Anybody have a RegDeleteTree[A|W] implementation lying around?

[Paul Vriens]

Hi,

I was browsing some code and found that we have several places where we
made our own implementation of a recursive delete of registry
keys/value.

Apparently there is an official one in Vista:

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/sysinfo/bas...

I don't know if somebody has this already lying around somewhere. If
not, I'll have a go (not soon though).

Would be a nice code cleanup, or are there any pitfalls?

Cheers,

Paul.

[Vitaliy Margolen]

Paul Vriens wrote:
> Hi,
>
> I was browsing some code and found that we have several places where we
> made our own implementation of a recursive delete of registry
> keys/value.
> See winecfg.

Vitaliy.

[Paul Vriens]

On Thu, 2006-09-28 at 07:01 -0600, Vitaliy Margolen wrote:
> Paul Vriens wrote:
> > Hi,
> >
> > I was browsing some code and found that we have several places where we
> > made our own implementation of a recursive delete of registry
> > keys/value.
> > See winecfg.
>
> Vitaliy.

Do you mean winecfg has it's implementation, or winecfg has the
solution?

I had another look, apparently shlwapi has SHDeleteKey[A|W]. It does the
same thing.

I'll have a few hours tomorrow and will have a look. The cleanest
solution for now seems to create RegDeleteTree[A|W] in advapi32 and use
that wherever possible (unless we find that it forwards to shlwapi).
Most DLL's already import advapi32. 

Cheers,

Paul.

[Frank Richter]

On 28.09.2006 15:26, Paul Vriens wrote:
> I'll have a few hours tomorrow and will have a look. The cleanest
> solution for now seems to create RegDeleteTree[A|W] in advapi32 and use
> that wherever possible (unless we find that it forwards to shlwapi).
> Most DLL's already import advapi32.

Hmm... isn't advapi32 "lower level" than shlwapi? I'd expect that if
forwarding takes place then it'd be shlwapi->advapi32.

-f.r.

[Brad DeMorrow]

Is Anyone currently working on this?

If Not - I would like to start working on this part myself, it seems like it
is fairly simple - but tedious work :-)

Shouldn't be too much of a hassle and should help a little

--Brad DeMorrow


On 9/28/06, Frank Richter <frank.richter(a)gmail.com> wrote:
> On 28.09.2006 15:26, Paul Vriens wrote:
> > I'll have a few hours tomorrow and will have a look. The cleanest
> > solution for now seems to create RegDeleteTree[A|W] in advapi32 and use
> > that wherever possible (unless we find that it forwards to shlwapi).
> > Most DLL's already import advapi32.
>
> Hmm... isn't advapi32 "lower level" than shlwapi? I'd expect that if
> forwarding takes place then it'd be shlwapi->advapi32.
>
> -f.r.

[Segin]

Brad DeMorrow wrote:
> Is Anyone currently working on this?
>
> If Not - I would like to start working on this part myself, it seems
> like it
> is fairly simple - but tedious work :-)
>
> Shouldn't be too much of a hassle and should help a little
>
> --Brad DeMorrow
>
> On 9/28/06, Frank Richter <frank.richter(a)gmail.com> wrote:
> > On 28.09.2006 15:26, Paul Vriens wrote:
> > > I'll have a few hours tomorrow and will have a look. The cleanest
> > > solution for now seems to create RegDeleteTree[A|W] in advapi32 and use
> > > that wherever possible (unless we find that it forwards to shlwapi).
> > > Most DLL's already import advapi32.
> >
> > Hmm... isn't advapi32 "lower level" than shlwapi? I'd expect that if
> > forwarding takes place then it'd be shlwapi->advapi32.
> >
> > -f.r.
>
> ------------------------------------------------------------------------

Some Google searching provides that there may or may not be an
implementation.

http://www.reactos.org/pipermail/ros-svn/2005-August/005303.html
([ros-svn] [weiden] 17572: implemented RegDeleteTree() (untested!))
mentions this, and nothing more:
--- CUT ---
implemented RegDeleteTree() (untested!)

Updated files:
trunk/reactos/lib/advapi32/reg/reg.c
--- CUT ---
The code moved around in their tree a bit, it's actually
trunk/reactos/dll/win32/advapi32/reg/reg.c, and you can view the code here:
http://svn.reactos.org/svn/reactos/trunk/reactos/dll/win32/advapi32/reg/reg....
-- 
The real problem with C++ for kernel modules is: the language just sucks.
	-- Linus Torvalds

[James Hawkins]

On 10/10/06, Segin <segin2005(a)gmail.com> wrote:
> >
>
> Some Google searching provides that there may or may not be an
> implementation.
>
> http://www.reactos.org/pipermail/ros-svn/2005-August/005303.html
> ([ros-svn] [weiden] 17572: implemented RegDeleteTree() (untested!))
> mentions this, and nothing more:
> --- CUT ---
> implemented RegDeleteTree() (untested!)
>
> Updated files:
> trunk/reactos/lib/advapi32/reg/reg.c
> --- CUT ---
> The code moved around in their tree a bit, it's actually
> trunk/reactos/dll/win32/advapi32/reg/reg.c, and you can view the code here:
> http://svn.reactos.org/svn/reactos/trunk/reactos/dll/win32/advapi32/reg/reg....

We don't accept code from the ReactOS project.

-- 
James Hawkins

[Segin]

James Hawkins wrote:
> On 10/10/06, Segin <segin2005(a)gmail.com> wrote:
> > >
> >
> > Some Google searching provides that there may or may not be an
> > implementation.
> >
> > http://www.reactos.org/pipermail/ros-svn/2005-August/005303.html
> > ([ros-svn] [weiden] 17572: implemented RegDeleteTree() (untested!))
> > mentions this, and nothing more:
> > --- CUT ---
> > implemented RegDeleteTree() (untested!)
> >
> > Updated files:
> > trunk/reactos/lib/advapi32/reg/reg.c
> > --- CUT ---
> > The code moved around in their tree a bit, it's actually
> > trunk/reactos/dll/win32/advapi32/reg/reg.c, and you can view the code
> > here:
> > http://svn.reactos.org/svn/reactos/trunk/reactos/dll/win32/advapi32/reg/reg....
>
> We don't accept code from the ReactOS project.

Thought it was a good idea at the time, but I can understand why you
wouldn't accept it. (I can assume that this has to do with ensuring that
the code is honest, due to the previous ReactOS "scandal")

-- 
The real problem with C++ for kernel modules is: the language just sucks.
	-- Linus Torvalds

[Vitaliy Margolen]

Segin wrote:
> James Hawkins wrote:
> > On 10/10/06, Segin <segin2005(a)gmail.com> wrote:
> > > Some Google searching provides that there may or may not be an
> > > implementation.
> > >
> > > http://www.reactos.org/pipermail/ros-svn/2005-August/005303.html
> > > ([ros-svn] [weiden] 17572: implemented RegDeleteTree() (untested!))
> > > mentions this, and nothing more:
> > > --- CUT ---
> > > implemented RegDeleteTree() (untested!)
> > >
> > > Updated files:
> > > trunk/reactos/lib/advapi32/reg/reg.c
> > > --- CUT ---
> > > The code moved around in their tree a bit, it's actually
> > > trunk/reactos/dll/win32/advapi32/reg/reg.c, and you can view the code
> > > here:
> > > http://svn.reactos.org/svn/reactos/trunk/reactos/dll/win32/advapi32/reg/reg....
> >
> > We don't accept code from the ReactOS project.
>
> Thought it was a good idea at the time, but I can understand why you
> wouldn't accept it. (I can assume that this has to do with ensuring that
> the code is honest, due to the previous ReactOS "scandal")

There are 2 (TWO) implementations of this functionality in Wine:
1. http://source.winehq.org/source/programs/winecfg/winecfg.c#L210
2. http://source.winehq.org/source/dlls/advapi32/registry.c#L938

If you guys need one more - have at it ;-)

Vitaliy

[Paul Vriens]

On Tue, 2006-10-10 at 16:32 -0500, Brad DeMorrow wrote:
> Is Anyone currently working on this?
>
> If Not - I would like to start working on this part myself, it seems
> like it is fairly simple - but tedious work :-)
>
> Shouldn't be too much of a hassle and should help a little
>
> --Brad DeMorrow
>
> On 9/28/06, Frank Richter <frank.richter(a)gmail.com> wrote:
>         On 28.09.2006 15:26, Paul Vriens wrote:
>         > I'll have a few hours tomorrow and will have a look. The
>         cleanest
>         > solution for now seems to create RegDeleteTree[A|W] in
>         advapi32 and use
>         > that wherever possible (unless we find that it forwards to
>         shlwapi). 
>         > Most DLL's already import advapi32.
>
> Hmm... isn't advapi32 "lower level" than shlwapi? I'd expect
>         that if
>         forwarding takes place then it'd be shlwapi->advapi32.
>
> -f.r.

I've created a wikipage for it:

http://wiki.winehq.org/RecursiveRegistryKeyDelete

There you can find pointers to several implementations already. 

When I raised the question I thought I could find some spare time. (I
didn't obviously).

As RegDeleteTree is a Vista function (and thus not officially there
yet), I'm not sure how Alexandre would react to this. Of course we need
some conformance tests for this, but the only OS's where it will run
will be Vista and Wine. I'm not sure whether winetest can cope with
Vista yet.

I'd say go for it and use one of the many implementations we already
have as a base. Maybe the best one is SHDeleteKey as that's an official
one (with tests already) and you'd only have to leave out the last
RegDeleteKey.

Cheers,

Paul.
>