CommConfigDialog() and serialui.dll

[Saveliy Tretiakov]

Hi all.
This is my changes :)

serialui
  - Started serialui dll
  - Implemented drvCommConfigDialogA and drvCommConfigDialogW
 
kernel32.diff
  - Implemented CommConfigDialogA and CommConfigDialogW
  - Removed @unimplemented comments for some implemented functions

include.diff
  - Add missing ioctls to ntddser.h:
    IOCTL_SERIAL_GET_COMMCONFIG and IOCTL_SERIAL_SET_COMMCONFIG
   
btw.
Screenshot is here:
http://www.ljplus.ru/img2/drg4njubas/commdlg.JPG

[Hervé Poussineau]

Saveliy Tretiakov a écrit :
>   serialui
>   - Started serialui dll
>   - Implemented drvCommConfigDialogA and drvCommConfigDialogW 
http://reactos.com:8080/pipermail/ros-svn/2005-April/002164.html

>   kernel32.diff
>   - Implemented CommConfigDialogA and CommConfigDialogW
>   - Removed @unimplemented comments for some implemented functions 
Not committed because they give warnings:
misc/comm.c: In function `CommConfigDialogA':
misc/comm.c:854: warning: use of cast expressions as lvalues is deprecated
misc/comm.c: In function `CommConfigDialogW':
misc/comm.c:887: warning: use of cast expressions as lvalues is deprecated
MAKE: *** [misc/comm.o] Error 1

>   include.diff
>   - Add missing ioctls to ntddser.h:
>     IOCTL_SERIAL_GET_COMMCONFIG and IOCTL_SERIAL_SET_COMMCONFIG 
They are not documented by Microsoft (see 
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/serial/hh/…

) and googling on them doesn't give good results.
Do you have some documentation?

[Saveliy Tretiakov]

Hervé Poussineau wrote:

>   Not committed because they give warnings:
>  misc/comm.c: In function `CommConfigDialogA':
>  misc/comm.c:854: warning: use of cast expressions as lvalues is 
>  deprecated
>  misc/comm.c: In function `CommConfigDialogW':
>  misc/comm.c:887: warning: use of cast expressions as lvalues is 
>  deprecated
>  MAKE: *** [misc/comm.o] Error 1
>  
Hmm, my compiler(gcc 3.3.3) didn't return any warnings. I will fix them 
tomorrow.

> >   include.diff
> >   - Add missing ioctls to ntddser.h:
> >     IOCTL_SERIAL_GET_COMMCONFIG and IOCTL_SERIAL_SET_COMMCONFIG 
>
>  They are not documented by Microsoft (see 
>
> http://msdn.microsoft.com/library/default.asp?url=/library/en-us/serial/hh/…
>
>  ) and googling on them doesn't give good results.
>  Do you have some documentation? 
I didn't know they are not documented :)
You can find this ioctls defined in microsoft's ntddser.h. They are used 
to acess SERIALCONFIG structure.
I need this to implement GetCommConfig and SetCommConfig apis. The 
SERIALCONFIG structure
is undocumented too. So what? Maybe remove it from our ntddser.h? :)

[Thomas Weidenmueller]

Hervé Poussineau wrote:

>   Saveliy Tretiakov a écrit :
>
> >   serialui
> >   - Started serialui dll
> >   - Implemented drvCommConfigDialogA and drvCommConfigDialogW 
>
>  http://reactos.com:8080/pipermail/ros-svn/2005-April/002164.html 
I just noticed wine also has a serialui.dll implementation. Is there a 
reason why Saveliy reinvented the wheel?

Best Regards,
Thomas

[Saveliy Tretiakov]

Thomas Weidenmueller wrote:

>  
>  I just noticed wine also has a serialui.dll implementation. Is there a 
>  reason why Saveliy reinvented the wheel? 
serialui.dll from wine is not implemented correctly.  Almost all code 
from there needs a rewrite\remove and can't be included into
reactos.  It was more simple for me to reimplement it from scratch.

[James Tabor]

Hi!
How could I miss this! I guess I was asleep!

Saveliy Tretiakov wrote:
>   Hi all.
>  This is my changes :)
>  
>  serialui
>   - Started serialui dll
>   - Implemented drvCommConfigDialogA and drvCommConfigDialogW
>  
>  kernel32.diff
>   - Implemented CommConfigDialogA and CommConfigDialogW
>   - Removed @unimplemented comments for some implemented functions
>  
>  include.diff
>   - Add missing ioctls to ntddser.h:
>     IOCTL_SERIAL_GET_COMMCONFIG and IOCTL_SERIAL_SET_COMMCONFIG
>    btw.
>  Screenshot is here:
>  http://www.ljplus.ru/img2/drg4njubas/commdlg.JPG
>   
You can attach graphic files, it's okay. How do you call up the
config menu?

Thanks,
James

[Thomas Weidenmueller]

James Tabor wrote:

>   Hi!
>  How could I miss this! I guess I was asleep!
>
>  Saveliy Tretiakov wrote:
>
> >   Hi all.
> >  This is my changes :)
> >
> >  serialui
> >   - Started serialui dll
> >   - Implemented drvCommConfigDialogA and drvCommConfigDialogW
> >
> >  kernel32.diff
> >   - Implemented CommConfigDialogA and CommConfigDialogW
> >   - Removed @unimplemented comments for some implemented functions
> >
> >  include.diff
> >   - Add missing ioctls to ntddser.h:
> >     IOCTL_SERIAL_GET_COMMCONFIG and IOCTL_SERIAL_SET_COMMCONFIG
> >    btw.
> >  Screenshot is here:
> >  http://www.ljplus.ru/img2/drg4njubas/commdlg.JPG
> >  
>  You can attach graphic files, it's okay. How do you call up the
>  config menu? 
As documented in the platform SDK?

Best Regards,
Thomas

[Saveliy Tretiakov]

James Tabor wrote:

>   How do you call up the
>  config menu? 
kernel32 loads serialui.dll and calls drvCommConfigDialog.

Thomas Weidenmueller wrote:

>   As documented in the platform SDK? 
It seems to me that it is not documented in platform SDK. But I"ve
tested my dll in windows environment and it worked.

 From platform SDK:

Remarks
The *CommConfigDialog* function requires a dynamic-link library (DLL) 
provided by the communications hardware vendor.

[Thomas Weidenmueller]

Saveliy Tretiakov wrote:

>   It seems to me that it is not documented in platform
> SDK. But I"ve
>  tested my dll in windows environment and it worked. 
I thought you meant CommConfigDialog. Btw i tried calling 
drvCommConfigDialogA in WinXP and it returned 
ERROR_CALL_NOT_IMPLEMENTED. So the return type is definitely not a BOOL 
but rather a DWORD returning a win32 error code (or 0 for success).

Best Regards,
Thomas

[Saveliy Tretiakov]

Thomas Weidenmueller wrote:

>   I thought you meant CommConfigDialog. Btw i tried
> calling 
>  drvCommConfigDialogA in WinXP and it returned 
>  ERROR_CALL_NOT_IMPLEMENTED. So the return type is definitely not a 
>  BOOL but rather a DWORD returning a win32 error code (or 0 for success). 
Maybe this is because you pressed cancel button and 
ERROR_CALL_NOT_IMPLEMENTED
was set somewhere else? It returns FALSE when cancel button is pressed.

 From CommConfig dialog description:
If the function succeeds, the return value is nonzero.
If the function fails, the return value is zero. To get extended error 
information, call *GetLastError*.

I think drvCommConfigDialog should return the same. Maybe adding 
SetLastError(ERROR_CANCELLED)
and SetLastError(ERROR_SUCCESS) will solve the problem?

[Thomas Weidenmueller]

Saveliy Tretiakov wrote:

>   Maybe this is because you pressed cancel button and 
>  ERROR_CALL_NOT_IMPLEMENTED
>  was set somewhere else? It returns FALSE when cancel button is pressed. 
no, in my copy of win XP calling drvCommConfigDialogA always returns 
ERROR_CALL_NOT_IMPLEMENTED. Calling the unicode version works, it 
returns FALSE/0/ERROR_SUCCESS or ERROR_CANCELLED when correct parameters 
are passed to it.

>   From CommConfig dialog description:
>  If the function succeeds, the return value is nonzero.
>  If the function fails, the return value is zero. To get extended error 
>  information, call *GetLastError*. 
I'm talking about drvCommConfigDialog, not CommConfig. However 
CommConfig calls drvCommConfigDialog, which appears to return 
ERROR_SUCCESS, ERROR_CANCELLED or another win32 error code, which is not 
a BOOL.

>   I think drvCommConfigDialog should return the same.
As i mentioned above, it appears to return a win32 error code, not a 
BOOL in windows XP, which leads to the conclusion that 
drvCommConfigDialog doesn't have the exact same prototype as CommConfig.

>   Maybe adding SetLastError(ERROR_CANCELLED) and 
>  SetLastError(ERROR_SUCCESS) will solve the problem? 
I propose only CommConfig should set the last error, if 
drvCommConfigDialog didn't return ERROR_SUCCESS or ERROR_CANCELLED. Then 
returning TRUE from CommConfig when drvCommConfigDialog returned 
ERROR_SUCCESS and returning FALSE in all other cases should implement it 
correctly. Instead of setting the last error code in 
drvCommConfigDialog, the function should return it directly.

Best Regards,
Thomas

[Saveliy Tretiakov]

Ok, thank you.
These patches correct return values.

Thomas Weidenmueller wrote:

>   Saveliy Tretiakov wrote:
>
> >   Maybe this is because you pressed cancel button
> > and 
> >  ERROR_CALL_NOT_IMPLEMENTED
> >  was set somewhere else? It returns FALSE when cancel button is pressed. 
>
>  no, in my copy of win XP calling drvCommConfigDialogA always returns 
>  ERROR_CALL_NOT_IMPLEMENTED. Calling the unicode version works, it 
>  returns FALSE/0/ERROR_SUCCESS or ERROR_CANCELLED when correct 
>  parameters are passed to it.
>
> >   From CommConfig dialog description:
> >  If the function succeeds, the return value is nonzero.
> >  If the function fails, the return value is zero. To get extended 
> >  error information, call *GetLastError*. 
>
>  I'm talking about drvCommConfigDialog, not CommConfig. However 
>  CommConfig calls drvCommConfigDialog, which appears to return 
>  ERROR_SUCCESS, ERROR_CANCELLED or another win32 error code, which is 
>  not a BOOL.
>
> >   I think drvCommConfigDialog should return the
> > same. 
>
>  As i mentioned above, it appears to return a win32 error code, not a 
>  BOOL in windows XP, which leads to the conclusion that 
>  drvCommConfigDialog doesn't have the exact same prototype as CommConfig.
>
> >   Maybe adding SetLastError(ERROR_CANCELLED) and 
> >  SetLastError(ERROR_SUCCESS) will solve the problem? 
>
>  I propose only CommConfig should set the last error, if 
>  drvCommConfigDialog didn't return ERROR_SUCCESS or ERROR_CANCELLED. 
>  Then returning TRUE from CommConfig when drvCommConfigDialog returned 
>  ERROR_SUCCESS and returning FALSE in all other cases should implement 
>  it correctly. Instead of setting the last error code in 
>  drvCommConfigDialog, the function should return it directly.
>
>  Best Regards,
>  Thomas
>  _______________________________________________
>  Ros-dev mailing list
>  Ros-dev(a)reactos.com
>  http://reactos.com:8080/mailman/listinfo/ros-dev
>
>