RFC: Patch to change what sets the is_win9x in riched20/tests

[James McKenzie]

All:

Upon examining other test code that creates a variable called is_win9x, 
I realized that using get_version and comparing it to a fixed value may 
not be best for the riched20 tests and have attached a proposed change 
to how this variable is set.  It uses a called function, lstrcmpW and if 
it does not exist, the variable is set to a false value.  This change 
has been tested on the testbot for 
Windows95/98/98SE/2K/2K3/XP/XP_64/Vista/Vista64/Win7/Win7_64 and no 
discrepancies were found.

Comments on this patch are appreciated.  I would like to submit this for 
inclusion into the Wine code base on Friday.

James McKenzie

[Paul Vriens]

On 02/22/2011 01:21 AM, James McKenzie wrote:
> All:
>
> Upon examining other test code that creates a variable called is_win9x,
> I realized that using get_version and comparing it to a fixed value may
> not be best for the riched20 tests and have attached a proposed change
> to how this variable is set. It uses a called function, lstrcmpW and if
> it does not exist, the variable is set to a false value. This change has
> been tested on the testbot for
> Windows95/98/98SE/2K/2K3/XP/XP_64/Vista/Vista64/Win7/Win7_64 and no
> discrepancies were found.

Win9x tests are no longer run with winetest. I also see that Austin sent 
some 9x cleanup patches. That said, I think the best way is to get rid 
of all the win9x 'hacks' in editor.c and rely on the fact that we have NT4+.

-- 
Cheers,

Paul.

[James McKenzie]

On 2/22/11 12:42 AM, Paul Vriens wrote:
> On 02/22/2011 01:21 AM, James McKenzie wrote:
> > All:
> >
> > Upon examining other test code that creates a variable called is_win9x,
> > I realized that using get_version and comparing it to a fixed value may
> > not be best for the riched20 tests and have attached a proposed change
> > to how this variable is set. It uses a called function, lstrcmpW and if
> > it does not exist, the variable is set to a false value. This change has
> > been tested on the testbot for
> > Windows95/98/98SE/2K/2K3/XP/XP_64/Vista/Vista64/Win7/Win7_64 and no
> > discrepancies were found.
>
> Win9x tests are no longer run with winetest. I also see that Austin 
> sent some 9x cleanup patches. That said, I think the best way is to 
> get rid of all the win9x 'hacks' in editor.c and rely on the fact that 
> we have NT4+.

Paul:

While that is true, I thought the consensus was that testing would still 
be available for Window9X/ME.  There are users (like me) that are 
running Windows9x/ME programs and don't want to loose the ability to run 
them under Wine.

This function may not exist in Windows versions after Windows2K either, 
that is why I proposed changing this from a version check to actually 
checking for the called function.

And lastly, I agree with adding tests to specifically check what happens 
in the riched20.dll for UNICODE calls.

James McKenzie