[ros-diffs] [greatlrd] 27767: fixed so metafile works bit better,

[James Tabor]

Hi!
We need to re-think Metafile trash inside win32k!

Hint: Win32k Gdi rewrite?

If developers are going to help me do this, help me do it right!

Remove metafile from win32k and implement it in gdi32 where it is done
in real windows. I did place examples in the tree on how to do this.

Poorly coded code built up upon poorly implement code equals more poorly
implemented code on code, or just trash code!

Thank you for your cooperation,
James

[Magnus Olsen]

Hi
what I can see metafile are done in two place in windows one in win32k
another in gdi32
some stuff can only be done in win32k in windows and reactos and metafile
need some setup
in win32k for it exists kernel api that need it.

Yes our gdi32 and win32k sucks it should be good a good rewriting was done,
for gdi subsystem. But before we can do that we need real testcase for each
NtGdi* stuff and test it in windows so we really known how they works.
after that is done we can start rewrite each NtGdi*. then later testcase for
each Gdi* api in gdi32, that will make our gdi substem good. and stable

But we need share memory betwin win32k and gdi/dx stuff

I known the metafile implement is not 100% ok,
it is base how wine doing it, and I did follow wine implement
and did not check how it really works in windows nt,

----- Original Message ----- 
From: "James Tabor" [e-mail]
To: "ReactOS Development List" [e-mail]
Sent: Sunday, July 22, 2007 4:07 AM
Subject: [ros-dev] [ros-diffs] [greatlrd] 27767: fixed so metafile works
bitbetter,
> Hi!
> We need to re-think Metafile trash inside win32k!
>
> Hint: Win32k Gdi rewrite?
>
> If developers are going to help me do this, help me do it right!
>
> Remove metafile from win32k and implement it in gdi32 where it is done
> in real windows. I did place examples in the tree on how to do this.
>
> Poorly coded code built up upon poorly implement code equals more poorly
> implemented code on code, or just trash code!
>
> Thank you for your cooperation,
> James
> _______________________________________________
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev

[James Tabor]

Hi all and Magnus,
Magnus Olsen wrote:
> Hi
> what I can see metafile are done in two place in windows one in win32k
> another in gdi32
> some stuff can only be done in win32k in windows and reactos and metafile
> need some setup
> in win32k for it exists kernel api that need it.

EnhMeta (not 16 bit) look at my code example, it's a DC with the EnhMeta bits set in
the handle table, most of the work is done in gdi32. Wine again is a good example.

Meta (16 bit) is created in gdi32,,,,, but,,,,, the dc handle is created via
nt create object. It works out side the normal DC mode of thinking.

Still, just like wine does it,,,,,, it is all handled in gdi32 not win32k, looks
like a gdi32 script player.
> Yes our gdi32 and win32k sucks it should be good a good rewriting was done,
> for gdi subsystem. But before we can do that we need real testcase for each
> NtGdi* stuff and test it in windows so we really known how they works.
> after that is done we can start rewrite each NtGdi*. then later testcase for
> each Gdi* api in gdi32, that will make our gdi substem good. and stable

Already done. Wine test with hooks can reveal the truth. Also there is nothing
wrong with that at all.
> But we need share memory betwin win32k and gdi/dx stuff

I have ideas to fix this. Blame GvG for this one!
> I known the metafile implement is not 100% ok,
> it is base how wine doing it, and I did follow wine implement
> and did not check how it really works in windows nt,

It's okay.

Thank you for your cooperation,
James