[ros-kernel] Header issues (final attempt)

[Waldo Alvarez Cañizares]

________________________________

From: ros-kernel-bounces at reactos.com on behalf of Eric Kohl
Sent: Tue 1/27/2004 6:50 AM
To: ReactOS Kernel
Subject: [ros-kernel] Header issues (final attempt)



Hi!

I want to get rid of the current header chaos as a part of my post-0.2.0 (or
pre-0.2.1) cleanup. I already replaced the WINBOOL type by BOOL and I am
currently replacing WIN32 types (DWORD, WORD, BYTE, BOOL, LP*) in the
kernel-mode and native code.

Please note that this is my final attempt to fix the headers. This means
that I _am_ going to fix them, whether you like it or not. Your only chance
it to raise your hand and notify me of potential problems but you won't stop
me from ripping out the old headers. I take this stance because none of the
previous discussions lead to a real change and I am sick of it.

My plan is to get rid of the MinGW headers because the MinGW team does not
accept undocumented declarations. It is their decision but it is a major
issue to us. We should neither use their headers nor their import libraries
because we have our own ones.

I there anything wrong with using extra headers for the undocummented features?

something like

#include <undoc/xxx.h>

or

#include <xxx_undoc.h>

I really don't think is a good idea to drop MingGW headers. New developers for ReactOS will heve to learn to use those
 new headers you speak about slowing down the learning curve and also you won't have people working for ReactOS indirectly from the
 mingw team. However if you really need to do it, define things like in the SDK and not like with Wine. I strongly oppose to drop mingw headers and/or use
those wrong Wine headers because of those reasons.

Best Regards

Waldo Alvarez

I want to use our current msvcrt headers (in include/msvcrt) as the new CRT
(C-Runtime-Libraty) headers and import the current Wine headers as the new
Win32 headers.

This approach fixes most of our problems but some minor issues still remain.

Several structures are declared in SDK and DDK headers and cause a lot of
trouble if SDK and DDK functions a used in some user-mode code. M$ solves
this issue by defining '_NT_INCLUDED_' at the top of 'ntddk.h' and by not
including some headers files in 'windows.h' if '_NT_INCLUDED_' is defined.
Tis is a much cleaner way than our current '_USE_WIN32_' approach but it
does not work with the old Win32 headers by Colin Peters.

IIRC, Alexandre is going to fix the '#include "header.h"' vs. '#include
<header.h>' issue.

Any comments?


Regards,
Eric

_______________________________________________
Ros-kernel mailing list
Ros-kernel at reactos.com
http://reactos.com/mailman/listinfo/ros-kernel <http://reactos.com/mailman/listinfo/ros-kernel> 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5555 bytes
Desc: not available
Url : http://reactos.com:8080/pipermail/ros-kernel/attachments/20040127/5d11a524/attachment-0001.bin