
implementation (in dlls/advapi32/registry.c) whereas under Win95-like
modes the RegQueryValueExA in memory/registry.c is used instead. Although
the memory/registry.c RegQueryValueExA is also used when wine is in
NT-like mode.

To progress any further would involve investing considerable amount of
time figuring out what all the calls are supposed to return (I'm not a
windows programmer!) I just don't have that time at the moment.

It does look like this one cuts deep (ish). One possible gambit is that
the advapi32 looks like it was based on the memory/registry.c version. If
the two copies diverged after the copy, then this would give the observed
result. It might be possible to look for patches that effected
memory/registry.c that didn't alter dlls/advapi32/registry.c in the same
way.

Cheers,

Paul.

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
Particle Physics (Theory & Experimental) Groups                   Paul Millar 
Department of Physics and Astronomy                     paulm at astro.gla.ac.uk
University of Glasgow                                 paulm at physics.gla.ac.uk
Glasgow, G12 8QQ, Scotland             http://www.astro.gla.ac.uk/users/paulm 
+44 (0)141 330 4717        A54C A9FC 6A77 1664 2E4E  90E3 FFD2 704B BF0F 03E9
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --