Renaming registry keys...

[Robert van Herk]

Hi!

Does anybody have a C algorithm to rename a registry section using the 
Win32 API?

I need such a thing for winecfg, but Win32 doesn't seem to have a 
RegKeyRename api-call...

Grtz,
Robert

[Dimitrie O. Paun]

On December 31, 2003 08:00 am, Robert van Herk wrote:
> I need such a thing for winecfg, but Win32 doesn't seem to have a
> RegKeyRename api-call...

Why do you need the rename a key in winecfg?

[Robert van Herk]

Dimitrie O. Paun wrote:
> On December 31, 2003 08:00 am, Robert van Herk wrote:
> > I need such a thing for winecfg, but Win32 doesn't seem to have a
> > RegKeyRename api-call...
>
> Why do you need the rename a key in winecfg?

When someone configures app1.exe, and later on decides that he actually 
meant app2.exe, he might want to rename app1.exe to app2.exe, and so 
migrate all the settings.

[Mike Hearn]

On Thu, 2004-01-01 at 13:38, Robert van Herk wrote:
> When someone configures app1.exe, and later on decides that he actually 
> meant app2.exe, he might want to rename app1.exe to app2.exe, and so 
> migrate all the settings.

Oh, I see.... I was intending to simply not support that :)

What you want is a deep tree copy function. You can then do renaming by
copying the contents of a key to a new key, and deleting the old one.

Here's one I just knocked up. Total pain in the ass isn't it! Hopefully
when we get a standard config system on Linux it will be a bit better
designed....

/* Copy the values and subkeys of the registry key source to dest recursively.
 * Returns NO_ERROR if the copy completed OK, otherwise it propagates the error.
 */
LONG copytree(HKEY source, HKEY dest) {
    int i = 0;
    LONG res;

/* first of all, copy the values */
    while (1) {
    char valname[MAX_VALUE_NAME];
    DWORD valname_size = sizeof(valname);
    DWORD size;
    DWORD typecode;
    char *value;
    
    /* determine size of the buffer we'll need */
    res = RegEnumValue(source, i, valname, &valname_size, NULL, &typecode, NULL, &size);
    if (res == ERROR_NO_MORE_ITEMS) break;
    if (res != NO_ERROR) return res;

value = HeapAlloc(GetProcessHeap(), 0, size);
    if (!value) return ERROR_OUTOFMEMORY;

valname_size = sizeof(valname); /* RegEnumValue blasts this even though it doesn't need to */
    res = RegEnumValue(source, i, valname, &valname_size, NULL, &typecode, value, &size);
    if (res != NO_ERROR) return res;

res = RegSetValueEx(dest, valname, 0, typecode, value, size);
    if (res != NO_ERROR) return res;

HeapFree(GetProcessHeap(), 0, value);
    
    i++;
    }

/* for each subkey, create an equivalent under dest then recurse into ourselves to complete the copy */
    i = 0;
    while (1) {
    HKEY subkey_src, subkey_dest;
    char name[MAX_KEY_LENGTH];
    DWORD size = sizeof(name);
    FILETIME time;

res = RegEnumKeyEx(source, i, name, &size, NULL, NULL, NULL, &time);
    if (res == ERROR_NO_MORE_ITEMS) break;
    if (res != NO_ERROR) return res;

/* create the destination subkey */
    res = RegCreateKey(dest, name, &subkey_dest);
    if (res != NO_ERROR) return res;

/* open the source subkey */
    res = RegOpenKey(source, name, &subkey_src);
    if (res != NO_ERROR) return res;

/* perform the recursive copy */
    res = copytree(subkey_src, subkey_dest);
    if (res != NO_ERROR) return res;

i++;
    }
    return NO_ERROR;
}

Hope that helps!
thanks -mike

[Dimitrie O. Paun]

On January 1, 2004 08:38 am, Robert van Herk wrote:
> When someone configures app1.exe, and later on decides that he actually
> meant app2.exe, he might want to rename app1.exe to app2.exe, and so
> migrate all the settings.

I'd say don't support that. At least for now, it's not that important
a functionality to worry about it...

[Robert van Herk]

Dimitrie O. Paun wrote:
> On January 1, 2004 08:38 am, Robert van Herk wrote:
> > When someone configures app1.exe, and later on decides that he actually
> > meant app2.exe, he might want to rename app1.exe to app2.exe, and so
> > migrate all the settings.
>
> I'd say don't support that. At least for now, it's not that important
> a functionality to worry about it...

Hm, maybe you're right. I will leave it for a moment then, and maybe 
include it in the future.

Thanx though for all the help, guys.

Grtz,
Robert