Re: kernel32/tests: Remove some expected values for test_DeleteFileA()

[Dmitry Timoshkov]

Zhenbo Li <litimetal(a)gmail.com> wrote:
> There're too many expected values from GetLastError(), and I think it 
> might be confusing.

Usually there was a reason why there are so many different results
in ok() statements, it would be a good idea to check git history and
find out that reasons it out instead of simply declaring them as
confusing and removing them altogether.

-- 
Dmitry.

kernel32/tests: Remove some expected values for test_DeleteFileA()

[Zhenbo Li]

2013/12/12 Dmitry Timoshkov <dmitry(a)baikal.ru>:
> Usually there was a reason why there are so many different results
> in ok() statements, it would be a good idea to check git history and
> find out that reasons it out instead of simply declaring them as
> confusing and removing them altogether.

Thank you, it's quite a good idea.
My patch is related to 3 parts:

PART 1
> ret = DeleteFileA(NULL);
> -    ok(!ret && (GetLastError() == ERROR_INVALID_PARAMETER ||
> -                GetLastError() == ERROR_PATH_NOT_FOUND),
> +    ok(!ret && GetLastError() == ERROR_PATH_NOT_FOUND,
>        "DeleteFileA(NULL) returned ret=%d error=%d\n",ret,GetLastError());

The old code came from commit e948ad1fc7e18a2
Author: Francois Gouget <fgouget(a)free.fr>  2002-12-12 11:54:01
Committer: Alexandre Julliard <julliard(a)winehq.org>  2002-12-12 11:54:01

+    ret = DeleteFileA(NULL);
+    ok(!ret && (GetLastError() == ERROR_INVALID_PARAMETER ||
+                GetLastError() == ERROR_PATH_NOT_FOUND),
+       "DeleteFileA(NULL) returned ret=%d error=%ld",ret,GetLastError());




PART 2
> ret = DeleteFileA("");
> -    ok(!ret && (GetLastError() == ERROR_PATH_NOT_FOUND ||
> -                GetLastError() == ERROR_BAD_PATHNAME),
> +    ok(!ret && GetLastError() == ERROR_PATH_NOT_FOUND,
>        "DeleteFileA(\"\") returned ret=%d error=%d\n",ret,GetLastError());

The old code came from commit e948ad1fc7e18a2
Author: Francois Gouget <fgouget(a)free.fr>  2002-12-12 11:54:01
Committer: Alexandre Julliard <julliard(a)winehq.org>  2002-12-12 11:54:01
He added ERROR_BAD_PATHNAME

As Francois mentioned, he "Adapt the DeleteFileA error code checks to
take into account variations
    between Win9x and NT."


PART 3
> ret = DeleteFileA("nul");
> -    ok(!ret && (GetLastError() == ERROR_FILE_NOT_FOUND ||
> -                GetLastError() == ERROR_INVALID_PARAMETER ||
> -                GetLastError() == ERROR_ACCESS_DENIED ||
> -                GetLastError() == ERROR_INVALID_FUNCTION),
> +    ok(!ret && GetLastError() == ERROR_INVALID_PARAMETER,
>        "DeleteFileA(\"nul\") returned ret=%d error=%d\n",ret,GetLastError());
>
> GetTempPathA(MAX_PATH, temp_path);

The old code was related to 2 commits

commit  c49b9485
Author: Jakob Eriksson <jakov(a)vmlinux.org>  2004-04-28 11:52:02
Committer: Alexandre Julliard <julliard(a)winehq.org>  2004-04-28 11:52:02
ERROR_INVALID_PARAMETER and  ERROR_ACCESS_DENIED were added.

and commit 6cb97534
Author: Jacek Caban <jack(a)itma.pwr.wroc.pl>  2005-06-27 17:49:26
Committer: Alexandre Julliard <julliard(a)winehq.org>  2005-06-27 17:49:26
ERROR_INVALID_FUNCTION was added for NT4

As they only exist in old WIndows versions, should we still keep them
in testcases? Or make them as broken?

Appreciate your reply.

-- 
Have a nice day!
Zhenbo Li