Re: ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Henri Verbeet]

2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> This fixes bug #9008.
> Please provide a more descriptive log message, and consider writing a
> test. I'm also pretty sure this will conflict with
> http://source.winehq.org/patches/data/63676.

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Oldřich Jedlička]

Hi Henri,

On Monday 19 July 2010 20:07:08 Henri Verbeet wrote:
> 2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> > This fixes bug #9008.
>
> Please provide a more descriptive log message, and consider writing a
> test. I'm also pretty sure this will conflict with
> http://source.winehq.org/patches/data/63676.

I will enhance the log message, no problem here. Should I consider the 
referenced patch as being applied (so that I should base my patch on it)?

The tests will be a problem, because I don't have the Windows machine as a 
reference. Also `make test` fails on DirectX tests for me because of r600 
driver bug...

When I write CreateSurface tests (for various interfaces), is there a way for 
me to run the tests on Windows? I have a Windows XP machine at work.

Thanks,
Oldrich.

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Austin English]

2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> The tests will be a problem, because I don't have the Windows machine as a
> reference. Also `make test` fails on DirectX tests for me because of r600
> driver bug...
>
> When I write CreateSurface tests (for various interfaces), is there a way for
> me to run the tests on Windows? I have a Windows XP machine at work.

If you have mingw installed, you can use 'make crosstest' to make a
test executable that will run on windows.

-- 
-Austin

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[James McKenzie]

Austin English wrote:
> 2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> > The tests will be a problem, because I don't have the Windows machine as a
> > reference. Also `make test` fails on DirectX tests for me because of r600
> > driver bug...
> >
> > When I write CreateSurface tests (for various interfaces), is there a way for
> > me to run the tests on Windows? I have a Windows XP machine at work.
>
> If you have mingw installed, you can use 'make crosstest' to make a
> test executable that will run on windows.

Don't forget that we have the testbot available as well, for comparing 
your tests against many versions of Windows (thanks Ge.)

James McKenzie

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Oldřich Jedlička]

Hi James,

On Tuesday 20 July 2010 03:43:39 James McKenzie wrote:
> Austin English wrote:
> > 2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> > > The tests will be a problem, because I don't have the Windows machine as
> > > a reference. Also `make test` fails on DirectX tests for me because of
> > > r600 driver bug...
> > >
> > > When I write CreateSurface tests (for various interfaces), is there a
> > > way for me to run the tests on Windows? I have a Windows XP machine at
> > > work.
> >
> > If you have mingw installed, you can use 'make crosstest' to make a
> > test executable that will run on windows.
>
> Don't forget that we have the testbot available as well, for comparing
> your tests against many versions of Windows (thanks Ge.)

This is somehow automatic testing I guess. Is it sufficient to send the patch 
with tests or how does it work? When the buildbot recognizes to apply the 
patch and test?

Oldrich.
> James McKenzie

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Henri Verbeet]

2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> I will enhance the log message, no problem here. Should I consider the
> referenced patch as being applied (so that I should base my patch on it)?

Yeah.
> The tests will be a problem, because I don't have the Windows machine as a
> reference. Also `make test` fails on DirectX tests for me because of r600
> driver bug...
>
> When I write CreateSurface tests (for various interfaces), is there a way for
> me to run the tests on Windows? I have a Windows XP machine at work.

As Austin mentioned you can use "make crosstest" if you have a wine32
cross compiler installed. You can also ask people on that bug,
wine-devel, or IRC to run a test for you on native Windows.

Wrt the patch itself, can't you just check for the appropriate flag in
IDirectDraw7Impl_CreateSurface(), and allow it in
IDirectDrawImpl_CreateSurface()? (I.e., avoid adding the
AllowBackBuffer parameter.)

ddraw: Allow creating back buffer for pre-DirectX 7 interfaces.

[Oldřich Jedlička]

On Monday 19 July 2010 21:41:26 Henri Verbeet wrote:
> 2010/7/19 Oldřich Jedlička <oldium.pro(a)seznam.cz>:
> > I will enhance the log message, no problem here. Should I consider the
> > referenced patch as being applied (so that I should base my patch on it)?
>
> Yeah.

Ok, I will update it according to your changes :-)
> > The tests will be a problem, because I don't have the Windows machine as
> > a reference. Also `make test` fails on DirectX tests for me because of
> > r600 driver bug...
> >
> > When I write CreateSurface tests (for various interfaces), is there a way
> > for me to run the tests on Windows? I have a Windows XP machine at work.
>
> As Austin mentioned you can use "make crosstest" if you have a wine32
> cross compiler installed. You can also ask people on that bug,
> wine-devel, or IRC to run a test for you on native Windows.

I will have a look at this possibility, I've found a nice hint on cross-
compiling for Gentoo at http://wiki.winehq.org/JohnKlehm. It looks like I can 
do it myself (for now).
> Wrt the patch itself, can't you just check for the appropriate flag in
> IDirectDraw7Impl_CreateSurface(), and allow it in
> IDirectDrawImpl_CreateSurface()? (I.e., avoid adding the
> AllowBackBuffer parameter.)

I just wanted to keep the debugging messages (dumping DDSD) where they are and 
fail afterwards. If I moved the check to the interface-specific function, there 
would be no such DDSD dump. But yes, this can be work-around to dump it just 
when the back-buffer test fails.

I will send the updated version (without tests for now) tomorrow for more 
comments :-)

Thanks,
Oldrich.