Re: Dmitry Timoshkov : winspool: Add a test for GetPrinterDriver, make it pass under Wine.

[Detlef Riekenberg]

Am Freitag, den 14.04.2006, 12:46 -0500 schrieb Alexandre Julliard:
> Author: Dmitry Timoshkov <dmitry(a)codeweavers.com>
> Date:   Fri Apr 14 22:34:57 2006 +0900
>
> winspool: Add a test for GetPrinterDriver, make it pass under Wine.

Thanks for working on winspool.drv, but i'm Sorry that I had not 
enough Time to look at the Patch before the commit.

- Version 0x0400 was correct for win9x-Drivers (4.00).
  Version 3 is for Usermode-Drivers on w2k and above.
  (Version 2 is Kernelmode-Driver for NT4.0 and w2k)
  (Version 1 is NT3.5x and Version 0 is NT 3.1)
  Our Registry-Location is for win9x only (OpenDriverReg)

- One Terminating Zero is is not enough for REG_MULTI_SZ 
  http://bugs.winehq.org/show_bug.cgi?id=3589

- Why was (PDRIVER_INFO_3W) changed to (PDRIVER_INFO_2W)?
  (PDRIVER_INFO_3W) is an extended Version of (PDRIVER_INFO_2W)

- Not all Levels are supported on all Windows-Versions
  http://msdn.microsoft.com/library/en-us/gdi/prntspol_7xo2.asp

- The tests are not alphabetical sorted after this commit

- Why was SetLastError(0xdeadbeef) used, when there is
  SetLastError(MAGIC_DEAD) in the other tests

- 11 Tests are now failing on win95, 11 Tests on win98,
  9 Tests on winme and 8 Tests on NT3.51.
  For NT4.0 (level 4, 5 and 6 as well as the Version-Test will fail)
  and w2k (version will fail, when an NT4.0-Driver is used), i can not
  test the exact number of failures atm.


What's the preferred way to fix this?
My Idea:
1. Move the Test without any Changes
2. Fix the Tests
3. Fix the Implementation


I know, that you want a fast fix for your commercial Application and
winspool.drv is very ugly. I'm on the way to fix winspool.drv, but my
Direction is going from the Bottom-Layer to the High-Level API.

"struct printerenv_t" was created (as requested by Alexandre) to handle
the the different Printing-Environments for win9x-Drivers (Windows 4.0)
and NT-Drivers (Windows NT x86).
My fix for OpenDriverReg already need an Extension here, but
DeleteMonitor comes first.

You can Read more Infos here: http://wiki.winehq.org/Printing

-- 
By By ...
      ... Detlef

Dmitry Timoshkov : winspool: Add a test for GetPrinterDriver, make it pass under Wine.

[Dmitry Timoshkov]

First of all thanks for reviewing the test!

"Detlef Riekenberg" <wine.dev(a)web.de> wrote:
> - Version 0x0400 was correct for win9x-Drivers (4.00).
>  Version 3 is for Usermode-Drivers on w2k and above.
>  (Version 2 is Kernelmode-Driver for NT4.0 and w2k)
>  (Version 1 is NT3.5x and Version 0 is NT 3.1)
>  Our Registry-Location is for win9x only (OpenDriverReg)

MSDN states that:
cVersion Specifies the operating system version for which the driver
was written. It can be one of the following.
Value    Meaning
0        Driver for Windows 95/98/Me.
1        Driver for Windows NT 3.51.
2        Driver for Windows NT 4.0.
3        Driver for Windows 2000/XP.

Since Wine by default reports either win9x or win2000/XP I've chosen
to report 0 or 3, I don't think there is any app that depends on
a particular value.
> - One Terminating Zero is is not enough for REG_MULTI_SZ 
>  http://bugs.winehq.org/show_bug.cgi?id=3589

Feel free to add a test for that and fix it.
> - Why was (PDRIVER_INFO_3W) changed to (PDRIVER_INFO_2W)?
>  (PDRIVER_INFO_3W) is an extended Version of (PDRIVER_INFO_2W)

PDRIVER_INFO_2W is the lowest denominator for all extended structures,
I think it's more correct to use it.
> - Not all Levels are supported on all Windows-Versions
>  http://msdn.microsoft.com/library/en-us/gdi/prntspol_7xo2.asp

Feel free to adopt the test to handle an error in that case and not report
a failure.
> - The tests are not alphabetical sorted after this commit

I don't see much point in that.
> - Why was SetLastError(0xdeadbeef) used, when there is
>  SetLastError(MAGIC_DEAD) in the other tests

All the tests in Wine I write and most others use 0xdeadbeef directly,
that way I don't need to inspect the whole file to see to which value
that magic macro is defined.
> - 11 Tests are now failing on win95, 11 Tests on win98,
>  9 Tests on winme and 8 Tests on NT3.51.
>  For NT4.0 (level 4, 5 and 6 as well as the Version-Test will fail)
>  and w2k (version will fail, when an NT4.0-Driver is used), i can not
>  test the exact number of failures atm.

I'll have a look once the test results appear on test.winehq.org/data,
but feel free to make the test fail more gracefully on that platforms.

-- 
Dmitry.