Re: [PATCH 1/1] msi: MsiGetPropertyA return wrong length when called from a custom action.

[Dmitry Timoshkov]

"Mike McCormack"  wrote:
> This is a hack, however the correct way to achieve bug compatiblity is 
> complicated.
>
> MSI should run custom actions in a separate process and communicate with 
> the main installer process via COM.  This is quite a lot of work which 
> needs to be done anyway, but this patch will make us bug compatible with 
> Microsoft's implementation in the mean time.
>
> Furthermore, writing a test case for this behavior requires adding a dll 
> contain custom action code to the dlls/msi/tests directory.  I can 
> supply an MSVC project showing the problem.
>
> See bug 5768 (http://bugs.winehq.org/show_bug.cgi?id=5768) for more details.

Does behaviour of MsiGetPropertyA and MsiGetPropertyW differ in that case?
I'd be interested to have a look at the test case.
> --- a/dlls/msi/custom.c
> +++ b/dlls/msi/custom.c
> @@ -439,6 +439,8 @@ static DWORD WINAPI ACTION_CallDllFunction(thread_struct *stuff)
>     TRACE("calling function (%s, %s)\n", debugstr_w(stuff->source),
>           debugstr_w(stuff->target));
> - TlsSetValue( tls_slot, (void*) 1 );

Shouldn't this be reset to 0 once a custom action has finished?

[PATCH 1/1] msi: MsiGetPropertyA return wrong length when called from a custom action.

[Mike McCormack]

Dmitry Timoshkov wrote:
> Does behaviour of MsiGetPropertyA and MsiGetPropertyW differ in that case?
> I'd be interested to have a look at the test case.

I've put the test case up here:
http://mandoo.dyndns.org/msitest.zip
> > - TlsSetValue( tls_slot, (void*) 1 );
>
> Shouldn't this be reset to 0 once a custom action has finished?

The thread running the custom action terminates immediately after the 
custom action finishes, so there shouldn't be any need to reset the tls 
value.

Mike