Re: RegDeleteTreeA

[Paul Vriens]

Stefan Leichter wrote:
> ChangeLog
>
> add some tests for RegDeleteTreeA

Hi Stefan

good to see someone picked up this thing. I ran your new tests on Vista Ultimate 
  and comments are below. (I ran the tests while bypassing UAC, as administrator).
> --- ../wine-git/dlls/advapi32/tests/registry.c	2007-03-17 08:36:33.000000000 +0100
> +++ dlls/advapi32/tests/registry.c	2007-04-01 14:00:12.000000000 +0200
> @@ -35,6 +35,7 @@
>
> static HMODULE hadvapi32;
>  static DWORD (WINAPI *pRegGetValueA)(HKEY,LPCSTR,LPCSTR,DWORD,LPDWORD,PVOID,LPDWORD);
> +static DWORD (WINAPI *pRegDeleteTreeA)(HKEY,LPCSTR);
>
> #define ADVAPI32_GET_PROC(func) \
>      p ## func = (void*)GetProcAddress(hadvapi32, #func); \
> @@ -47,6 +48,7 @@
>  /* This function was introduced with Windows 2003 SP1 */
>  ADVAPI32_GET_PROC(RegGetValueA)
> - ADVAPI32_GET_PROC(RegDeleteTreeA)
>
> }
>
> /* delete key and all its subkeys */
> @@ -872,6 +874,53 @@
>      RegDeleteKeyA(subkey, "");
>  }
>
> +static void test_reg_delete_tree(void)
> +{
> - CHAR buffer[MAX_PATH];
> - HKEY subkey, subkey2;
> - LONG size, ret;
> -
> - if(!pRegDeleteTreeA) return;

Can you add a skip here?
> -
> - ret = RegCreateKeyA(hkey_main, "subkey", &subkey);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = RegCreateKeyA(subkey, "subkey2", &subkey2);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = RegSetValueA(subkey, NULL, REG_SZ, "data", 4);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = RegSetValueA(subkey2, NULL, REG_SZ, "data2", 5);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = RegCloseKey(subkey2);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> -
> - ret = pRegDeleteTreeA(subkey, "subkey2");
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ok(RegOpenKeyA(subkey, "subkey2", &subkey2),
> -
>    "subkey2 was not deleted\n");
> - size = MAX_PATH;
> - ok(!RegQueryValueA(subkey, "subkey", buffer, &size),
> -
>    "Default value of subkey not longer present\n");

This test fails so it must mean the default value should have been deleted.
> -
> - ret = RegCreateKeyA(subkey, "subkey2", &subkey2);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = RegCloseKey(subkey2);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ret = pRegDeleteTreeA(hkey_main, "subkey\subkey2");
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ok(RegOpenKeyA(subkey, "subkey2", &subkey2),
> -
>    "subkey2 was not deleted\n");
> - ok(!RegQueryValueA(subkey, "subkey", buffer, &size),
> -
>    "Default value of subkey not longer present\n");

This test fails so it must mean the default value should have been deleted.
> -
> - ret = pRegDeleteTreeA(subkey, NULL);
> - ok(ret == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", ret);
> - ok(RegOpenKeyA(hkey_main, "subkey", &subkey),
> -
>    "subkey was not deleted\n");

The subkey will not be deleted by RegDeleteTree (see MSDN). This is the main 
difference between RegDeleteTree and SHDeleteKey (where the main key is deleted).
> -
> - ret = pRegDeleteTreeA(hkey_main, "not-here");
> - trace("deleting an not existing key returned %d\n", ret);

The answer to this one is 2 : ERROR_FILE_NOT_FOUND
> +}
> -
>
> START_TEST(registry)
>  {
>      /* Load pointers for functions that are not available in all Windows versions */
> @@ -901,6 +950,8 @@
>          set_privileges(SE_RESTORE_NAME, FALSE);
>      }
> - test_reg_delete_tree();
> - /* cleanup */
>    delete_key( hkey_main );

Cheers,

Paul.

RegDeleteTreeA

[Stefan Leichter]

Am Sunday 01 April 2007 20:19 schrieb Paul Vriens:
> Stefan Leichter wrote:
> > ChangeLog
> >
> > add some tests for RegDeleteTreeA
>
> Hi Stefan
>
> good to see someone picked up this thing. I ran your new tests on Vista
> Ultimate and comments are below. (I ran the tests while bypassing UAC, as
> administrator).

Hello Paul,

thank you for verifying the tests before they get into git. This will save us 
some git cycles to get all tests pass. Can you please retest the new patch.

Thanks
Stefan

RegDeleteTreeA

[Paul Vriens]

On 4/2/07, Stefan Leichter  wrote:
> Am Sunday 01 April 2007 20:19 schrieb Paul Vriens:
> > Stefan Leichter wrote:
> > > ChangeLog
> > > add some tests for RegDeleteTreeA
> >
> > Hi Stefan
> >
> > good to see someone picked up this thing. I ran your new tests on Vista
> > Ultimate and comments are below. (I ran the tests while bypassing UAC, as
> > administrator).
>
> Hello Paul,
>
> thank you for verifying the tests before they get into git. This will save us
> some git cycles to get all tests pass. Can you please retest the new patch.
>
> Thanks
> Stefan

Hi Stefan,

only one failure left. When 'ret = pRegDeleteTreeA(subkey, NULL);' is
called the system deletes the original default but creates a new one.
This means the last test 'ok(RegQueryValueA(subkey, NULL, buffer,
&size)' fails. Maybe this check should be against the original default
(where some data is present).

Cheers,

Paul.