What is an invalid HKEY?

[Stefan Kuhr]

Hello everyone,

I am just wondering if code with registry functions that I am used to
write for years is correct. Maybe this is also my own little personal
paranoia...

The problem with registry functions like RegOpenKeyEx is that they both
have a return value that indicates success or failure and that they also
fill in a Handle. I usually have at the end of my routines a general
resource cleanup section that frees all resources a function claims by
testing handle values or ptrs for invalid values and closing/freeing the
handles/ptrs if they are valid. In order to do so I initialize them with
an invalid value like NULL (e.g. for heap allocations) or
INVALID_HANDLE_VALUE.

To illustrate this, I usually use for CreateFile something like this:

BOOL Foo (...)
{
HANDLE hFile = INVALID_HANDLE_VALUE;
BOOL bResult = TRUE;

// lots of stuff follows ...

hFile = CreateFile(...);

if (INVALID_HANDLE_VALUE==hFile)
{
bResult = FALSE;
goto CLEANUP;
}

/// lots
/// of
/// stuff here can follow on success...

CLEANUP:

// this is the function's general cleanup section

if (INVALID_HANDLE_VALUE!=hFile)
CloseHandle(hFile);

return bResult;
}

Now back to registry functions like RegOpenKeyEx. What is the correct
initialization value for an HKEY? Is it NULL or is it
INVALID_HANDLE_VALUE? Or are both possible valid HKEY values and I have
to keep around an additional BOOL for the return value of RegOpenKeyEx
and the like? Up to now I inizialize all HKEYs with NULL (and test for
non-NULL in the cleanup section for a RegCloseKey) but I am getting
doubts if this is correct.

Any insights? Please don't give me answers like "Goto considered
harmful" - people who answer this haven't understood my dilemma.

TIA,

--
Stefan

[Ara Avanesyan]

0 is an invalid HKEY so it's okay to setup the hkey to 0 (at least registry
functions check their HKEY args agains 0:)).

Btw, if you are programming in C++, you are using quite old method for
cleanup, instead you would better use one of the auto_ cleaners. This makes
the code significantly easy-to-read and less buggy.

Thanks,
Ara,
0xLLC

"Stefan Kuhr" <kust...@gmx.li> wrote in message
news:406DD912...@gmx.li...

[Stefan Kuhr]

Hi Ara

Ara Avanesyan wrote:
>
> 0 is an invalid HKEY so it's okay to setup the hkey to 0 (at least registry
> functions check their HKEY args agains 0:)).

Thanks. So it seems like I always did the right thing, which is a big
relief for me. But how do you know that the Reg* functions internally
check for NULL HKEYs?

> Btw, if you are programming in C++, you are using quite old method for
> cleanup, instead you would better use one of the auto_ cleaners. This makes
> the code significantly easy-to-read and less buggy.

Thanks for your advice. I typically use this approach, combined with SEH
in plain C code. I sometimes use it with C++ code that aims to be a
better type-checked C code as well. Nevertheless, I don't think you can
judge from the abbreviated snippet I posted how easy to read and buggy
my code typically is :-)

--
Stefan

[Ara Avanesyan]

Hi Stefan,

"Stefan Kuhr" <kust...@gmx.li> wrote in message

news:406ED261...@gmx.li...

> Hi Ara
>
> Ara Avanesyan wrote:
> >
> > 0 is an invalid HKEY so it's okay to setup the hkey to 0 (at least
registry
> > functions check their HKEY args agains 0:)).
>
> Thanks. So it seems like I always did the right thing, which is a big
> relief for me. But how do you know that the Reg* functions internally
> check for NULL HKEYs?

Just debug the first few instructions of the functions

>
> Nevertheless, I don't think you can
> judge from the abbreviated snippet I posted how easy to read and buggy
> my code typically is :-)

I never meant something like that about your code:) I was just talking about
the mechanisms...

>
> --
> Stefan